<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: James Basset</title>
    <description>The latest articles on DEV Community by James Basset (@jamesbasset).</description>
    <link>https://dev.to/jamesbasset</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4257730%2F6b1ea788-bbc8-4428-a22f-14a4e7663d9c.png</url>
      <title>DEV Community: James Basset</title>
      <link>https://dev.to/jamesbasset</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/jamesbasset"/>
    <language>en</language>
    <item>
      <title>Build a Slack alert for people asking to replace [Tool] in 20 lines</title>
      <dc:creator>James Basset</dc:creator>
      <pubDate>Sat, 21 Nov 2026 17:36:06 +0000</pubDate>
      <link>https://dev.to/jamesbasset/build-a-slack-alert-for-people-asking-to-replace-tool-in-20-lines-4peb</link>
      <guid>https://dev.to/jamesbasset/build-a-slack-alert-for-people-asking-to-replace-tool-in-20-lines-4peb</guid>
      <description>&lt;p&gt;You do not need a social-listening stack to find out when someone wants to replace a tool you compete with.&lt;/p&gt;

&lt;p&gt;If the author already posted it in public — "looking for an alternative to X", "need to switch before renewal" — you can poll a scored feed, keep only the rows that name that tool, and push a Slack message with the original thread link. The script at the heart of this is 20 lines of Python. The rest of the article is the request shape, the response shape, and how not to waste quota or annoy people.&lt;/p&gt;

&lt;p&gt;The feed I will use is &lt;a href="https://rapidapi.com/JamesBasset/api/b2b-buying-intent-signals" rel="noopener noreferrer"&gt;B2B Buying Intent Signals&lt;/a&gt; on RapidAPI. It scans public posts on Reddit, Hacker News, Lobsters, Dev.to, and Stack Exchange Software Recommendations every six hours, scores buying intent from 1 to 10, and keeps the high-intent ones (score 7+ by default). At the time of writing the live feed holds ~100 high-intent B2B signals. The tools that show up most often today are Shopify, then Supabase, then ecommerce and payments tools such as Veeqo, Vesyl and Stripe, and hosting/dev tools such as Railway, Lovable, Bolt and GitHub.&lt;/p&gt;

&lt;h2&gt;
  
  
  What you need
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;A RapidAPI account and a subscription to the API. The BASIC plan is free (200 calls/month).&lt;/li&gt;
&lt;li&gt;Your &lt;code&gt;X-RapidAPI-Key&lt;/code&gt; from the API playground. Also copy the &lt;code&gt;X-RapidAPI-Host&lt;/code&gt; value shown there; it is typically &lt;code&gt;b2b-buying-intent-signals.p.rapidapi.com&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;A Slack &lt;a href="https://api.slack.com/messaging/webhooks" rel="noopener noreferrer"&gt;Incoming Webhook&lt;/a&gt; for the channel that should receive alerts.&lt;/li&gt;
&lt;li&gt;Python 3 and &lt;code&gt;requests&lt;/code&gt; (&lt;code&gt;pip install requests&lt;/code&gt;).&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Set two environment variables and, optionally, the tool name (the default below is Shopify, currently the most-mentioned tool in the feed — change it to the product you actually care about):&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;&lt;span class="kd"&gt;export&lt;/span&gt; &lt;span class="nv"&gt;RAPIDAPI_KEY&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s2"&gt;"your-rapidapi-key"&lt;/span&gt;
&lt;span class="kd"&gt;export&lt;/span&gt; &lt;span class="nv"&gt;SLACK_WEBHOOK_URL&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s2"&gt;"https://hooks.slack.com/services/..."&lt;/span&gt;
&lt;span class="kd"&gt;export&lt;/span&gt; &lt;span class="nv"&gt;TOOL&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s2"&gt;"Shopify"&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;h2&gt;
  
  
  The endpoint
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight http"&gt;&lt;code&gt;&lt;span class="no"&gt;GET&lt;/span&gt; &lt;span class="n"&gt;https&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="c1"&gt;//b2b-buying-intent-signals.p.rapidapi.com/api/v1/signals&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Query parameters, all optional:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Parameter&lt;/th&gt;
&lt;th&gt;Default&lt;/th&gt;
&lt;th&gt;Notes&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;tool&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;(none)&lt;/td&gt;
&lt;td&gt;Case-insensitive &lt;strong&gt;exact&lt;/strong&gt; match on a &lt;code&gt;mentioned_tools&lt;/code&gt; entry. &lt;code&gt;tool=shopify&lt;/code&gt; matches &lt;code&gt;"Shopify"&lt;/code&gt;, but not &lt;code&gt;"Shopify Plus"&lt;/code&gt; unless that exact string is in the list. Max 80 characters.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;min_score&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;7&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Integer 1–10. 7–8 is a clear need; 9–10 is urgent (budget, deadline, "need this now").&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;urgency&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;(none)&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;low&lt;/code&gt;, &lt;code&gt;medium&lt;/code&gt;, or &lt;code&gt;high&lt;/code&gt;.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;category&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;(none)&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;ai_tools&lt;/code&gt;, &lt;code&gt;saas_alternative&lt;/code&gt;, &lt;code&gt;automation&lt;/code&gt;, &lt;code&gt;dev_tools&lt;/code&gt;, or &lt;code&gt;marketing_growth&lt;/code&gt;.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;platform&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;(none)&lt;/td&gt;
&lt;td&gt;
&lt;code&gt;reddit&lt;/code&gt;, &lt;code&gt;hackernews&lt;/code&gt;, &lt;code&gt;lobsters&lt;/code&gt;, &lt;code&gt;devto&lt;/code&gt;, or &lt;code&gt;stackexchange&lt;/code&gt;.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;code&gt;limit&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;&lt;code&gt;20&lt;/code&gt;&lt;/td&gt;
&lt;td&gt;Maximum 50.&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Authenticate with both headers, and go through RapidAPI — do not call the origin server directly:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight http"&gt;&lt;code&gt;&lt;span class="no"&gt;X&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;RapidAPI&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Key&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="no"&gt;YOUR_RAPIDAPI_KEY&lt;/span&gt;
&lt;span class="no"&gt;X&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;RapidAPI&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nc"&gt;Host&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;b2b&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="n"&gt;buying&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="n"&gt;intent&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="n"&gt;signals&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;p&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;rapidapi&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;com&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;A successful body always looks like &lt;code&gt;{ "success": true, "count": &amp;lt;n&amp;gt;, "data": [ ... ] }&lt;/code&gt;. The script below calls &lt;code&gt;GET /api/v1/signals?tool=shopify&amp;amp;min_score=7&lt;/code&gt;. I am showing a real public signal from the same endpoint with &lt;code&gt;tool=veeqo&lt;/code&gt; (Veeqo and Vesyl both appear in the current feed), so every field is one you can actually get back. The author has been anonymized.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="nl"&gt;"success"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="nl"&gt;"count"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="nl"&gt;"data"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;
    &lt;span class="p"&gt;{&lt;/span&gt;
      &lt;span class="nl"&gt;"id"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"16430170-0028-4416-8c4a-96e3699d9f7d"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="nl"&gt;"platform"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"reddit"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="nl"&gt;"source_url"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"https://www.reddit.com/r/ecommerce/comments/1y0su3v/veeqo_alternatives_that_arent_tied_to_amazon/"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="nl"&gt;"title"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"Veeqo alternatives that arent tied to Amazon?"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="nl"&gt;"author"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"user_820ea5"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="nl"&gt;"intent_score"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mi"&gt;8&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="nl"&gt;"category"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"saas_alternative"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="nl"&gt;"problem_summary"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"Ecommerce seller using Amazon-owned Veeqo wants a carrier- and channel-neutral order routing/shipping alternative because Amazon is only 20% of sales and they are uneasy routing order data through a competitor."&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="nl"&gt;"mentioned_tools"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"Veeqo"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"Vesyl"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="nl"&gt;"urgency"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"medium"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="nl"&gt;"created_at"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"2026-11-20T07:23:19.512Z"&lt;/span&gt;
    &lt;span class="p"&gt;}&lt;/span&gt;
  &lt;span class="p"&gt;]&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;created_at&lt;/code&gt; is UTC, and the list is newest first. Dedup on &lt;code&gt;id&lt;/code&gt;, not on the title. Other &lt;code&gt;tool&lt;/code&gt; values that currently return rows include &lt;code&gt;shopify&lt;/code&gt;, &lt;code&gt;supabase&lt;/code&gt;, &lt;code&gt;stripe&lt;/code&gt; and &lt;code&gt;railway&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;If something is wrong you get &lt;code&gt;{ "success": false, "error": "..." }&lt;/code&gt;. Typical cases:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;401&lt;/code&gt; — missing or invalid RapidAPI key, or the request did not go through the RapidAPI gateway&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;400&lt;/code&gt; — invalid &lt;code&gt;category&lt;/code&gt;, &lt;code&gt;platform&lt;/code&gt;, &lt;code&gt;urgency&lt;/code&gt;, &lt;code&gt;min_score&lt;/code&gt;, &lt;code&gt;limit&lt;/code&gt;, or &lt;code&gt;tool&lt;/code&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  The 20-line script
&lt;/h2&gt;

&lt;p&gt;Save this as &lt;code&gt;slack_intent_alert.py&lt;/code&gt;.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight python"&gt;&lt;code&gt;&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;json&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;os&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;pathlib&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;requests&lt;/span&gt;
&lt;span class="no"&gt;TOOL&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="no"&gt;SEEN&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;os&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;environ&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"TOOL"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"Shopify"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;pathlib&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nc"&gt;Path&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"seen_ids.json"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="n"&gt;seen&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="kt"&gt;set&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;json&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;loads&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="no"&gt;SEEN&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;read_text&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="no"&gt;SEEN&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;exists&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;else&lt;/span&gt; &lt;span class="kt"&gt;set&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="n"&gt;host&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;"b2b-buying-intent-signals.p.rapidapi.com"&lt;/span&gt;
&lt;span class="n"&gt;res&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;requests&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"https://{host}/api/v1/signals"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;timeout&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;20&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;headers&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="s2"&gt;"X-RapidAPI-Key"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;os&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;environ&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"RAPIDAPI_KEY"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"X-RapidAPI-Host"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;host&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;params&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="s2"&gt;"tool"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="no"&gt;TOOL&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"min_score"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mi"&gt;7&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"limit"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mi"&gt;20&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="n"&gt;res&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;raise_for_status&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="n"&gt;body&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;res&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;json&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="k"&gt;assert&lt;/span&gt; &lt;span class="n"&gt;body&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"success"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;body&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"error"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"API error"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="n"&gt;fresh&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;s&lt;/span&gt; &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;s&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;body&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"data"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;s&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"id"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;seen&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;
&lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;s&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;fresh&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;tools&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;", "&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;join&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;s&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"mentioned_tools"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;text&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"*Buying intent {s['intent_score']}/10 · {s['urgency']} · {s['platform']}*\n"&lt;/span&gt;
            &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"{s['title']}\n{s['problem_summary']}\n"&lt;/span&gt;
            &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"Tools: {tools or 'n/a'} · &amp;lt;{s['source_url']}|Open thread&amp;gt;"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;requests&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;post&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;os&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;environ&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"SLACK_WEBHOOK_URL"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;json&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="s2"&gt;"text"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;text&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;timeout&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;20&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;raise_for_status&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;seen&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;add&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;s&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"id"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="no"&gt;SEEN&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;write_text&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;json&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;dumps&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nb"&gt;sorted&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;seen&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="k"&gt;print&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"Posted {len(fresh)} alert(s) for tool={TOOL}"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Run it once:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;python slack_intent_alert.py
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The first run may post several messages if the feed already has matching signals. After that, &lt;code&gt;seen_ids.json&lt;/code&gt; keeps it quiet unless a new &lt;code&gt;id&lt;/code&gt; appears. If you only want the urgent ones, add &lt;code&gt;"urgency": "high"&lt;/code&gt; to &lt;code&gt;params&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;A Slack message from the example payload above looks like this:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;Buying intent 8/10 · medium · reddit&lt;/strong&gt;&lt;br&gt;
Veeqo alternatives that arent tied to Amazon?&lt;br&gt;
Ecommerce seller using Amazon-owned Veeqo wants a carrier- and channel-neutral order routing/shipping alternative because Amazon is only 20% of sales and they are uneasy routing order data through a competitor.&lt;br&gt;
Tools: Veeqo, Vesyl · &lt;a href="https://www.reddit.com/r/ecommerce/comments/1y0su3v/veeqo_alternatives_that_arent_tied_to_amazon/" rel="noopener noreferrer"&gt;Open thread&lt;/a&gt;&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;The script defaults to &lt;code&gt;TOOL=Shopify&lt;/code&gt;. Use the same file with &lt;code&gt;TOOL=Veeqo&lt;/code&gt;, &lt;code&gt;TOOL=Stripe&lt;/code&gt;, &lt;code&gt;TOOL=Supabase&lt;/code&gt; or &lt;code&gt;TOOL=Railway&lt;/code&gt; if that is the market you sell into.&lt;/p&gt;

&lt;h2&gt;
  
  
  Put it on a schedule (and do not over-poll)
&lt;/h2&gt;

&lt;p&gt;New signals are collected every six hours, so polling faster than that mostly burns quota. Four runs a day is about 120 calls a month, which fits in the 200-call free plan and leaves room to try filters in the playground.&lt;/p&gt;

&lt;p&gt;A cron line that trails the collector slightly:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;25 */6 * * * RAPIDAPI_KEY=... SLACK_WEBHOOK_URL=... TOOL=Shopify /usr/bin/python3 /path/to/slack_intent_alert.py
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;On GitHub Actions, a scheduled workflow with those values in Actions secrets is the same idea. Use a persistent &lt;code&gt;seen_ids.json&lt;/code&gt; (a repo-untracked file on a small VM, or an artifact / a tiny database). If the seen-id file resets, you will re-alert on old rows.&lt;/p&gt;

&lt;p&gt;&lt;code&gt;/api/v1/stats&lt;/code&gt; is there if you want a weekly pulse rather than a per-tool alert. It returns 7-day volume and average intent score, overall and by category. The shape is:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight json"&gt;&lt;code&gt;&lt;span class="p"&gt;{&lt;/span&gt;
  &lt;span class="nl"&gt;"success"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
  &lt;span class="nl"&gt;"data"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="nl"&gt;"period"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="nl"&gt;"days"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mi"&gt;7&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"from"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;ISO-8601 UTC start&amp;gt;"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"to"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;ISO-8601 UTC end&amp;gt;"&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="nl"&gt;"total_signals"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;integer&amp;gt;"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="nl"&gt;"avg_intent_score"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;number, rounded to 2 decimals&amp;gt;"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="nl"&gt;"by_category"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;
      &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="nl"&gt;"category"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"saas_alternative"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"count"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;integer&amp;gt;"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"avg_intent_score"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;number&amp;gt;"&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="nl"&gt;"category"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"ai_tools"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"count"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;integer&amp;gt;"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"avg_intent_score"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;number&amp;gt;"&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="nl"&gt;"category"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"automation"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"count"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;integer&amp;gt;"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"avg_intent_score"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;number&amp;gt;"&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="nl"&gt;"category"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"dev_tools"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"count"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;integer&amp;gt;"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"avg_intent_score"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;number&amp;gt;"&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
      &lt;span class="p"&gt;{&lt;/span&gt; &lt;span class="nl"&gt;"category"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"marketing_growth"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"count"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;integer&amp;gt;"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="nl"&gt;"avg_intent_score"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"&amp;lt;number&amp;gt;"&lt;/span&gt; &lt;span class="p"&gt;}&lt;/span&gt;
    &lt;span class="p"&gt;]&lt;/span&gt;
  &lt;span class="p"&gt;}&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Those five categories are always present in &lt;code&gt;by_category&lt;/code&gt;, including when a count is zero. I am showing the schema only — not a live snapshot, because the numbers change as the feed does.&lt;/p&gt;

&lt;h2&gt;
  
  
  A few sharp edges worth knowing
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;tool&lt;/code&gt; is exact, not a substring.&lt;/strong&gt; That is deliberate: &lt;code&gt;tool=meta&lt;/code&gt; should not match a tool whose name merely contains "meta". &lt;code&gt;tool=shopify&lt;/code&gt; matches &lt;code&gt;"Shopify"&lt;/code&gt;, but if the model stored &lt;code&gt;"Shopify Plus"&lt;/code&gt;, that filter will not hit it. When in doubt, pull a page without &lt;code&gt;tool&lt;/code&gt; and look at &lt;code&gt;mentioned_tools&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;&lt;code&gt;mentioned_tools&lt;/code&gt; can be empty.&lt;/strong&gt; The scorer only lists names that were explicitly cited. Your Slack line should tolerate an empty list.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Authors are whatever the source published.&lt;/strong&gt; Some platforms have a public username, some do not, and deleted authors come through as &lt;code&gt;null&lt;/code&gt;. Do not make the Slack message depend on &lt;code&gt;author&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Read the thread before you talk to anyone.&lt;/strong&gt; The valuable field is &lt;code&gt;source_url&lt;/code&gt;. Join the public conversation, or write a note that refers to the actual problem. A generic "saw you were looking for a tool" email is how this kind of feed gets a bad name.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you outgrow a webhook, the same loop can post to a CRM, a sequencer, or a Slack bot via &lt;code&gt;chat.postMessage&lt;/code&gt;. The API part does not change: poll &lt;code&gt;/api/v1/signals&lt;/code&gt;, filter, remember &lt;code&gt;id&lt;/code&gt;.&lt;/p&gt;

</description>
      <category>python</category>
      <category>slack</category>
      <category>api</category>
      <category>tutorial</category>
    </item>
    <item>
      <title>How I detect B2B buying intent in public developer forums with Python + an LLM</title>
      <dc:creator>James Basset</dc:creator>
      <pubDate>Sat, 21 Nov 2026 17:36:01 +0000</pubDate>
      <link>https://dev.to/jamesbasset/how-i-detect-b2b-buying-intent-in-public-developer-forums-with-python-an-llm-1fo</link>
      <guid>https://dev.to/jamesbasset/how-i-detect-b2b-buying-intent-in-public-developer-forums-with-python-an-llm-1fo</guid>
      <description>&lt;p&gt;Keyword alerts are a bad way to find software buyers.&lt;/p&gt;

&lt;p&gt;Search Reddit or Hacker News for "alternative to" and you will get launch posts, affiliate roundups, venting, and the occasional genuine "I need to replace this before renewal." The last one is the only row I want. Everything else is noise that burns time and, if you score it with a model, tokens.&lt;/p&gt;

&lt;p&gt;This article walks through the pipeline I use to detect &lt;strong&gt;B2B buying intent&lt;/strong&gt; in public developer and founder forums: fetch recent posts, deduplicate them, pre-filter with cheap text rules, then ask an LLM for a strict JSON score. I run the production collector in TypeScript; the approach is easier to read in Python, which is what I will use here.&lt;/p&gt;

&lt;p&gt;At the end I will point to a hosted API that already does this, if you would rather not operate a collector yourself. The useful part is the method. You can stop after the script and still have something that works.&lt;/p&gt;

&lt;h2&gt;
  
  
  What "buying intent" means here
&lt;/h2&gt;

&lt;p&gt;I am not trying to detect curiosity or brand mentions. I want posts where the &lt;strong&gt;author&lt;/strong&gt; is trying to buy, switch, or replace software: a SaaS tool, an AI product, automation, a developer tool, or a marketing/growth tool.&lt;/p&gt;

&lt;p&gt;A practical scoring scale, used as an integer from 1 to 10:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;1–3&lt;/strong&gt; — no purchase intent: news, opinion, self-promotion, a launch announcement, hiring, or a rant with no need&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;4–6&lt;/strong&gt; — vague interest or early research&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;7–8&lt;/strong&gt; — a clear need, actively evaluating solutions&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;9–10&lt;/strong&gt; — urgent, explicit intent to buy or switch now (a budget, a deadline, "need it this week")&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Two rules matter more than the rest of the prompt:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;If the author is promoting their &lt;strong&gt;own&lt;/strong&gt; product, the score must be 3 or below.&lt;/li&gt;
&lt;li&gt;Only scores &lt;strong&gt;7 and above&lt;/strong&gt; are kept. That cutoff is the whole product: a short list of people who are actually shopping.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;I also ask the model for a category, an urgency (&lt;code&gt;low&lt;/code&gt;, &lt;code&gt;medium&lt;/code&gt;, or &lt;code&gt;high&lt;/code&gt;), the software names explicitly cited, and one sentence that describes the buyer's need. The sentence is what you will actually use later, in Slack or in a CRM note.&lt;/p&gt;

&lt;h2&gt;
  
  
  The pipeline
&lt;/h2&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;public posts
    → SHA-256 hash of platform + external id (dedup)
    → keyword pre-filter, plus a looser software + question pass
    → LLM JSON analysis (capped per run and per day)
    → keep only if intent_score &amp;gt;= 7
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;In production this runs every six hours, looking back 36 hours, against Reddit, Hacker News, Lobsters, Dev.to, and Stack Exchange Software Recommendations. The Python script below uses the three sources that are easiest to call from a laptop: the HN Algolia API, the Dev.to API (&lt;code&gt;help&lt;/code&gt; and &lt;code&gt;discuss&lt;/code&gt;), and the Stack Exchange API for Software Recommendations. The same pre-filter and scorer apply if you add Reddit later.&lt;/p&gt;

&lt;p&gt;A note on Reddit, because it is the noisiest source and the one people reach for first: official JSON and RSS endpoints often return 403 or 429 from cloud IPs. A production collector needs a polite User-Agent, delays between requests, and a fallback. I use a near-real-time public archive (Arctic Shift) as the primary Reddit path and official Reddit as a fallback, and I skip any subreddit that errors so one 403 does not kill the run.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 1 — Fetch public posts, nothing private
&lt;/h2&gt;

&lt;p&gt;Stick to public APIs, send a real User-Agent, and sleep between pages. Software Recommendations is a special case: almost every question is "please recommend software," so I treat those posts as buying-intent &lt;strong&gt;candidates&lt;/strong&gt; even when they do not match my keyword list. They still go through the LLM. The pre-filter is skipped; the score threshold is not.&lt;/p&gt;

&lt;p&gt;Hacker News is searched with quoted intent phrases (&lt;code&gt;"alternative to"&lt;/code&gt;, &lt;code&gt;"looking to replace"&lt;/code&gt;, &lt;code&gt;"willing to pay"&lt;/code&gt;, and so on) across Ask HN, stories, and comments. Comments are noisy, so I drop any comment that does not also look like a software/tool discussion. Dev.to is simpler: page recent articles tagged &lt;code&gt;help&lt;/code&gt; or &lt;code&gt;discuss&lt;/code&gt; and keep those inside the time window.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight python"&gt;&lt;code&gt;&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;time&lt;/span&gt;
&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;requests&lt;/span&gt;

&lt;span class="no"&gt;USER_AGENT&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;"BuyingIntentTutorial/1.0 (dev.to article)"&lt;/span&gt;
&lt;span class="no"&gt;HEADERS&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="s2"&gt;"User-Agent"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="no"&gt;USER_AGENT&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"Accept"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"application/json"&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;
&lt;span class="no"&gt;HN_ALGOLIA&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;"https://hn.algolia.com/api/v1/search_by_date"&lt;/span&gt;

&lt;span class="no"&gt;INTENT_SEARCH_PHRASES&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="s1"&gt;'"looking for"'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s1"&gt;'"alternative to"'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s1"&gt;'"recommend a"'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s1"&gt;'"any suggestions"'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s1"&gt;'"switching from"'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s1"&gt;'"willing to pay"'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s1"&gt;'"need a tool"'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s1"&gt;'"replacement for"'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s1"&gt;'"looking to replace"'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s1"&gt;'"frustrated with"'&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
&lt;span class="p"&gt;)&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;get_json&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;url&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;params&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="kc"&gt;None&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;timeout&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;15&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;res&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;requests&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;url&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;headers&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="no"&gt;HEADERS&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;params&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;params&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;timeout&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;timeout&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;res&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;raise_for_status&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;res&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;json&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;fetch_hn&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;since_unix&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;until_unix&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;max_pages&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;2&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;posts&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;seen&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="kt"&gt;set&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;phrase&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="no"&gt;INTENT_SEARCH_PHRASES&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;tags&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"ask_hn"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"story"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"comment"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;page&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="nb"&gt;range&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;max_pages&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                &lt;span class="n"&gt;data&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;get_json&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
                    &lt;span class="no"&gt;HN_ALGOLIA&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                    &lt;span class="p"&gt;{&lt;/span&gt;
                        &lt;span class="s2"&gt;"query"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;phrase&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                        &lt;span class="s2"&gt;"tags"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;tags&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                        &lt;span class="s2"&gt;"numericFilters"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"created_at_i&amp;gt;={since_unix},created_at_i&amp;lt;{until_unix}"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                        &lt;span class="s2"&gt;"hitsPerPage"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="mi"&gt;50&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                        &lt;span class="s2"&gt;"page"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;page&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                    &lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                &lt;span class="p"&gt;)&lt;/span&gt;
                &lt;span class="n"&gt;hits&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;data&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"hits"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;
                &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;hit&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;hits&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                    &lt;span class="n"&gt;object_id&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;hit&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"objectID"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="n"&gt;object_id&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="n"&gt;object_id&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;seen&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                        &lt;span class="k"&gt;continue&lt;/span&gt;
                    &lt;span class="n"&gt;title&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;hit&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"title"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="n"&gt;hit&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"story_title"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="s2"&gt;"HN comment"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;strip&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                    &lt;span class="n"&gt;body&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;hit&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"comment_text"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="n"&gt;hit&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"story_text"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="s2"&gt;""&lt;/span&gt;
                    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;tags&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"comment"&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="no"&gt;SOFTWARE_QUALIFIER&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;search&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"{title}\n{body}"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                        &lt;span class="k"&gt;continue&lt;/span&gt;
                    &lt;span class="n"&gt;seen&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;add&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;object_id&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
                    &lt;span class="n"&gt;posts&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
                        &lt;span class="p"&gt;{&lt;/span&gt;
                            &lt;span class="s2"&gt;"platform"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"hackernews"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                            &lt;span class="s2"&gt;"external_id"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="kt"&gt;str&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;object_id&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                            &lt;span class="s2"&gt;"source"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="s2"&gt;"hackernews"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                            &lt;span class="s2"&gt;"url"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"https://news.ycombinator.com/item?id={object_id}"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                            &lt;span class="s2"&gt;"title"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;title&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                            &lt;span class="s2"&gt;"body"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nf"&gt;strip_html&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;body&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                            &lt;span class="s2"&gt;"author"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;hit&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"author"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
                        &lt;span class="p"&gt;}&lt;/span&gt;
                    &lt;span class="p"&gt;)&lt;/span&gt;
                &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="n"&gt;hits&lt;/span&gt; &lt;span class="ow"&gt;or&lt;/span&gt; &lt;span class="n"&gt;page&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt; &lt;span class="o"&gt;&amp;gt;=&lt;/span&gt; &lt;span class="n"&gt;data&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"nbPages"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="mi"&gt;1&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
                    &lt;span class="k"&gt;break&lt;/span&gt;
                &lt;span class="n"&gt;time&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;sleep&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="mf"&gt;0.15&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;posts&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;SOFTWARE_QUALIFIER&lt;/code&gt; and &lt;code&gt;strip_html&lt;/code&gt; are in the complete script at the end. Dev.to and Software Recommendations follow the same shape: normalize each item into &lt;code&gt;platform&lt;/code&gt;, &lt;code&gt;external_id&lt;/code&gt;, &lt;code&gt;url&lt;/code&gt;, &lt;code&gt;title&lt;/code&gt;, &lt;code&gt;body&lt;/code&gt;, and &lt;code&gt;author&lt;/code&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 2 — Dedup before you spend a token
&lt;/h2&gt;

&lt;p&gt;Hash &lt;code&gt;platform&lt;/code&gt; and the source's own id, not the title. Titles get edited; ids do not.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight python"&gt;&lt;code&gt;&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;hashlib&lt;/span&gt;

&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;source_hash&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;key&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"{post['platform']}:{post['external_id']}"&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;hashlib&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;sha256&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;key&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;encode&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;hexdigest&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;In production I store two things: accepted signals, and a memory of every post already analysed (including rejects). Without the second table, a post that scored 4 yesterday will be sent to the model again tomorrow. That is wasted spend, and it also hides your true yield (accepted / analysed), which you need if you ever tune keywords.&lt;/p&gt;

&lt;p&gt;For a local script, a JSON file of hashes is enough.&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 3 — Pre-filter, then fill leftover budget with a looser pass
&lt;/h2&gt;

&lt;p&gt;If you send every Ask HN and every &lt;code&gt;#discuss&lt;/code&gt; post to an LLM, most of your bill will be spent confirming that a post is not about buying software. A case-insensitive substring check against a short phrase list removes the bulk of that.&lt;/p&gt;

&lt;p&gt;Phrases that have been useful in practice:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;alternative to&lt;/code&gt;, &lt;code&gt;looking to replace&lt;/code&gt;, &lt;code&gt;replacement for&lt;/code&gt;, &lt;code&gt;switching from&lt;/code&gt;, &lt;code&gt;migrate from&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;looking for&lt;/code&gt;, &lt;code&gt;recommend a&lt;/code&gt;, &lt;code&gt;any suggestions&lt;/code&gt;, &lt;code&gt;any recommendations&lt;/code&gt;, &lt;code&gt;can anyone recommend&lt;/code&gt;, &lt;code&gt;seeking recommendations&lt;/code&gt;, &lt;code&gt;tool recommendations&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;need a tool&lt;/code&gt;, &lt;code&gt;need software&lt;/code&gt;, &lt;code&gt;need a saas&lt;/code&gt;, &lt;code&gt;software for&lt;/code&gt;, &lt;code&gt;best tool&lt;/code&gt;, &lt;code&gt;what tool do you&lt;/code&gt;, &lt;code&gt;how do you automate&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;willing to pay&lt;/code&gt;, &lt;code&gt;ready to pay&lt;/code&gt;, &lt;code&gt;budget for&lt;/code&gt;, &lt;code&gt;in the market for&lt;/code&gt;, &lt;code&gt;frustrated with&lt;/code&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Two exceptions keep this from being too tight:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Software Recommendations questions skip the keyword filter.&lt;/strong&gt; They are candidates by design.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;A looser second pass&lt;/strong&gt; fills leftover model budget. The post must look like a software conversation (&lt;code&gt;saas&lt;/code&gt;, &lt;code&gt;tool&lt;/code&gt;, &lt;code&gt;crm&lt;/code&gt;, &lt;code&gt;automation&lt;/code&gt;, &lt;code&gt;vendor&lt;/code&gt;, and similar) &lt;em&gt;and&lt;/em&gt; look like a question or a switch (&lt;code&gt;?&lt;/code&gt;, &lt;code&gt;alternative&lt;/code&gt;, &lt;code&gt;recommend&lt;/code&gt;, &lt;code&gt;replace&lt;/code&gt;, &lt;code&gt;switching from&lt;/code&gt;, "what is the best", and similar). Those loose matches are queued &lt;em&gt;after&lt;/em&gt; exact keyword hits.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Then rank: more keyword hits first, then longer bodies. Short titles with no context are a poor use of the cap.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight python"&gt;&lt;code&gt;&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;match_keywords&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;text&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;keywords&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;haystack&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;" "&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;join&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;text&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;lower&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;split&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="n"&gt;k&lt;/span&gt; &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;k&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;keywords&lt;/span&gt; &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;k&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;lower&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;haystack&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;is_loose_buying_candidate&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;title&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;body&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;text&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"{title}\n{body}"&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="ow"&gt;not&lt;/span&gt; &lt;span class="no"&gt;SOFTWARE_QUALIFIER&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;search&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;text&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="kc"&gt;False&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="nb"&gt;any&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;cue&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;search&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;text&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;cue&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="no"&gt;BUYING_QUESTION_CUES&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;to_candidates&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;posts&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;keywords&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;keyword_hits&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;loose_hits&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;
    &lt;span class="k"&gt;for&lt;/span&gt; &lt;span class="n"&gt;post&lt;/span&gt; &lt;span class="k"&gt;in&lt;/span&gt; &lt;span class="n"&gt;posts&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"platform"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"stackexchange"&lt;/span&gt; &lt;span class="ow"&gt;and&lt;/span&gt; &lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"source"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;==&lt;/span&gt; &lt;span class="s2"&gt;"softwarerecs"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="n"&gt;keyword_hits&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="o"&gt;**&lt;/span&gt;&lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"matched_keywords"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"softwarerecs"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
            &lt;span class="k"&gt;continue&lt;/span&gt;
        &lt;span class="n"&gt;matched&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;match_keywords&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"{post['title']}\n{post['body']}"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;keywords&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;matched&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="n"&gt;keyword_hits&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="o"&gt;**&lt;/span&gt;&lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"matched_keywords"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;matched&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
        &lt;span class="k"&gt;elif&lt;/span&gt; &lt;span class="nf"&gt;is_loose_buying_candidate&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"title"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"body"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
            &lt;span class="n"&gt;loose_hits&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;append&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="o"&gt;**&lt;/span&gt;&lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"matched_keywords"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"loose_intent"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;

    &lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;rank&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nb"&gt;len&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"matched_keywords"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="o"&gt;-&lt;/span&gt;&lt;span class="nb"&gt;len&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"body"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;

    &lt;span class="n"&gt;keyword_hits&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;sort&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;key&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;rank&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;loose_hits&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;sort&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;key&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;rank&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;keyword_hits&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="n"&gt;loose_hits&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Only then do you slice the queue to a budget (&lt;code&gt;MAX_AI_CALLS&lt;/code&gt; per run, plus a daily cap if you put this on a cron).&lt;/p&gt;

&lt;h2&gt;
  
  
  Step 4 — Ask the model for one JSON object, then distrust it
&lt;/h2&gt;

&lt;p&gt;Use an OpenAI-compatible chat API at a low temperature (0.1). The exact provider does not matter. The contract does: one JSON object, no markdown, no commentary, with a fixed set of keys.&lt;/p&gt;

&lt;p&gt;Categories I use:&lt;/p&gt;

&lt;p&gt;&lt;code&gt;ai_tools&lt;/code&gt;, &lt;code&gt;saas_alternative&lt;/code&gt;, &lt;code&gt;automation&lt;/code&gt;, &lt;code&gt;dev_tools&lt;/code&gt;, &lt;code&gt;marketing_growth&lt;/code&gt;&lt;/p&gt;

&lt;p&gt;The user message is boring on purpose: platform, title, and a body truncated to 2,000 characters. Long threads do not need to be fully scored; the question is usually in the first few paragraphs.&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight python"&gt;&lt;code&gt;&lt;span class="no"&gt;CATEGORIES&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;
    &lt;span class="s2"&gt;"ai_tools"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s2"&gt;"saas_alternative"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s2"&gt;"automation"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s2"&gt;"dev_tools"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="s2"&gt;"marketing_growth"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
&lt;span class="p"&gt;}&lt;/span&gt;
&lt;span class="no"&gt;URGENCIES&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;{&lt;/span&gt;&lt;span class="s2"&gt;"low"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"medium"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="s2"&gt;"high"&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;

&lt;span class="no"&gt;SYSTEM_PROMPT&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s2"&gt;"""
You are a B2B sales-intelligence analyst. You read a public post from Reddit,
Hacker News, Lobsters, Dev.to or Stack Exchange (Software Recommendations)
and decide whether its author shows REAL buying intent for software (SaaS, AI
tools, automation, developer tools, marketing/growth tools): actively searching
for a tool, asking for alternatives, frustrated with a current vendor, ready
to switch or pay.

Scoring guide for intent_score (integer 1-10):
1-3 = no purchase intent (news, opinion, self-promotion, launch announcement,
      hiring, rant without need)
4-6 = vague interest or early research
7-8 = clear need, actively evaluating solutions
9-10 = urgent, explicit intent to buy/switch now (budget, deadline,
       "need it this week")
Posts where the author promotes their OWN product must score &amp;lt;= 3.

Respond with ONLY one JSON object, no markdown, no commentary, exactly these keys:
{
  "intent_score": &amp;lt;integer 1-10&amp;gt;,
  "category": "ai_tools" | "saas_alternative" | "automation" | "dev_tools" | "marketing_growth",
  "problem_summary": "&amp;lt;one actionable sentence in English describing the buyer's need&amp;gt;",
  "mentioned_tools": ["&amp;lt;software/product names explicitly cited&amp;gt;"],
  "urgency": "low" | "medium" | "high"
}
"""&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;strip&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;


&lt;span class="k"&gt;def&lt;/span&gt; &lt;span class="nf"&gt;build_user_prompt&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;max_chars&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;2000&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
    &lt;span class="n"&gt;body&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="s2"&gt;"body"&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt;
    &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="nb"&gt;len&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;body&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;&amp;gt;&lt;/span&gt; &lt;span class="n"&gt;max_chars&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;
        &lt;span class="n"&gt;body&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;body&lt;/span&gt;&lt;span class="p"&gt;[&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="n"&gt;max_chars&lt;/span&gt;&lt;span class="p"&gt;]&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="s2"&gt;"…"&lt;/span&gt;
    &lt;span class="n"&gt;source&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;" ({post['source']})"&lt;/span&gt; &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="n"&gt;post&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;get&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s2"&gt;"source"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="k"&gt;else&lt;/span&gt; &lt;span class="s2"&gt;""&lt;/span&gt;
    &lt;span class="k"&gt;return&lt;/span&gt; &lt;span class="n"&gt;f&lt;/span&gt;&lt;span class="s2"&gt;"Platform: {post['platform']}{source}\nTitle: {post['title']}\nBody: {body or '(empty)'}"&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Models will still wrap the object in &lt;code&gt;&lt;/code&gt;`&lt;code&gt;json&lt;/code&gt; fences, emit smart quotes, leave a trailing comma, or add a &lt;code&gt;&amp;lt;think&amp;gt;&lt;/code&gt; block. Parse defensively, then &lt;strong&gt;validate&lt;/strong&gt;:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;code&gt;intent_score&lt;/code&gt; must be numeric; round it and clamp it to 1–10&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;category&lt;/code&gt; must be one of the five values above, or the whole analysis is invalid&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;problem_summary&lt;/code&gt; must be a string of at least a few characters (I cap it at 500)&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;urgency&lt;/code&gt; must be &lt;code&gt;low&lt;/code&gt;, &lt;code&gt;medium&lt;/code&gt;, or &lt;code&gt;high&lt;/code&gt;; if it is missing or malformed I default it to &lt;code&gt;low&lt;/code&gt; rather than dropping a otherwise-valid analysis&lt;/li&gt;
&lt;li&gt;
&lt;code&gt;mentioned_tools&lt;/code&gt; is a de-duplicated list of short strings (I cap each name at 60 characters and the list at 15)&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If validation fails, retry a few times with backoff. If it still fails, skip the post and move on. Do not insert a half-parsed row.&lt;/p&gt;

&lt;p&gt;`&lt;code&gt;&lt;/code&gt;python&lt;br&gt;
def validate_analysis(raw):&lt;br&gt;
    if not isinstance(raw, dict):&lt;br&gt;
        return None&lt;br&gt;
    try:&lt;br&gt;
        score = int(round(float(raw.get("intent_score"))))&lt;br&gt;
    except (TypeError, ValueError):&lt;br&gt;
        return None&lt;br&gt;
    score = min(10, max(1, score))&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;category = str(raw.get("category") or "").strip().lower().replace("-", "_").replace(" ", "_")
if category not in CATEGORIES:
    return None

summary = " ".join(str(raw.get("problem_summary") or "").split()).strip()
if len(summary) &amp;lt; 5:
    return None

urgency = str(raw.get("urgency") or "").strip().lower()
if urgency not in URGENCIES:
    urgency = "low"

tools, seen = [], set()
for item in raw.get("mentioned_tools") or []:
    if not isinstance(item, str):
        continue
    name = item.strip()[:60]
    key = name.lower()
    if not name or key in seen:
        continue
    seen.add(key)
    tools.append(name)
    if len(tools) &amp;gt;= 15:
        break

return {
    "intent_score": score,
    "category": category,
    "problem_summary": summary[:500],
    "mentioned_tools": tools,
    "urgency": urgency,
}
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;

&lt;p&gt;&lt;code&gt;&lt;/code&gt;`&lt;/p&gt;

&lt;p&gt;Only persist a signal when &lt;code&gt;intent_score &amp;gt;= 7&lt;/code&gt;. Still record the hash of rejects, so you never pay to score them again.&lt;/p&gt;

&lt;h2&gt;
  
  
  What this looks like on a real post
&lt;/h2&gt;

&lt;p&gt;This is a public Reddit thread, with the kind of structured result the scorer should return. The username is omitted on purpose.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Title:&lt;/strong&gt; Veeqo alternatives that arent tied to Amazon?&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;intent_score:&lt;/strong&gt; 8&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;category:&lt;/strong&gt; &lt;code&gt;saas_alternative&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;urgency:&lt;/strong&gt; &lt;code&gt;medium&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;mentioned_tools:&lt;/strong&gt; &lt;code&gt;["Veeqo", "Vesyl"]&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;problem_summary:&lt;/strong&gt; an ecommerce seller using Amazon-owned Veeqo wants a carrier- and channel-neutral order routing and shipping alternative, because Amazon is only 20% of their sales and they are uneasy routing order data through a competitor.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;That is a keep: a named current tool, a specific constraint, and an active search for a replacement. A "what's the best AI tool?" post with no constraint would land in the 4–6 band and be dropped.&lt;/p&gt;

&lt;h2&gt;
  
  
  A script you can actually run
&lt;/h2&gt;

&lt;p&gt;Save the complete file as &lt;code&gt;detect_intent.py&lt;/code&gt;, then:&lt;/p&gt;

&lt;p&gt;`&lt;code&gt;&lt;/code&gt;bash&lt;br&gt;
pip install requests openai&lt;br&gt;
export OPENAI_API_KEY=sk-...&lt;br&gt;
export OPENAI_MODEL=gpt-4o-mini   # or whatever your provider expects&lt;/p&gt;

&lt;h1&gt;
  
  
  Optional, if you use an OpenAI-compatible router instead of api.openai.com:
&lt;/h1&gt;

&lt;h1&gt;
  
  
  export OPENAI_BASE_URL=&lt;a href="https://api.example.com/v1" rel="noopener noreferrer"&gt;https://api.example.com/v1&lt;/a&gt;
&lt;/h1&gt;

&lt;p&gt;python detect_intent.py&lt;br&gt;
&lt;code&gt;&lt;/code&gt;`&lt;/p&gt;

&lt;p&gt;The script fetches about 36 hours of HN / Dev.to / Software Recommendations posts, pre-filters, scores at most 15 candidates, and prints anything that meets the threshold. It writes analysed hashes to &lt;code&gt;processed_hashes.json&lt;/code&gt;.&lt;/p&gt;

&lt;p&gt;`&lt;code&gt;&lt;/code&gt;python&lt;/p&gt;

&lt;h1&gt;
  
  
  !/usr/bin/env python3
&lt;/h1&gt;

&lt;p&gt;"""Detect B2B buying intent in public HN, Dev.to and Software Recs posts."""&lt;/p&gt;

&lt;p&gt;from &lt;strong&gt;future&lt;/strong&gt; import annotations&lt;/p&gt;

&lt;p&gt;import hashlib&lt;br&gt;
import json&lt;br&gt;
import os&lt;br&gt;
import re&lt;br&gt;
import time&lt;br&gt;
from datetime import datetime, timezone&lt;br&gt;
from pathlib import Path&lt;/p&gt;

&lt;p&gt;import requests&lt;br&gt;
from openai import OpenAI&lt;/p&gt;

&lt;p&gt;LOOKBACK_HOURS = 36&lt;br&gt;
MAX_AI_CALLS = 15&lt;br&gt;
MIN_INTENT_THRESHOLD = 7&lt;br&gt;
BODY_MAX_CHARS = 2000&lt;br&gt;
USER_AGENT = "BuyingIntentTutorial/1.0 (dev.to article)"&lt;br&gt;
HEADERS = {"User-Agent": USER_AGENT, "Accept": "application/json"}&lt;br&gt;
HASH_FILE = Path("processed_hashes.json")&lt;/p&gt;

&lt;p&gt;KEYWORDS = [&lt;br&gt;
    "alternative to",&lt;br&gt;
    "looking for",&lt;br&gt;
    "recommend a",&lt;br&gt;
    "any suggestions",&lt;br&gt;
    "any recommendations",&lt;br&gt;
    "suggestions for",&lt;br&gt;
    "switching from",&lt;br&gt;
    "migrate from",&lt;br&gt;
    "budget for",&lt;br&gt;
    "willing to pay",&lt;br&gt;
    "ready to pay",&lt;br&gt;
    "need a tool",&lt;br&gt;
    "need software",&lt;br&gt;
    "need a saas",&lt;br&gt;
    "looking to replace",&lt;br&gt;
    "replacement for",&lt;br&gt;
    "frustrated with",&lt;br&gt;
    "can anyone recommend",&lt;br&gt;
    "seeking recommendations",&lt;br&gt;
    "in the market for",&lt;br&gt;
    "software for",&lt;br&gt;
    "best tool",&lt;br&gt;
    "how do you automate",&lt;br&gt;
    "what tool do you",&lt;br&gt;
    "tool recommendations",&lt;br&gt;
]&lt;/p&gt;

&lt;p&gt;INTENT_SEARCH_PHRASES = (&lt;br&gt;
    '"looking for"',&lt;br&gt;
    '"alternative to"',&lt;br&gt;
    '"recommend a"',&lt;br&gt;
    '"any suggestions"',&lt;br&gt;
    '"any recommendations"',&lt;br&gt;
    '"switching from"',&lt;br&gt;
    '"willing to pay"',&lt;br&gt;
    '"ready to pay"',&lt;br&gt;
    '"need a tool"',&lt;br&gt;
    '"need software"',&lt;br&gt;
    '"replacement for"',&lt;br&gt;
    '"looking to replace"',&lt;br&gt;
    '"can anyone recommend"',&lt;br&gt;
    '"in the market for"',&lt;br&gt;
    '"frustrated with"',&lt;br&gt;
)&lt;/p&gt;

&lt;p&gt;CATEGORIES = {&lt;br&gt;
    "ai_tools",&lt;br&gt;
    "saas_alternative",&lt;br&gt;
    "automation",&lt;br&gt;
    "dev_tools",&lt;br&gt;
    "marketing_growth",&lt;br&gt;
}&lt;br&gt;
URGENCIES = {"low", "medium", "high"}&lt;/p&gt;

&lt;p&gt;SOFTWARE_QUALIFIER = re.compile(&lt;br&gt;
    r"\b(saas|software|tool|tools|app|apps|platform|crm|api|"&lt;br&gt;
    r"automat(?:e|ion|ing)?|subscription|vendor|workspace|plugin|"&lt;br&gt;
    r"integration|dashboard|productiv(?:e|ity)|analytics|billing|"&lt;br&gt;
    r"invoice|payroll|onboard(?:ing)?)\b",&lt;br&gt;
    re.I,&lt;br&gt;
)&lt;br&gt;
BUYING_QUESTION_CUES = [&lt;br&gt;
    re.compile(r"\?"),&lt;br&gt;
    re.compile(r"\bwhat (?:do you|would you|should i|can i)\b", re.I),&lt;br&gt;
    re.compile(r"\bhow do (?:you|i|we)\b", re.I),&lt;br&gt;
    re.compile(r"\b(?:anyone|any one) (?:here )?(?:use|using|know|tried|recommend)", re.I),&lt;br&gt;
    re.compile(r"\binstead of\b", re.I),&lt;br&gt;
    re.compile(r"\brecommend(?:ation)?s?\b", re.I),&lt;br&gt;
    re.compile(r"\balternative\b", re.I),&lt;br&gt;
    re.compile(r"\bwhat(?:'s| is) the best\b", re.I),&lt;br&gt;
    re.compile(r"\blooking to\b", re.I),&lt;br&gt;
    re.compile(r"\bneed (?:help finding|something|an? (?:tool|app|software|saas|platform))", re.I),&lt;br&gt;
    re.compile(r"\bswitch(?:ing)? (?:from|to)\b", re.I),&lt;br&gt;
    re.compile(r"\breplace\b", re.I),&lt;br&gt;
]&lt;/p&gt;

&lt;p&gt;SYSTEM_PROMPT = """&lt;br&gt;
You are a B2B sales-intelligence analyst. You read a public post from Reddit,&lt;br&gt;
Hacker News, Lobsters, Dev.to or Stack Exchange (Software Recommendations)&lt;br&gt;
and decide whether its author shows REAL buying intent for software (SaaS, AI&lt;br&gt;
tools, automation, developer tools, marketing/growth tools): actively searching&lt;br&gt;
for a tool, asking for alternatives, frustrated with a current vendor, ready&lt;br&gt;
to switch or pay.&lt;/p&gt;

&lt;p&gt;Scoring guide for intent_score (integer 1-10):&lt;br&gt;
1-3 = no purchase intent (news, opinion, self-promotion, launch announcement,&lt;br&gt;
      hiring, rant without need)&lt;br&gt;
4-6 = vague interest or early research&lt;br&gt;
7-8 = clear need, actively evaluating solutions&lt;br&gt;
9-10 = urgent, explicit intent to buy/switch now (budget, deadline,&lt;br&gt;
       "need it this week")&lt;br&gt;
Posts where the author promotes their OWN product must score &amp;lt;= 3.&lt;/p&gt;

&lt;p&gt;Respond with ONLY one JSON object, no markdown, no commentary, exactly these keys:&lt;br&gt;
{&lt;br&gt;
  "intent_score": ,&lt;br&gt;
  "category": "ai_tools" | "saas_alternative" | "automation" | "dev_tools" | "marketing_growth",&lt;br&gt;
  "problem_summary": "",&lt;br&gt;
  "mentioned_tools": [""],&lt;br&gt;
  "urgency": "low" | "medium" | "high"&lt;br&gt;
}&lt;br&gt;
""".strip()&lt;/p&gt;

&lt;p&gt;def strip_html(text: str) -&amp;gt; str:&lt;br&gt;
    text = re.sub(r"&amp;lt;[^&amp;gt;]+&amp;gt;", " ", text or "")&lt;br&gt;
    return re.sub(r"\s+", " ", text).strip()&lt;/p&gt;

&lt;p&gt;def get_json(url: str, params=None):&lt;br&gt;
    res = requests.get(url, headers=HEADERS, params=params, timeout=15)&lt;br&gt;
    res.raise_for_status()&lt;br&gt;
    return res.json()&lt;/p&gt;

&lt;p&gt;def source_hash(post: dict) -&amp;gt; str:&lt;br&gt;
    return hashlib.sha256(f"{post['platform']}:{post['external_id']}".encode()).hexdigest()&lt;/p&gt;

&lt;p&gt;def load_hashes() -&amp;gt; set[str]:&lt;br&gt;
    if HASH_FILE.exists():&lt;br&gt;
        return set(json.loads(HASH_FILE.read_text()))&lt;br&gt;
    return set()&lt;/p&gt;

&lt;p&gt;def save_hashes(hashes: set[str]) -&amp;gt; None:&lt;br&gt;
    HASH_FILE.write_text(json.dumps(sorted(hashes), indent=2))&lt;/p&gt;

&lt;p&gt;def fetch_hn(since: int, until: int) -&amp;gt; list[dict]:&lt;br&gt;
    posts, seen = [], set()&lt;br&gt;
    for phrase in INTENT_SEARCH_PHRASES:&lt;br&gt;
        for tags in ("ask_hn", "story", "comment"):&lt;br&gt;
            for page in range(2):&lt;br&gt;
                data = get_json(&lt;br&gt;
                    "&lt;a href="https://hn.algolia.com/api/v1/search_by_date" rel="noopener noreferrer"&gt;https://hn.algolia.com/api/v1/search_by_date&lt;/a&gt;",&lt;br&gt;
                    {&lt;br&gt;
                        "query": phrase,&lt;br&gt;
                        "tags": tags,&lt;br&gt;
                        "numericFilters": f"created_at_i&amp;gt;={since},created_at_i&amp;lt;{until}",&lt;br&gt;
                        "hitsPerPage": 50,&lt;br&gt;
                        "page": page,&lt;br&gt;
                    },&lt;br&gt;
                )&lt;br&gt;
                hits = data.get("hits") or []&lt;br&gt;
                for hit in hits:&lt;br&gt;
                    object_id = hit.get("objectID")&lt;br&gt;
                    if not object_id or object_id in seen:&lt;br&gt;
                        continue&lt;br&gt;
                    title = (hit.get("title") or hit.get("story_title") or "HN comment").strip()&lt;br&gt;
                    body = hit.get("comment_text") or hit.get("story_text") or ""&lt;br&gt;
                    if tags == "comment" and not SOFTWARE_QUALIFIER.search(f"{title}\n{body}"):&lt;br&gt;
                        continue&lt;br&gt;
                    seen.add(object_id)&lt;br&gt;
                    posts.append(&lt;br&gt;
                        {&lt;br&gt;
                            "platform": "hackernews",&lt;br&gt;
                            "external_id": str(object_id),&lt;br&gt;
                            "source": "hackernews",&lt;br&gt;
                            "url": f"&lt;a href="https://news.ycombinator.com/item?id=%7Bobject_id%7D" rel="noopener noreferrer"&gt;https://news.ycombinator.com/item?id={object_id}&lt;/a&gt;",&lt;br&gt;
                            "title": title,&lt;br&gt;
                            "body": strip_html(body),&lt;br&gt;
                            "author": hit.get("author"),&lt;br&gt;
                        }&lt;br&gt;
                    )&lt;br&gt;
                if not hits or page + 1 &amp;gt;= data.get("nbPages", 1):&lt;br&gt;
                    break&lt;br&gt;
                time.sleep(0.15)&lt;br&gt;
    return posts&lt;/p&gt;

&lt;p&gt;def fetch_devto(since_ms: int, until_ms: int) -&amp;gt; list[dict]:&lt;br&gt;
    posts, seen = [], set()&lt;br&gt;
    for tag in ("help", "discuss"):&lt;br&gt;
        reached_old = False&lt;br&gt;
        for page in range(1, 6):&lt;br&gt;
            items = get_json(&lt;br&gt;
                "&lt;a href="https://dev.to/api/articles"&gt;https://dev.to/api/articles&lt;/a&gt;",&lt;br&gt;
                {"per_page": 30, "page": page, "tag": tag},&lt;br&gt;
            )&lt;br&gt;
            if not items:&lt;br&gt;
                break&lt;br&gt;
            for item in items:&lt;br&gt;
                item_id = item.get("id")&lt;br&gt;
                published = item.get("published_at")&lt;br&gt;
                posted_ms = (&lt;br&gt;
                    datetime.fromisoformat(published.replace("Z", "+00:00")).timestamp() * 1000&lt;br&gt;
                    if published&lt;br&gt;
                    else None&lt;br&gt;
                )&lt;br&gt;
                if posted_ms is not None and posted_ms &amp;lt; since_ms:&lt;br&gt;
                    reached_old = True&lt;br&gt;
                    continue&lt;br&gt;
                if not item_id or item_id in seen:&lt;br&gt;
                    continue&lt;br&gt;
                if posted_ms is not None and not (since_ms &amp;lt;= posted_ms &amp;lt; until_ms):&lt;br&gt;
                    continue&lt;br&gt;
                seen.add(item_id)&lt;br&gt;
                username = (item.get("user") or {}).get("username") or "dev"&lt;br&gt;
                posts.append(&lt;br&gt;
                    {&lt;br&gt;
                        "platform": "devto",&lt;br&gt;
                        "external_id": str(item_id),&lt;br&gt;
                        "source": "devto",&lt;br&gt;
                        "url": item.get("url") or f"&lt;a href="https://dev.to/%7Busername%7D/%7Bitem.get('slug"&gt;https://dev.to/{username}/{item.get('slug&lt;/a&gt;', item_id)}",&lt;br&gt;
                        "title": (item.get("title") or "").strip(),&lt;br&gt;
                        "body": (item.get("description") or "").strip(),&lt;br&gt;
                        "author": username,&lt;br&gt;
                    }&lt;br&gt;
                )&lt;br&gt;
            if reached_old:&lt;br&gt;
                break&lt;br&gt;
            time.sleep(0.8)&lt;br&gt;
    return posts&lt;/p&gt;

&lt;p&gt;def fetch_software_recs(since: int, until: int) -&amp;gt; list[dict]:&lt;br&gt;
    posts = []&lt;br&gt;
    for page in range(1, 4):&lt;br&gt;
        data = get_json(&lt;br&gt;
            "&lt;a href="https://api.stackexchange.com/2.3/questions" rel="noopener noreferrer"&gt;https://api.stackexchange.com/2.3/questions&lt;/a&gt;",&lt;br&gt;
            {&lt;br&gt;
                "order": "desc",&lt;br&gt;
                "sort": "creation",&lt;br&gt;
                "site": "softwarerecs",&lt;br&gt;
                "pagesize": 50,&lt;br&gt;
                "page": page,&lt;br&gt;
                "filter": "withbody",&lt;br&gt;
                "fromdate": since,&lt;br&gt;
                "todate": until,&lt;br&gt;
            },&lt;br&gt;
        )&lt;br&gt;
        for item in data.get("items") or []:&lt;br&gt;
            question_id = item.get("question_id")&lt;br&gt;
            if not question_id or not item.get("title"):&lt;br&gt;
                continue&lt;br&gt;
            owner = (item.get("owner") or {}).get("display_name")&lt;br&gt;
            posts.append(&lt;br&gt;
                {&lt;br&gt;
                    "platform": "stackexchange",&lt;br&gt;
                    "external_id": str(question_id),&lt;br&gt;
                    "source": "softwarerecs",&lt;br&gt;
                    "url": item.get("link")&lt;br&gt;
                    or f"&lt;a href="https://softwarerecs.stackexchange.com/questions/%7Bquestion_id%7D" rel="noopener noreferrer"&gt;https://softwarerecs.stackexchange.com/questions/{question_id}&lt;/a&gt;",&lt;br&gt;
                    "title": item["title"].strip(),&lt;br&gt;
                    "body": strip_html(item.get("body") or ""),&lt;br&gt;
                    "author": owner,&lt;br&gt;
                }&lt;br&gt;
            )&lt;br&gt;
        if not data.get("has_more"):&lt;br&gt;
            break&lt;br&gt;
        time.sleep(0.8)&lt;br&gt;
    return posts&lt;/p&gt;

&lt;p&gt;def match_keywords(text: str, keywords: list[str]) -&amp;gt; list[str]:&lt;br&gt;
    haystack = " ".join(text.lower().split())&lt;br&gt;
    return [k for k in keywords if k.lower() in haystack]&lt;/p&gt;

&lt;p&gt;def is_loose_buying_candidate(title: str, body: str) -&amp;gt; bool:&lt;br&gt;
    text = f"{title}\n{body}"&lt;br&gt;
    return bool(SOFTWARE_QUALIFIER.search(text) and any(cue.search(text) for cue in BUYING_QUESTION_CUES))&lt;/p&gt;

&lt;p&gt;def to_candidates(posts: list[dict]) -&amp;gt; list[dict]:&lt;br&gt;
    keyword_hits, loose_hits = [], []&lt;br&gt;
    for post in posts:&lt;br&gt;
        if post["platform"] == "stackexchange" and post["source"] == "softwarerecs":&lt;br&gt;
            keyword_hits.append({&lt;strong&gt;post, "matched_keywords": ["softwarerecs"]})&lt;br&gt;
            continue&lt;br&gt;
        matched = match_keywords(f"{post['title']}\n{post['body']}", KEYWORDS)&lt;br&gt;
        if matched:&lt;br&gt;
            keyword_hits.append({&lt;/strong&gt;post, "matched_keywords": matched})&lt;br&gt;
        elif is_loose_buying_candidate(post["title"], post["body"]):&lt;br&gt;
            loose_hits.append({**post, "matched_keywords": ["loose_intent"]})&lt;br&gt;
    rank = lambda p: (-len(p["matched_keywords"]), -len(p["body"]))&lt;br&gt;
    keyword_hits.sort(key=rank)&lt;br&gt;
    loose_hits.sort(key=rank)&lt;br&gt;
    return keyword_hits + loose_hits&lt;/p&gt;

&lt;p&gt;def parse_json_safe(raw: str | None):&lt;br&gt;
    if not raw:&lt;br&gt;
        return None&lt;br&gt;
    text = re.sub(r"[\s\S]&lt;em&gt;?&lt;/em&gt;&lt;em&gt;", "", raw, flags=re.I).strip()&lt;br&gt;
    candidates = re.findall(r"&lt;code&gt;&lt;/code&gt;`(?:json)?\s&lt;/em&gt;([\s\S]&lt;em&gt;?)`&lt;code&gt;&lt;/code&gt;", text, flags=re.I)&lt;br&gt;
    candidates.append(text)&lt;br&gt;
    start = text.find("{")&lt;br&gt;
    end = text.rfind("}")&lt;br&gt;
    if start != -1 and end &amp;gt; start:&lt;br&gt;
        candidates.append(text[start : end + 1])&lt;br&gt;
    for candidate in candidates:&lt;br&gt;
        repaired = candidate.replace("“", '"').replace("”", '"').replace("’", "'")&lt;br&gt;
        repaired = re.sub(r",\s&lt;/em&gt;([}]])", r"\1", repaired)&lt;br&gt;
        try:&lt;br&gt;
            return json.loads(repaired)&lt;br&gt;
        except json.JSONDecodeError:&lt;br&gt;
            continue&lt;br&gt;
    return None&lt;/p&gt;

&lt;p&gt;def validate_analysis(raw) -&amp;gt; dict | None:&lt;br&gt;
    if not isinstance(raw, dict):&lt;br&gt;
        return None&lt;br&gt;
    try:&lt;br&gt;
        score = min(10, max(1, int(round(float(raw.get("intent_score"))))))&lt;br&gt;
    except (TypeError, ValueError):&lt;br&gt;
        return None&lt;br&gt;
    category = str(raw.get("category") or "").strip().lower().replace("-", "&lt;em&gt;").replace(" ", "&lt;/em&gt;")&lt;br&gt;
    if category not in CATEGORIES:&lt;br&gt;
        return None&lt;br&gt;
    summary = " ".join(str(raw.get("problem_summary") or "").split()).strip()&lt;br&gt;
    if len(summary) &amp;lt; 5:&lt;br&gt;
        return None&lt;br&gt;
    urgency = str(raw.get("urgency") or "").strip().lower()&lt;br&gt;
    if urgency not in URGENCIES:&lt;br&gt;
        urgency = "low"&lt;br&gt;
    tools, seen = [], set()&lt;br&gt;
    for item in raw.get("mentioned_tools") or []:&lt;br&gt;
        if not isinstance(item, str):&lt;br&gt;
            continue&lt;br&gt;
        name = item.strip()[:60]&lt;br&gt;
        if not name or name.lower() in seen:&lt;br&gt;
            continue&lt;br&gt;
        seen.add(name.lower())&lt;br&gt;
        tools.append(name)&lt;br&gt;
        if len(tools) &amp;gt;= 15:&lt;br&gt;
            break&lt;br&gt;
    return {&lt;br&gt;
        "intent_score": score,&lt;br&gt;
        "category": category,&lt;br&gt;
        "problem_summary": summary[:500],&lt;br&gt;
        "mentioned_tools": tools,&lt;br&gt;
        "urgency": urgency,&lt;br&gt;
    }&lt;/p&gt;

&lt;p&gt;def analyse(client: OpenAI, model: str, post: dict) -&amp;gt; dict | None:&lt;br&gt;
    body = post["body"]&lt;br&gt;
    if len(body) &amp;gt; BODY_MAX_CHARS:&lt;br&gt;
        body = body[:BODY_MAX_CHARS] + "…"&lt;br&gt;
    source = f" ({post['source']})" if post.get("source") else ""&lt;br&gt;
    user = f"Platform: {post['platform']}{source}\nTitle: {post['title']}\nBody: {body or '(empty)'}"&lt;br&gt;
    completion = client.chat.completions.create(&lt;br&gt;
        model=model,&lt;br&gt;
        temperature=0.1,&lt;br&gt;
        max_tokens=2000,&lt;br&gt;
        messages=[&lt;br&gt;
            {"role": "system", "content": SYSTEM_PROMPT},&lt;br&gt;
            {"role": "user", "content": user},&lt;br&gt;
        ],&lt;br&gt;
    )&lt;br&gt;
    parsed = parse_json_safe(completion.choices[0].message.content)&lt;br&gt;
    return validate_analysis(parsed)&lt;/p&gt;

&lt;p&gt;def main() -&amp;gt; None:&lt;br&gt;
    now = datetime.now(timezone.utc)&lt;br&gt;
    until = int(now.timestamp())&lt;br&gt;
    since = until - LOOKBACK_HOURS * 3600&lt;br&gt;
    since_ms, until_ms = since * 1000, until * 1000&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;posts = fetch_hn(since, until) + fetch_devto(since_ms, until_ms) + fetch_software_recs(since, until)
known = load_hashes()
fresh = [p for p in posts if source_hash(p) not in known]
queued = to_candidates(fresh)[:MAX_AI_CALLS]
print(f"fetched={len(posts)} new={len(fresh)} to_analyse={len(queued)}")

client = OpenAI(base_url=os.getenv("OPENAI_BASE_URL") or None)
model = os.getenv("OPENAI_MODEL") or "gpt-4o-mini"
kept = 0
for post in queued:
    digest = source_hash(post)
    try:
        analysis = analyse(client, model, post)
    except Exception as exc:
        print(f"failed {post['url']}: {exc}")
        continue
    known.add(digest)
    if not analysis:
        print(f"invalid JSON {post['url']}")
        continue
    if analysis["intent_score"] &amp;lt; MIN_INTENT_THRESHOLD:
        print(f"reject {analysis['intent_score']} {post['title'][:80]}")
        continue
    kept += 1
    print("=" * 72)
    print(f"{analysis['intent_score']}/10  {analysis['category']}  {analysis['urgency']}")
    print(post["title"])
    print(analysis["problem_summary"])
    print("tools:", ", ".join(analysis["mentioned_tools"]) or "(none)")
    print(post["url"])

save_hashes(known)
print(f"kept={kept}  remembered_hashes={len(known)}")
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;

&lt;p&gt;if &lt;strong&gt;name&lt;/strong&gt; == "&lt;strong&gt;main&lt;/strong&gt;":&lt;br&gt;
    main()&lt;br&gt;
&lt;code&gt;&lt;/code&gt;`&lt;/p&gt;

&lt;p&gt;If you put this on a schedule, keep the per-run cap, persist hashes, and do not poll faster than the forums themselves update. Be conservative in how you follow up: these are public posts, not a license to spam. Read the thread, and if you reach out, talk about the actual problem in the post.&lt;/p&gt;

&lt;p&gt;A useful later upgrade, which I run in production on Sundays, is to look at yield (accepted vs analysed) by source and keyword, and adjust the keyword list and the threshold with guardrails — for example a minimum and maximum number of keywords, and no more than a one-point threshold change at a time. Do not let a model freely rewrite the strategy without those bounds.&lt;/p&gt;

&lt;h2&gt;
  
  
  If you want the feed without running the collector
&lt;/h2&gt;

&lt;p&gt;I expose the same scored feed as &lt;a href="https://rapidapi.com/JamesBasset/api/b2b-buying-intent-signals" rel="noopener noreferrer"&gt;B2B Buying Intent Signals&lt;/a&gt; on RapidAPI. Collection covers Reddit, Hacker News, Lobsters, Dev.to, and Software Recommendations, every six hours. The default list is score 7+. At the time of writing the live feed holds ~100 high-intent B2B signals.&lt;/p&gt;

&lt;p&gt;&lt;code&gt;`bash&lt;br&gt;
curl --request GET \&lt;br&gt;
  --url 'https://b2b-buying-intent-signals.p.rapidapi.com/api/v1/signals?tool=shopify&amp;amp;min_score=7&amp;amp;limit=10' \&lt;br&gt;
  --header 'X-RapidAPI-Host: b2b-buying-intent-signals.p.rapidapi.com' \&lt;br&gt;
  --header 'X-RapidAPI-Key: YOUR_RAPIDAPI_KEY'&lt;br&gt;
`&lt;/code&gt;&lt;/p&gt;

&lt;p&gt;You can also filter by &lt;code&gt;tool&lt;/code&gt; (a case-insensitive exact match on &lt;code&gt;mentioned_tools&lt;/code&gt;, so &lt;code&gt;tool=shopify&lt;/code&gt; matches &lt;code&gt;"Shopify"&lt;/code&gt;), &lt;code&gt;category&lt;/code&gt;, &lt;code&gt;platform&lt;/code&gt;, and &lt;code&gt;urgency&lt;/code&gt;. Tools that actually appear in the current feed include Shopify, Supabase, Veeqo, Vesyl, Stripe, Railway, Lovable, Bolt and GitHub — useful starting filters if you sell into ecommerce, payments, or hosting. There is a free BASIC plan (200 calls/month) if you want to inspect real payloads before deciding whether to run any of this yourself.&lt;/p&gt;

&lt;p&gt;If you would rather stay in code, the script above is the whole idea: cheap filters first, a strict JSON contract second, and a threshold that is allowed to say "no."&lt;/p&gt;

</description>
      <category>python</category>
      <category>llm</category>
      <category>api</category>
      <category>ai</category>
    </item>
  </channel>
</rss>
