<?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: consolata gicheru</title>
    <description>The latest articles on DEV Community by consolata gicheru (@consolatacleah).</description>
    <link>https://dev.to/consolatacleah</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%2F439820%2F1f79126b-3fe1-48db-8ec7-671968140802.jpeg</url>
      <title>DEV Community: consolata gicheru</title>
      <link>https://dev.to/consolatacleah</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/consolatacleah"/>
    <language>en</language>
    <item>
      <title>Claude Meets Copilot Studio</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Sun, 08 Nov 2026 17:06:45 +0000</pubDate>
      <link>https://dev.to/consolatacleah/claude-meets-copilot-studio-4pk0</link>
      <guid>https://dev.to/consolatacleah/claude-meets-copilot-studio-4pk0</guid>
      <description>&lt;p&gt;This is the one I've been building toward.&lt;/p&gt;

&lt;p&gt;Over the last few posts we deployed Claude in Microsoft Foundry, said hello to it from code, and taught it to turn a messy support ticket into clean, structured data. All useful, all a bit back-end. Today it comes together into something you can actually hand to a colleague: a working agent, built in Copilot Studio, the low-code tool you already know, powered by Claude.&lt;/p&gt;

&lt;p&gt;This is the moment the two halves of a Power Platform person's brain finally meet. The business side that knows what a good customer answer looks like, and the AI side that can produce one at scale. And the nicest part is how little new there is to learn, because most of it happens in a tool you've probably already got open in another tab.&lt;/p&gt;

&lt;h2&gt;
  
  
  The pleasant surprise
&lt;/h2&gt;

&lt;p&gt;I spent the first few posts carefully wiring Claude up by hand in Foundry, so I braced myself for a fiddly connection step here. There isn't one.&lt;/p&gt;

&lt;p&gt;Claude is already in Copilot Studio. Microsoft opened the tool up to more than one model, and Anthropic's models sit right there in the picker alongside the OpenAI ones. You choose Claude from a dropdown, the same way you'd choose a connector or a data source. No endpoint to paste, no key to manage, no code.&lt;/p&gt;

&lt;p&gt;If you've spent years in the Power Platform, sit with that for a second. A frontier model, available as a dropdown, next to the tools you already use every day. That is the whole promise of this series in one small menu.&lt;/p&gt;

&lt;h2&gt;
  
  
  One thing to check before you start
&lt;/h2&gt;

&lt;p&gt;There's a catch, and it's worth knowing up front so you don't waste an hour wondering why Claude isn't in your dropdown.&lt;/p&gt;

&lt;p&gt;An admin has to turn Anthropic's models on first, in the Microsoft 365 Admin Center. Once that switch is flipped, the models are on by default in the Power Platform Admin Center, where an admin can decide which environments makers can use them in. So this is a tenant-level decision, not just a personal one.&lt;/p&gt;

&lt;p&gt;The sneaky part: if Anthropic models are switched off, an agent you built on Claude doesn't error. It quietly falls back to a different model. So if your agent suddenly sounds different one morning, this is the first place to look. If you run your own tenant, flip the switch and move on. If you don't, this is the one short conversation to have with whoever does before you build.&lt;/p&gt;

&lt;h2&gt;
  
  
  Build the agent, and point it at Claude
&lt;/h2&gt;

&lt;p&gt;Now the fun part, and it's genuinely about ten minutes.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Go to &lt;strong&gt;copilotstudio.microsoft.com&lt;/strong&gt; and select &lt;strong&gt;Agents &amp;gt; New agent&lt;/strong&gt;. You can let Copilot interview you, but I like &lt;strong&gt;Skip to configure&lt;/strong&gt; so I can fill in the fields myself.&lt;/li&gt;
&lt;li&gt;Give it a name and a short description. Mine is &lt;strong&gt;Harbor Help&lt;/strong&gt;, "the first line of support for Harbor Supply Co. customers."&lt;/li&gt;
&lt;li&gt;In the &lt;strong&gt;Instructions&lt;/strong&gt;, tell it who it is and how to behave. This is the system prompt, and it's where your business judgement lives. Something like: &lt;em&gt;You are Harbor Supply Co.'s support agent. Be warm and brief. Confirm order numbers back to the customer. If a request involves damaged goods or a missed deadline, treat it as urgent and offer to escalate to a human.&lt;/em&gt;
&lt;/li&gt;
&lt;li&gt;Open the agent's &lt;strong&gt;settings&lt;/strong&gt; and set its primary model to &lt;strong&gt;Claude&lt;/strong&gt;. That's the whole integration. From now on, the reasoning behind every reply is Claude's.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;That last step is the one that would have been a week of work a couple of years ago. Today it's a dropdown. If you'd rather mix models, you can also pick a model per prompt, so a cheap task and a hard one can use different models in the same agent, but for today, one good model everywhere is plenty.&lt;/p&gt;

&lt;h2&gt;
  
  
  Give it a bit of Harbor's world
&lt;/h2&gt;

&lt;p&gt;Right now the agent is clever but empty. Ask it about a specific order and it will cheerfully make something up, because it doesn't know anything about Harbor beyond what you put in the instructions.&lt;/p&gt;

&lt;p&gt;For today, a little goes a long way. Add one knowledge source, a short document with Harbor's returns and replacement policy, so the agent has real facts to lean on. In Copilot Studio that's the &lt;strong&gt;Knowledge&lt;/strong&gt; section: point it at a file or a site and it will use it when answering. Even a one-page policy is enough to stop the worst of the guessing.&lt;/p&gt;

&lt;p&gt;I'm keeping this deliberately light, because grounding is a proper topic in its own right, and it's next time's whole post. Today the goal is just to get the agent answering in Harbor's voice, with enough real context that it isn't inventing the company from scratch.&lt;/p&gt;

&lt;h2&gt;
  
  
  Watch it answer
&lt;/h2&gt;

&lt;p&gt;Here's the difference the whole exercise buys you. Same question, Dawn from the last post, asking about her bent racking.&lt;/p&gt;

&lt;p&gt;A raw, empty chatbot gives you the beige non-answer we've all seen:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;Dawn:&lt;/strong&gt; 2 of the uprights on order HS-44821 came bent, I need them replaced before the weekend.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Generic bot:&lt;/strong&gt; I'm sorry to hear about your experience. Please contact our support team and we will be happy to assist you.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;That helps no one. Now the Harbor Help agent, same question, with its instructions and that one policy document behind it:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;Dawn:&lt;/strong&gt; 2 of the uprights on order HS-44821 came bent, I need them replaced before the weekend.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Harbor Help:&lt;/strong&gt; Thanks Dawn, I've got order HS-44821. Damaged uprights are covered by our replacement policy, and because this is blocking a weekend move I'm flagging it as urgent. I'm connecting you to a person now who can get replacements out to you. Is the delivery address on the order still correct?&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;Same model, same ten minutes of setup. The second answer knows the order, applies the real policy, spots the urgency, and does the one thing the first one refused to: it actually moves things forward.&lt;/p&gt;

&lt;h2&gt;
  
  
  So why did we bother with Foundry?
&lt;/h2&gt;

&lt;p&gt;Fair question. If Claude is just a dropdown in Copilot Studio, why did we spend three posts deploying it in Foundry and calling it from code?&lt;/p&gt;

&lt;p&gt;Because they're two doors to the same room, for two different jobs.&lt;/p&gt;

&lt;p&gt;The &lt;strong&gt;Copilot Studio picker&lt;/strong&gt; is the fast door. Low-code, no plumbing, perfect for an agent a business team will own and tweak. You trade some control for speed, and most of the time that's exactly the right trade.&lt;/p&gt;

&lt;p&gt;The &lt;strong&gt;Foundry route&lt;/strong&gt; from the earlier posts is the deep door. You call Claude from your own code, shape the request precisely, enforce a strict schema like we did with the tickets, and wire it into systems Copilot Studio never touches. More control, more responsibility.&lt;/p&gt;

&lt;p&gt;The quiet superpower of this series is that you don't have to choose a camp. A Copilot Studio agent can call a custom prompt or an action that runs your Foundry code underneath, so the low-code front and the pro-code engine live in the same solution. You reach for whichever door the job in front of you needs.&lt;/p&gt;

&lt;h2&gt;
  
  
  The two halves, finally in one place
&lt;/h2&gt;

&lt;p&gt;Step back and look at what you've got. An agent that reasons with a frontier model, answers in your company's voice, knows when to pull in a human, and lives in the same low-code tool you've been using for years. The business-process brain and the AI brain, in one place, on your turf.&lt;/p&gt;

&lt;p&gt;That's the whole thesis of this series, and you just built it in an afternoon.&lt;/p&gt;

&lt;p&gt;The agent still has one honest weakness: it only knows the little we told it. Ask something outside that one policy document and it's back to guessing. So next time we fix that properly. We ground the agent in Harbor's real world, its policies, its catalog, its orders, so its answers come from facts instead of vibes. That's the difference between a demo and something you'd actually let talk to a customer.&lt;/p&gt;

&lt;p&gt;If you build your own version, I'd genuinely love to see it. Show me what your agent said back.&lt;/p&gt;

&lt;p&gt;See you next time. 😊&lt;/p&gt;




&lt;p&gt;&lt;strong&gt;Sources&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;a href="https://www.microsoft.com/en-us/copilot/blog/copilot-studio/anthropic-joins-the-multi-model-lineup-in-microsoft-copilot-studio/" rel="noopener noreferrer"&gt;Anthropic joins the multi-model lineup in Microsoft Copilot Studio — Microsoft&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href="https://learn.microsoft.com/en-us/microsoft-copilot-studio/bring-your-own-model-prompts" rel="noopener noreferrer"&gt;Bring your own model for your prompts — Microsoft Learn&lt;/a&gt;&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>ai</category>
      <category>copilotstudio</category>
      <category>powerplatform</category>
      <category>azure</category>
    </item>
    <item>
      <title>From a Messy Ticket to Clean Data</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Sun, 08 Nov 2026 15:56:35 +0000</pubDate>
      <link>https://dev.to/consolatacleah/from-a-messy-ticket-to-clean-data-no4</link>
      <guid>https://dev.to/consolatacleah/from-a-messy-ticket-to-clean-data-no4</guid>
      <description>&lt;p&gt;Last time, we got Claude running inside Microsoft Foundry and said hello to it. Fun, but useless. A model that can describe Seattle is not going to help Harbor Supply Co. ship a single pallet.&lt;/p&gt;

&lt;p&gt;So today Claude gets its first real job. We're going to take one of Harbor's support tickets, the kind that arrives as a half-typed mess at 4:58 on a Friday, and turn it into clean, structured data your systems can actually act on. No chat window. No human reading each one. Just messy text in, tidy fields out.&lt;/p&gt;

&lt;p&gt;This is the moment the whole thing stops being a clever toy and starts being plumbing. And plumbing, if you've built Power Platform flows, is your home turf.&lt;/p&gt;

&lt;h2&gt;
  
  
  Meet a real Harbor ticket
&lt;/h2&gt;

&lt;p&gt;Here's an actual message from Harbor's support inbox. Well, actual for a fictional company. You've seen a hundred like it:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;hi so we ordered the heavy duty pallet racking (order HS-44821 i think?) and 2 of the uprights came bent, like properly bent not just scuffed. warehouse cant use them and were meant to be doing a big stock move THIS WEEKEND. need these replaced asap or at least know whats happening. called twice already. - Dawn, Westlake Logistics&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;A human reads that in three seconds and knows exactly what to do. The trouble is, a human has to read every single one.&lt;/p&gt;

&lt;p&gt;Your flow can't. Power Automate has no idea whether this is urgent, which order it's about, or who should pick it up. A dashboard can't count it. A routing rule can't route it. As far as any of your systems are concerned, this is just a blob of text. Useful to a person, invisible to a process.&lt;/p&gt;

&lt;p&gt;The job is to turn that blob into something a machine can read.&lt;/p&gt;

&lt;h2&gt;
  
  
  Decide what "clean" looks like
&lt;/h2&gt;

&lt;p&gt;Before we ask Claude for anything, we decide the shape we want back. This is the most Power-Platform part of the whole exercise: you're designing the columns of a table, same as you always have.&lt;/p&gt;

&lt;p&gt;For Harbor's tickets, I want six things out of every message:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;category&lt;/strong&gt;: what kind of issue it is (damaged goods, delivery, billing, and so on)&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;urgency&lt;/strong&gt;: low, medium, or high&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;customer&lt;/strong&gt;: who sent it&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;order_ref&lt;/strong&gt;: the order number, if there is one&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;summary&lt;/strong&gt;: one plain line a human can scan&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;needs_human&lt;/strong&gt;: true when this is too thorny for an automated reply&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;That's it. Give me those six fields, filled in correctly, for any ticket Dawn or anyone else throws at us, and the rest of Harbor's automation suddenly has something to work with.&lt;/p&gt;

&lt;h2&gt;
  
  
  The obvious way, and why it bites
&lt;/h2&gt;

&lt;p&gt;The first thing everyone tries is to just ask. You paste the ticket, add "reply with JSON containing category, urgency, customer, order_ref, summary, needs_human," and hit send. And honestly? Most of the time it works. Claude is good at this, and you'll get back something that looks perfect.&lt;/p&gt;

&lt;p&gt;Most of the time is the problem.&lt;/p&gt;

&lt;p&gt;When you're processing one ticket, "most of the time" is fine. When you're processing every ticket, automatically, with no human watching, the few that go sideways are exactly the ones that break your flow at 2am. The failures are small and annoying:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;A friendly sentence before the JSON: "Sure! Here's the structured data:" Your &lt;code&gt;json.loads()&lt;/code&gt; chokes on the first word.&lt;/li&gt;
&lt;li&gt;The JSON wrapped in a markdown code fence, so now you're stripping backticks with string hacks.&lt;/li&gt;
&lt;li&gt;A seventh field Claude thought would be helpful, that your code wasn't expecting.&lt;/li&gt;
&lt;li&gt;"urgency": "pretty urgent" instead of one of your three allowed values.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;None of these are hard to patch. But you end up writing a pile of defensive string-wrangling to clean up output that should have been clean in the first place. There's a better way, and it's built in.&lt;/p&gt;

&lt;h2&gt;
  
  
  The reliable way: give Claude a schema
&lt;/h2&gt;

&lt;p&gt;Claude has a feature called structured outputs. Instead of asking nicely for JSON and hoping, you hand it a schema, and the model is constrained to produce valid JSON that matches it. Not "usually valid." Valid. No stray sentences, no code fences, no surprise fields.&lt;/p&gt;

&lt;p&gt;The cleanest way to do this in Python is to describe the shape as a small class, then let the SDK handle the schema for you. Here it is, built straight on the &lt;code&gt;AnthropicFoundry&lt;/code&gt; client from last time:&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="kn"&gt;from&lt;/span&gt; &lt;span class="n"&gt;anthropic&lt;/span&gt; &lt;span class="kn"&gt;import&lt;/span&gt; &lt;span class="n"&gt;AnthropicFoundry&lt;/span&gt;
&lt;span class="kn"&gt;from&lt;/span&gt; &lt;span class="n"&gt;azure.identity&lt;/span&gt; &lt;span class="kn"&gt;import&lt;/span&gt; &lt;span class="n"&gt;DefaultAzureCredential&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;get_bearer_token_provider&lt;/span&gt;
&lt;span class="kn"&gt;from&lt;/span&gt; &lt;span class="n"&gt;pydantic&lt;/span&gt; &lt;span class="kn"&gt;import&lt;/span&gt; &lt;span class="n"&gt;BaseModel&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;Field&lt;/span&gt;

&lt;span class="n"&gt;base_url&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;https://&amp;lt;your-resource-name&amp;gt;.services.ai.azure.com/anthropic&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;
&lt;span class="n"&gt;deployment_name&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;claude-sonnet-4-6&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;

&lt;span class="n"&gt;token_provider&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;get_bearer_token_provider&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="nc"&gt;DefaultAzureCredential&lt;/span&gt;&lt;span class="p"&gt;(),&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;https://ai.azure.com/.default&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;
&lt;span class="p"&gt;)&lt;/span&gt;
&lt;span class="n"&gt;client&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nc"&gt;AnthropicFoundry&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;azure_ad_token_provider&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;token_provider&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;base_url&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;base_url&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;

&lt;span class="c1"&gt;# This is our table definition. Claude is forced to fill exactly these fields.
&lt;/span&gt;&lt;span class="k"&gt;class&lt;/span&gt; &lt;span class="nc"&gt;Ticket&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;BaseModel&lt;/span&gt;&lt;span class="p"&gt;):&lt;/span&gt;
    &lt;span class="n"&gt;category&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nb"&gt;str&lt;/span&gt;
    &lt;span class="n"&gt;urgency&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nb"&gt;str&lt;/span&gt;
    &lt;span class="n"&gt;customer&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nb"&gt;str&lt;/span&gt;
    &lt;span class="n"&gt;order_ref&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nb"&gt;str&lt;/span&gt; &lt;span class="o"&gt;|&lt;/span&gt; &lt;span class="bp"&gt;None&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nc"&gt;Field&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;description&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;Order number if one is mentioned, else null&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;summary&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nb"&gt;str&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nc"&gt;Field&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;description&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;One short line a human can scan&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
    &lt;span class="n"&gt;needs_human&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="nb"&gt;bool&lt;/span&gt;

&lt;span class="n"&gt;ticket_text&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="sh"&gt;"""&lt;/span&gt;&lt;span class="s"&gt;hi so we ordered the heavy duty pallet racking (order HS-44821 i think?)
and 2 of the uprights came bent ... need these replaced asap ... - Dawn, Westlake Logistics&lt;/span&gt;&lt;span class="sh"&gt;"""&lt;/span&gt;

&lt;span class="n"&gt;result&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;client&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;messages&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;parse&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="n"&gt;model&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;deployment_name&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;max_tokens&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;512&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;messages&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="p"&gt;[{&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;role&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;user&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;content&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="sa"&gt;f&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;Extract the ticket fields:&lt;/span&gt;&lt;span class="se"&gt;\n\n&lt;/span&gt;&lt;span class="si"&gt;{&lt;/span&gt;&lt;span class="n"&gt;ticket_text&lt;/span&gt;&lt;span class="si"&gt;}&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="p"&gt;}],&lt;/span&gt;
    &lt;span class="n"&gt;output_format&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;Ticket&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
&lt;span class="p"&gt;)&lt;/span&gt;

&lt;span class="nf"&gt;print&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;result&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;parsed_output&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;And what comes back is not a hopeful blob of text. It's a real &lt;code&gt;Ticket&lt;/code&gt; object, already parsed and validated:&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="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"category"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"damaged goods"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"urgency"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"high"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"customer"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"Dawn, Westlake Logistics"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"order_ref"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"HS-44821"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"summary"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="s2"&gt;"Two pallet racking uprights arrived bent; replacement needed before a weekend stock move."&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;&lt;span class="w"&gt;
  &lt;/span&gt;&lt;span class="nl"&gt;"needs_human"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt;&lt;span class="w"&gt; &lt;/span&gt;&lt;span class="kc"&gt;true&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;span class="p"&gt;}&lt;/span&gt;&lt;span class="w"&gt;
&lt;/span&gt;&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;That is a row. You could drop it straight into Dataverse, raise it in a queue, or hand it to a Power Automate flow without touching a single string function. The messy blob became clean data, and it will be clean every single time.&lt;/p&gt;

&lt;h2&gt;
  
  
  Tightening it up
&lt;/h2&gt;

&lt;p&gt;The first version works, but a couple of fields are looser than I'd like. Building in the open means showing the two small changes that made it genuinely production-ready.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Pin down urgency.&lt;/strong&gt; Right now &lt;code&gt;urgency&lt;/code&gt; is a plain string, so one ticket says "high" and the next says "urgent" or "High-ish." I don't want to find out at 2am that my routing rule missed one. So I turn it into a fixed set of choices, and now Claude can only ever return one of three values:&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="kn"&gt;from&lt;/span&gt; &lt;span class="n"&gt;enum&lt;/span&gt; &lt;span class="kn"&gt;import&lt;/span&gt; &lt;span class="n"&gt;Enum&lt;/span&gt;

&lt;span class="k"&gt;class&lt;/span&gt; &lt;span class="nc"&gt;Urgency&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="nb"&gt;str&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;Enum&lt;/span&gt;&lt;span class="p"&gt;):&lt;/span&gt;
    &lt;span class="n"&gt;low&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;low&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;
    &lt;span class="n"&gt;medium&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;medium&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;
    &lt;span class="n"&gt;high&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;high&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;

&lt;span class="c1"&gt;# in the model:
&lt;/span&gt;    &lt;span class="n"&gt;urgency&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="n"&gt;Urgency&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;strong&gt;Make the model admit when it's unsure.&lt;/strong&gt; A ticket that just says "where is my stuff" with no order number shouldn't be guessed at. I add a short instruction in the prompt: &lt;em&gt;set needs_human to true whenever a field is missing or the request is ambiguous.&lt;/em&gt; That one line turns needs_human from decoration into a real safety valve, so the awkward cases get a person instead of a confident wrong answer.&lt;/p&gt;

&lt;p&gt;That's the whole loop. Decide the shape, run a few real tickets through it, and tighten the fields that wobble. You're not prompt-whispering; you're doing the same iterative tidying you'd do to any flow that has to survive real-world input.&lt;/p&gt;

&lt;h2&gt;
  
  
  A few things that will trip you
&lt;/h2&gt;

&lt;p&gt;Short list, learned so you don't have to:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Check why it stopped.&lt;/strong&gt; Look at &lt;code&gt;stop_reason&lt;/code&gt; on the response. If it's &lt;code&gt;max_tokens&lt;/code&gt;, your output was cut off mid-JSON, so raise the limit. If it's &lt;code&gt;refusal&lt;/code&gt;, the model declined and the output may not match your schema, so don't blindly parse it.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Compare enums case-insensitively.&lt;/strong&gt; A value can come back differing only in capitalisation ("High" vs "high"), so don't do a strict string match in your routing.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Objects need &lt;code&gt;additionalProperties: false&lt;/code&gt;.&lt;/strong&gt; The Pydantic helper handles this for you, but if you hand-write a raw JSON schema, set it, or the request is rejected.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Keep real customer data out of the schema.&lt;/strong&gt; Field names and descriptions are fine, but never bake an actual person's details into the schema definition itself. Harbor is fictional on purpose, and your real tickets should stay in the data, not the code.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Why this is the real unlock
&lt;/h2&gt;

&lt;p&gt;Step back and look at what just happened. We took something only a human could use, a rambling message, and turned it into something a machine can route, count, and act on. That is the entire bridge from "chatbot" to "automation," and it runs right through the skills you already have. A Power Platform person who can design a clean table and reason about edge cases is exactly the person who's good at this.&lt;/p&gt;

&lt;p&gt;Everything from here builds on this one move. An agent that answers tickets, escalates the hard ones, or looks up an order is really just this extraction with actions bolted on.&lt;/p&gt;

&lt;p&gt;Next time we bolt on the first action. We'll take this tidy ticket data into Copilot Studio and stand up an actual agent that can answer a Harbor customer, grounded in Harbor's own world rather than whatever the model happens to remember. That's the post where the two halves of your brain, the business side and the AI side, finally meet.&lt;/p&gt;

&lt;p&gt;If you run a few of your own messy tickets through this, I'd love to see what came back. Show me your worst one.&lt;/p&gt;

&lt;p&gt;See you next time. 😊&lt;/p&gt;

</description>
      <category>ai</category>
      <category>automation</category>
      <category>claude</category>
      <category>llm</category>
    </item>
    <item>
      <title>Deploying Your First Claude Model</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Mon, 26 Oct 2026 20:15:59 +0000</pubDate>
      <link>https://dev.to/consolatacleah/deploying-your-first-claude-model-126o</link>
      <guid>https://dev.to/consolatacleah/deploying-your-first-claude-model-126o</guid>
      <description>&lt;p&gt;Last time I made a promise: that if you can build a Power Automate flow, you're most of the way to building an AI agent. This time we start proving it, and we do it the only way that really counts: by actually building something.&lt;/p&gt;

&lt;p&gt;This is a hands-on walkthrough, not concepts. We're going to stand up a real Claude model inside Microsoft Foundry, say hello to it, and get a reply back. That's it. One small, complete, working thing. Everything else in this series stacks on top of this one step, so it's worth doing slowly and getting right.&lt;/p&gt;

&lt;h2&gt;
  
  
  Before you start
&lt;/h2&gt;

&lt;p&gt;Nothing to install on your machine yet. But there are a few boxes to tick on the Azure side, and it's less painful to know them up front than to hit a wall halfway through.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;A &lt;strong&gt;paid Azure subscription&lt;/strong&gt; with pay-as-you-go billing. This is the one that trips people up: a free trial, student credits, or a sponsored "Azure credits only" subscription won't work for Claude. You need real pay-as-you-go billing turned on.&lt;/li&gt;
&lt;li&gt;A &lt;strong&gt;Microsoft Foundry project&lt;/strong&gt; in a supported region. If you don't have one, you'll create it the first time you open the portal.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Azure Marketplace access&lt;/strong&gt;, because Claude is a partner model you subscribe to through the Marketplace.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Contributor or Owner&lt;/strong&gt; on the resource group. If someone else runs your Azure tenant, this is the thing to go ask them for.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you've got all four, you're good. If you're missing the billing one, sort that out first, because everything downstream depends on it.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;"Do I have to pay to try this?"&lt;/strong&gt; Barely. There's no fully free way to run Claude in Foundry. The free Azure account gives you $200 of credit, but that's credit-only, and Claude won't run on credit-only or trial subscriptions. What you actually need is a pay-as-you-go subscription: a card on file, but no upfront charge. You only pay for the tokens you use, and for today's hello-world that's a few cents. So it's less "buy a subscription" and more "put a card down and pay pennies for the experiment."&lt;/p&gt;

&lt;h2&gt;
  
  
  Deploying Claude from the catalog
&lt;/h2&gt;

&lt;p&gt;This is the part that sounds intimidating and turns out to be about eight clicks. Here's the whole path:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Sign in to Microsoft Foundry at &lt;strong&gt;ai.azure.com&lt;/strong&gt;, with the &lt;strong&gt;New Foundry&lt;/strong&gt; toggle on.&lt;/li&gt;
&lt;li&gt;Go to &lt;strong&gt;Discover &amp;gt; Models&lt;/strong&gt; and pick a Claude model. I went with &lt;strong&gt;Claude Sonnet 4.6&lt;/strong&gt;, the one Microsoft uses in all their examples and a sensible, capable default to learn on.&lt;/li&gt;
&lt;li&gt;Select &lt;strong&gt;Deploy &amp;gt; Custom settings&lt;/strong&gt;. (There's a Default settings option too, which just picks the Azure-hosted version for you. We'll use Custom so you can see what the choices actually are.)&lt;/li&gt;
&lt;li&gt;Accept the Azure Marketplace terms, pick an industry, and hit &lt;strong&gt;Agree and Proceed&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Choose the &lt;strong&gt;model version&lt;/strong&gt; and &lt;strong&gt;region scope&lt;/strong&gt;. More on what those mean in the next section.&lt;/li&gt;
&lt;li&gt;Rename the deployment if you like. Worth knowing: &lt;strong&gt;the deployment name is what you'll pass as the &lt;code&gt;model&lt;/code&gt; in your code later&lt;/strong&gt;, so give it something you'll recognise.&lt;/li&gt;
&lt;li&gt;Hit &lt;strong&gt;Deploy&lt;/strong&gt;. When it finishes you land straight in the Playground, where you can type at Claude and get a reply without writing a line of code.&lt;/li&gt;
&lt;li&gt;Pop over to the &lt;strong&gt;Details&lt;/strong&gt; tab and check the status says &lt;strong&gt;Succeeded&lt;/strong&gt;.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;That's genuinely it. You now have a frontier model running under your own Azure subscription.&lt;/p&gt;

&lt;h2&gt;
  
  
  Two flavors of "hosted", and which to pick
&lt;/h2&gt;

&lt;p&gt;At the version step you'll see two options, and it's worth a sentence each so you're choosing on purpose rather than clicking past them.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Version 2, Hosted on Azure.&lt;/strong&gt; This is the default, and for most people it's the right answer. The model runs on Azure infrastructure, so it sits neatly inside your existing Azure identity, networking, and governance. If you just want to get going, pick this.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Version 1, Hosted on Anthropic infrastructure.&lt;/strong&gt; This routes to Anthropic's own hosting. You'd reach for it if you need an Anthropic feature or a model that isn't on the Azure-hosted option yet.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;There's also a &lt;strong&gt;region scope&lt;/strong&gt; choice. &lt;strong&gt;Global Standard&lt;/strong&gt; works for every Claude model and version, and is the safe default. &lt;strong&gt;Data Zone Standard (US)&lt;/strong&gt; keeps things in a US data zone and is available for some of the Azure-hosted versions. Handy if you have data-residency requirements, which, spoiler, is exactly the kind of thing Harbor's eventual security review will care about.&lt;/p&gt;

&lt;p&gt;For today: Hosted on Azure, Global Standard. We'll come back to the residency question in a later post when the stakes are higher.&lt;/p&gt;

&lt;h2&gt;
  
  
  Saying hello to Claude
&lt;/h2&gt;

&lt;p&gt;The Playground is lovely, but the whole point of this series is to call Claude from code so we can wire it into real things later. So let's make one request.&lt;/p&gt;

&lt;p&gt;First, two things to grab from your deployment's &lt;strong&gt;Details&lt;/strong&gt; tab:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Your &lt;strong&gt;endpoint&lt;/strong&gt;, which looks like &lt;code&gt;https://&amp;lt;your-resource-name&amp;gt;.services.ai.azure.com/anthropic&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;Your &lt;strong&gt;deployment name&lt;/strong&gt; (the one you set earlier). This is the &lt;code&gt;model&lt;/code&gt; value.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Then pick how you authenticate. There are two ways:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Microsoft Entra ID&lt;/strong&gt;, the keyless, enterprise-friendly option. Nothing secret sits in your code; Azure hands out a short-lived token. This is the one your security team will prefer, so it's the one I'll show.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;API key&lt;/strong&gt;, simpler to start with, copied straight from the deployment details. Fine for a quick experiment. (One gotcha: the newest Mythos models are Entra-only and don't take API keys.)&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Install the SDK with &lt;code&gt;pip install -U anthropic azure-identity&lt;/code&gt;, then here's the whole hello-world with Entra ID:&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="kn"&gt;from&lt;/span&gt; &lt;span class="n"&gt;anthropic&lt;/span&gt; &lt;span class="kn"&gt;import&lt;/span&gt; &lt;span class="n"&gt;AnthropicFoundry&lt;/span&gt;
&lt;span class="kn"&gt;from&lt;/span&gt; &lt;span class="n"&gt;azure.identity&lt;/span&gt; &lt;span class="kn"&gt;import&lt;/span&gt; &lt;span class="n"&gt;DefaultAzureCredential&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="n"&gt;get_bearer_token_provider&lt;/span&gt;

&lt;span class="n"&gt;base_url&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;https://&amp;lt;your-resource-name&amp;gt;.services.ai.azure.com/anthropic&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;
&lt;span class="n"&gt;deployment_name&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;claude-sonnet-4-6&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;  &lt;span class="c1"&gt;# whatever you named your deployment
&lt;/span&gt;
&lt;span class="n"&gt;token_provider&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nf"&gt;get_bearer_token_provider&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="nc"&gt;DefaultAzureCredential&lt;/span&gt;&lt;span class="p"&gt;(),&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;https://ai.azure.com/.default&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;
&lt;span class="p"&gt;)&lt;/span&gt;

&lt;span class="n"&gt;client&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="nc"&gt;AnthropicFoundry&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="n"&gt;azure_ad_token_provider&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;token_provider&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;base_url&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;base_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;message&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="n"&gt;client&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;messages&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;create&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;
    &lt;span class="n"&gt;model&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="n"&gt;deployment_name&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;max_tokens&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="mi"&gt;512&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt;
    &lt;span class="n"&gt;messages&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="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;role&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;user&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="p"&gt;,&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;content&lt;/span&gt;&lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="p"&gt;:&lt;/span&gt; &lt;span class="sh"&gt;"&lt;/span&gt;&lt;span class="s"&gt;In one sentence, what does Harbor Supply Co. sell?&lt;/span&gt;&lt;span class="sh"&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;print&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;message&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="n"&gt;content&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. If a sentence about pallets and forklifts comes back, congratulations: you just called a frontier model running under your own Azure subscription. If you'd rather use the API key, swap the client line for &lt;code&gt;AnthropicFoundry(api_key="YOUR_KEY", base_url=base_url)&lt;/code&gt; and drop the token bits.&lt;/p&gt;

&lt;p&gt;Tiny as it is, this is the hinge the whole series turns on. Every agent we build from here is just this call, with more interesting things wrapped around it.&lt;/p&gt;

&lt;h2&gt;
  
  
  The billing bit nobody reads until it's too late
&lt;/h2&gt;

&lt;p&gt;Quick, boring, important. Skipping this is how you get a surprise later.&lt;/p&gt;

&lt;p&gt;Claude usage shows up on your &lt;strong&gt;normal Azure invoice&lt;/strong&gt;. No separate Anthropic bill, no separate vendor to onboard. For a business that already lives in Azure, that's the quiet win here: no new procurement, no new contract, it just lands on the invoice you already pay. Usage is metered in Claude Consumption Units, and if your organisation has a &lt;strong&gt;Microsoft Azure Consumption Commitment (MACC)&lt;/strong&gt;, eligible Claude spend can count toward it. That's the kind of sentence that makes finance teams suddenly interested.&lt;/p&gt;

&lt;p&gt;The flip side is the eligibility rules, which are stricter than you'd expect. These subscription types &lt;strong&gt;can't&lt;/strong&gt; use Claude: free trials and student or startup-credit subscriptions, sponsored "Azure credits only" subscriptions, Cloud Solution Provider subscriptions, and enterprise accounts in a couple of specific regions. The common thread is the same as the prerequisites: real pay-as-you-go billing, or it won't run.&lt;/p&gt;

&lt;p&gt;None of this costs anything until you actually make calls, so experimenting is cheap. Just know where the meter is before you build something that hammers it.&lt;/p&gt;

&lt;h2&gt;
  
  
  When it breaks (because it will)
&lt;/h2&gt;

&lt;p&gt;I promised at the start of this series you'd see the bits that don't work on the first try. The classic first-run error here is a 403. Here's the short list of what goes wrong and the fix, so you don't lose an afternoon to a stack trace:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;What you see&lt;/th&gt;
&lt;th&gt;What it usually means&lt;/th&gt;
&lt;th&gt;The fix&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;401&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Auth isn't right&lt;/td&gt;
&lt;td&gt;Check the API key, or confirm your Entra scope is exactly &lt;code&gt;https://ai.azure.com/.default&lt;/code&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;403&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;You're in, but not allowed&lt;/td&gt;
&lt;td&gt;Check your role; for Entra ID, assign yourself &lt;strong&gt;Cognitive Services User&lt;/strong&gt;
&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;404&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;It can't find the model&lt;/td&gt;
&lt;td&gt;Re-check the base URL format and that &lt;code&gt;model&lt;/code&gt; matches your &lt;strong&gt;deployment name&lt;/strong&gt;, not the model ID&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;429&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;You're going too fast&lt;/td&gt;
&lt;td&gt;Back off and retry, or request a quota increase&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The 403 trips people up because Contributor on the resource group looks like it should be enough. For the actual calls, Entra ID also wants &lt;strong&gt;Cognitive Services User&lt;/strong&gt;. Thirty seconds to fix once you know; twenty minutes of confusion if you don't.&lt;/p&gt;

&lt;h2&gt;
  
  
  Next up
&lt;/h2&gt;

&lt;p&gt;Right now Claude will happily tell us about Seattle or Harbor's forklifts, which is fun and completely useless for actual work. Next time we put it to a real job: taking one of Harbor's messy, half-typed support tickets and turning it into clean, structured data your systems can act on. That's where this stops being a chatbot and starts being useful.&lt;/p&gt;

&lt;p&gt;If you got your first reply back today, you're officially off the starting line. Tell me what came back from your hello-world. I'd love to hear it.&lt;/p&gt;

&lt;p&gt;See you next time. 😊&lt;/p&gt;

</description>
      <category>ai</category>
      <category>azure</category>
      <category>powerplatform</category>
    </item>
    <item>
      <title>You're Already an Agent Builder</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Sun, 18 Oct 2026 18:02:48 +0000</pubDate>
      <link>https://dev.to/consolatacleah/youre-already-an-agent-builder-5943</link>
      <guid>https://dev.to/consolatacleah/youre-already-an-agent-builder-5943</guid>
      <description>&lt;p&gt;If you’ve spent the last few years building Power Automate flows, making Power Apps that people slowly came to rely on, or fixing a chatbot that keeps breaking, here’s some good news. You’re already most of the way to building AI agents. You just haven’t been calling them that.&lt;/p&gt;

&lt;p&gt;I've basically kept a Power Platform diary on this blog for years now. And lately one word kept following me around: &lt;em&gt;agents&lt;/em&gt;. Foundry this, Copilot Studio that, Claude the other. For a hot minute it felt like a whole new universe, like I'd have to forget everything I knew and reintroduce myself as An AI Person.&lt;/p&gt;

&lt;p&gt;Reader, I did not. And neither will you.&lt;/p&gt;

&lt;h2&gt;
  
  
  From flows to agents: what actually changes
&lt;/h2&gt;

&lt;p&gt;Here's the reframe that made it click for me.&lt;/p&gt;

&lt;p&gt;A Power Automate flow is a recipe you write out by hand: &lt;em&gt;when this happens, do that, then that, and if it's a Tuesday, do the other thing.&lt;/em&gt; It's wonderful, and it's completely deterministic. It does exactly what you told it and not one step more.&lt;/p&gt;

&lt;p&gt;An agent is what you get when you stop spelling out every branch and hand some of the deciding to a model. You give it a goal, a few tools, and some knowledge, and it works out the steps itself.&lt;/p&gt;

&lt;p&gt;Now look at what &lt;em&gt;doesn't&lt;/em&gt; change. You're still thinking in triggers, connectors, data, and "what should actually happen when a customer asks for a refund." That business-process brain, knowing what good looks like, is the genuinely hard part, and you already have it. What's new is the plumbing underneath.&lt;/p&gt;

&lt;p&gt;The happy surprise: that plumbing now lives right inside the stack you already know. Claude is generally available in Microsoft Foundry on Azure, so you build these agents with your existing Azure sign-in, billing, and governance. No new universe. Same Azure, new toy.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fkedgh5o41zy5jztqztat.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fkedgh5o41zy5jztqztat.png" alt="Two-panel diagram. On the left, a flow: … ends in triaged, answered, or escalated." width="800" height="500"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Meet Harbor Supply Co.
&lt;/h2&gt;

&lt;p&gt;Every decent tutorial needs a guinea pig, so say hello to &lt;strong&gt;Harbor Supply Co.&lt;/strong&gt; — our invented distributor for this series. They sell deeply unglamorous things (boxes, pallets, the occasional forklift), they've got a product catalog, a stack of orders, and a support inbox that is perpetually a little on fire.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F77s8t9zf9pxl6fulm4zf.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F77s8t9zf9pxl6fulm4zf.png" alt="Harbor Supply Co. logo — an anchor badge for the fictional distributor used throughout this series." width="800" height="400"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;We'll use Harbor for everything, cover to cover. Same company, same slightly chaotic data. That way each post stands on the one before it instead of starting from a blank slate, and you can follow the whole thing like a story rather than a few disconnected demos.&lt;/p&gt;

&lt;h2&gt;
  
  
  What we're building
&lt;/h2&gt;

&lt;p&gt;Few posts, from now through the end of December. We start gentle: stand up a Foundry project, find Claude in the catalog, make our very first call to it. Then we climb.&lt;/p&gt;

&lt;p&gt;By the end, Harbor will have a Copilot Studio agent that can actually &lt;em&gt;do&lt;/em&gt; things: look up an order, answer from the real returns policy instead of cheerfully inventing one, and know when a question is over its head.&lt;/p&gt;

&lt;p&gt;Along the way we'll get into the parts nobody warns you about. How to ground an agent so it stops guessing. How to let it take actions without handing it the keys to everything. When an agent is honestly overkill and a plain Power Automate flow was the right answer all along. And the questions your security team &lt;em&gt;will&lt;/em&gt; ask the moment this gets anywhere near production.&lt;/p&gt;

&lt;p&gt;I'll be building it in real time, which means you'll see the bits that don't work on the first try too. That's the fun part.&lt;/p&gt;

&lt;h2&gt;
  
  
  Where we start
&lt;/h2&gt;

&lt;p&gt;First stop: setting up Microsoft Foundry and tracking down where Claude lives in the model catalog. Nothing to install today, just an Azure account and a little curiosity when you're ready to follow along.&lt;/p&gt;

&lt;p&gt;If you've been quietly Power-Platform-curious about this whole agent thing but weren't sure where someone who isn't a card-carrying "AI Person" fits in, this series is aimed squarely at you. Follow along, build Harbor with me, and let's find out together.&lt;/p&gt;

&lt;p&gt;Let's build. 😊&lt;/p&gt;

</description>
      <category>powerplatform</category>
      <category>copilotstudio</category>
      <category>azure</category>
      <category>ai</category>
    </item>
    <item>
      <title>Top 5 Use Cases for Automation in Humanitarian Aid Organizations</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Thu, 18 Sep 2025 06:37:37 +0000</pubDate>
      <link>https://dev.to/consolatacleah/top-5-use-cases-for-automation-in-humanitarian-aid-organizations-ip1</link>
      <guid>https://dev.to/consolatacleah/top-5-use-cases-for-automation-in-humanitarian-aid-organizations-ip1</guid>
      <description>&lt;p&gt;In this era of Artificial Intelligence, the humanitarian sector can no longer afford to sit on the sidelines. Automation, once seen as a luxury, is now a lifeline. It’s the quiet miracle helping aid organizations streamline operations, reduce errors, and free up precious time for what truly matters,serving communities in crisis.&lt;/p&gt;

&lt;p&gt;Let’s take a closer look at how automation is already making a difference across five critical areas.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Donor Reporting and Compliance&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;For many humanitarian organizations, the end of the year brings a whirlwind of donor reports, compliance reviews, and impact documentation. In 2025, this workload has intensified, especially following funding reductions from USAID. NGOs have found themselves submitting more frequent reports, crafting detailed justifications for grant extensions, and preparing new proposals to diversify their funding streams all while operating with leaner teams due to budget constraints.&lt;/p&gt;

&lt;p&gt;These reporting tasks are vital, but they’re also time-consuming and vulnerable to human error. That’s where automation is stepping in to lighten the load. By pulling data from financial systems, field reports, and CRM platforms, automation tools are generating clean, donor-ready templates with minimal manual input. &lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Procurement and Vendor Processing&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Emergency response often demands rapid procurement, but speed must be balanced with accountability. Automation is helping NGOs onboard vendors faster by verifying documents, checking compliance, and flagging risks all in real time. &lt;/p&gt;

&lt;p&gt;Mercy Corps, for example, has integrated automated procurement tools to streamline vendor registration and contract management, reducing turnaround times and improving transparency across their supply chain.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;IT Support&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Field teams working in remote or low-connectivity areas frequently encounter technical issues that can stall operations. AI-powered chatbots are now providing real-time IT support, guiding users through troubleshooting steps, resetting credentials, and escalating tickets when needed. &lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Field Data Monitoring&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Data from the field is the heartbeat of humanitarian work. Whether it’s attendance logs, distribution records, or health screenings, timely insights are critical. Automation tools are helping organizations scan incoming data for anomalies such as missing supplies, low turnout, or delayed deliveries and trigger alerts instantly. &lt;br&gt;
The Danish Refugee Council has piloted automated monitoring systems that flag operational risks early, enabling faster interventions and better resource allocation.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Inventory Tracking&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;In humanitarian logistics, visibility isn’t just helpful, it’s life-saving. Knowing what’s in stock, where it’s located, and when it’s moving can mean the difference between timely aid and missed opportunities. That’s why automation is becoming a vital ally in supply chain operations, helping organizations manage stock levels, track shipments and reconcile inventory across multiple warehouses with precision and speed.&lt;/p&gt;

&lt;p&gt;UNICEF’s Supply Division has been leading the charge with its&lt;a href="https://www.unicef.org/supply/vaccine-stock-management-tool" rel="noopener noreferrer"&gt; Stock Management Tool (SMT)&lt;/a&gt;a ready to use digital platform designed to bring end-to-end visibility to vaccine and therapeutic supply chains. Developed in partnership with WHO, the SMT is tailored for use in low-resource settings, offering both online and offline functionality to ensure accessibility even in areas with unreliable internet.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;In conclusion&lt;/strong&gt;, automation isn’t just a technical upgrade, it’s a gift of time. These digital helpers are quietly working behind the scenes, enabling humanitarian teams to deliver with precision and purpose, even in the face of growing complexity and limited resources. And that, perhaps, is the most meaningful transformation of all.&lt;/p&gt;

</description>
      <category>automation</category>
    </item>
    <item>
      <title>UiPath vs Power Platform: Same Goals, Different Soap Opera</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Sun, 14 Sep 2025 19:07:00 +0000</pubDate>
      <link>https://dev.to/consolatacleah/uipath-vs-power-platform-same-goals-different-soap-opera-3207</link>
      <guid>https://dev.to/consolatacleah/uipath-vs-power-platform-same-goals-different-soap-opera-3207</guid>
      <description>&lt;h2&gt;
  
  
  1. Meet the Cast
&lt;/h2&gt;

&lt;p&gt;Before we dive into the drama, let’s introduce our stars:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;&lt;em&gt;UiPath Character&lt;/em&gt;&lt;/th&gt;
&lt;th&gt;&lt;em&gt;Power Platform Counterpart&lt;/em&gt;&lt;/th&gt;
&lt;th&gt;&lt;em&gt;Personality&lt;/em&gt;&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;em&gt;Studio&lt;/em&gt;&lt;/td&gt;
&lt;td&gt;&lt;em&gt;Power Apps&lt;/em&gt;&lt;/td&gt;
&lt;td&gt;The  developer, enables creation of graphic interfaces that allows users to connect with data or agents in the case of studio.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;em&gt;Orchestrator&lt;/em&gt;&lt;/td&gt;
&lt;td&gt;&lt;em&gt;Power Automate&lt;/em&gt;&lt;/td&gt;
&lt;td&gt;The bossy scheduler, enables creation of automation workflows with orchestrator being able to handle complex processes&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;em&gt;Maestro&lt;/em&gt;&lt;/td&gt;
&lt;td&gt;&lt;em&gt;Power Platform Pipelines&lt;/em&gt;&lt;/td&gt;
&lt;td&gt;The process diva,lives for long-running workflows and dramatic pauses.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;em&gt;Agents&lt;/em&gt;&lt;/td&gt;
&lt;td&gt;&lt;em&gt;AI Builder / Custom Connectors&lt;/em&gt;&lt;/td&gt;
&lt;td&gt;The mysterious sidekick shows up when things get complex.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;em&gt;UiPath Chatbots&lt;/em&gt;&lt;/td&gt;
&lt;td&gt;&lt;em&gt;Power Virtual Agents&lt;/em&gt;&lt;/td&gt;
&lt;td&gt;The smooth talker, great with users, secretly powered by backend drama.&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;Each platform has its own ensemble cast, but they’re all chasing the same plot: automate processes, empower users, and reduce manual chaos.&lt;/p&gt;

&lt;h1&gt;
  
  
  2. The Relationship Status
&lt;/h1&gt;

&lt;p&gt;UiPath’s components are tightly coupled &lt;em&gt;Studio&lt;/em&gt; builds the bot, &lt;em&gt;Orchestrator&lt;/em&gt; deploys it, &lt;em&gt;Maestro&lt;/em&gt; choreographs it, and &lt;em&gt;Agents&lt;/em&gt; fill in the gaps. It’s like a well-rehearsed theatre troupe: everyone knows their role, cues are timed, and the performance runs smoothly.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Power Platform&lt;/em&gt;, on the other hand, is more like a group project in college: &lt;em&gt;Power Apps&lt;/em&gt; builds the front-end, &lt;em&gt;Power Automate&lt;/em&gt; handles the logic, and &lt;em&gt;AI Builder&lt;/em&gt; occasionally shows up when needed.&lt;/p&gt;

&lt;p&gt;They work together, but sometimes you wonder if they’ve even met.&lt;/p&gt;

&lt;h1&gt;
  
  
  3. Emotional Intelligence or Just Good UX?
&lt;/h1&gt;

&lt;p&gt;UiPath Studio is powerful but intimidating. Power Apps, meanwhile, is the friendly neighbor who helps you build a form in five minutes and then asks if you want to add AI.&lt;/p&gt;

&lt;h1&gt;
  
  
  4. Secrets, Scandals, and Shadow IT
&lt;/h1&gt;

&lt;p&gt;Governance is where the drama hits peak ratings.&lt;br&gt;
UiPath is built for centralized control. Power Platform needs strong governance policies and CoE frameworks to avoid chaos.&lt;/p&gt;

&lt;p&gt;UiPath Orchestrator is strict all roles, permissions, audit logs, and robot management are baked in.&lt;/p&gt;

&lt;p&gt;Power Platform is more democratic until someone builds a Flow that emails 500 people every 10 minutes and IT gets involved.&lt;/p&gt;

&lt;h1&gt;
  
  
  5. Who Gets the Final Rose?
&lt;/h1&gt;

&lt;p&gt;It depends on your storyline:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Scenario&lt;/th&gt;
&lt;th&gt;Winner&lt;/th&gt;
&lt;th&gt;Why&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Enterprise-grade automation with complex workflows&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;UiPath&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Maestro, Orchestrator, and Agents shine here.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Rapid app development for business users&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;Power Platform&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Power Apps + Power Automate = instant MVP.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Humanitarian tech in low-connectivity environments&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;Tie&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;UiPath’s offline bots vs Power Platform’s mobile-first design which can be configured for offline use.&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Citizen developer empowerment&lt;/td&gt;
&lt;td&gt;&lt;strong&gt;Power Platform&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Studio can be abit scary.would need some more experience as compared to powerapps.&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h2&gt;
  
  
  🎬 Final Scene
&lt;/h2&gt;

&lt;p&gt;Like any good soap opera, the automation saga ends not with a clear winner but with a cliffhanger.&lt;/p&gt;

&lt;p&gt;UiPath and Power Platform may chase the same goals, but they bring wildly different personalities to the stage. One thrives on orchestration and precision; the other on accessibility and speed. And in the middle of it all? You the architect, the strategist, the storyteller deciding which cast best fits your organization’s plot.&lt;/p&gt;

&lt;p&gt;Whether you're building for enterprise scale, humanitarian impact, or citizen empowerment, the real power lies in knowing your audience, your constraints, and your endgame. Choose your tools like you cast your leads: not just for talent, but for chemistry.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;So what’s next&lt;/strong&gt;&lt;br&gt;
Start your own pilot episode. Map your workflows, sketch your personas, and test your platforms. And if the drama gets too real, remember even bots need good direction.&lt;/p&gt;

</description>
      <category>automation</category>
      <category>powerplatform</category>
      <category>rpa</category>
    </item>
    <item>
      <title>How I met Data Pipelines😊</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Sun, 05 Mar 2023 18:24:26 +0000</pubDate>
      <link>https://dev.to/consolatacleah/how-i-met-data-pipelines-5661</link>
      <guid>https://dev.to/consolatacleah/how-i-met-data-pipelines-5661</guid>
      <description>&lt;p&gt;Which is your favorite TV show? Mine is  the classic, "How I met your mother". In the entertainment industry, data pipelines can be equated to the process of creating and delivering a film or television show. &lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fyvtkaoidx8wtwp8lnb2g.jpg" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fyvtkaoidx8wtwp8lnb2g.jpg" alt=" " width="799" height="491"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Just as data pipelines extract, transform, and load data from various sources into a central repository for analysis and reporting, the process of creating a film or television show involves extracting ideas and inspiration from various sources, transforming them into a cohesive story, and delivering them to the audience.&lt;/p&gt;

&lt;p&gt;The production process for a film or television show typically involves several stages, including pre-production, production, and post-production. During pre-production, ideas and inspiration are gathered from various sources, such as books, scripts, or real-life events, and transformed into a cohesive story that can be turned into a film or television show. This process is similar to the data extraction and transformation stage of a data pipeline.&lt;/p&gt;

&lt;p&gt;During production, the story is brought to life through the use of actors, sets, and special effects. This stage is similar to the data loading stage of a data pipeline, where the transformed data is loaded into a central repository for analysis and reporting.&lt;/p&gt;

&lt;p&gt;Finally, during post-production, the film or television show is edited, polished, and delivered to the audience. This stage is similar to the reporting stage of a data pipeline, where the data is analyzed, processed, and presented to stakeholders.&lt;/p&gt;

&lt;p&gt;Data pipelines have revolutionized the way businesses and organizations process, manage, and analyze their data. These pipelines are designed to extract, transform, and load data from various sources into a central repository for analysis and reporting. &lt;/p&gt;

&lt;p&gt;In this blog post, we'll also get to explore the advantages of data pipelines and why they are an essential component of any modern data architecture.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Scalability and Flexibility:&lt;/strong&gt; One of the most significant advantages of data pipelines is their scalability and flexibility. With the ability to scale up or down based on the volume of data, data pipelines allow organizations to process large volumes of data efficiently. Additionally, they are flexible enough to handle data from various sources, including structured, semi-structured, and unstructured data, making them ideal for modern data architectures.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Automated Data Processing&lt;/strong&gt;: Data pipelines automate the process of data extraction, transformation, and loading. This automation reduces the risk of human error, increases processing speed, and ensures data accuracy. Additionally, data pipelines allow organizations to process data in near-real-time, enabling timely and accurate decision-making.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Data Quality&lt;/strong&gt;: Data pipelines help maintain data quality by identifying and addressing data inconsistencies, errors, and redundancies. By performing data quality checks during the extraction, transformation, and loading process, organizations can ensure that the data they analyze is accurate, complete, and consistent.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Cost Savings&lt;/strong&gt;: Data pipelines help organizations save money by reducing the need for manual data processing and maintenance. By automating the data processing process, organizations can significantly reduce labor costs, improve data accuracy, and make better use of their resources.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Improved Data Governance:&lt;/strong&gt; Data pipelines facilitate better data governance by providing a centralized repository for data storage, processing, and analysis. This centralization allows organizations to manage data access, security, and compliance more effectively, reducing the risk of data breaches and ensuring regulatory compliance.&lt;/p&gt;

&lt;p&gt;In conclusion, data pipelines are a crucial component of any modern data architecture, providing organizations with the scalability, flexibility, automation, data quality, cost savings, and improved data governance they need to make informed decisions. By adopting data pipelines, organizations can effectively process and analyze data to gain valuable insights and stay ahead of the competition.&lt;/p&gt;

</description>
    </item>
    <item>
      <title>Data! Data! Data!</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Thu, 09 Feb 2023 14:03:00 +0000</pubDate>
      <link>https://dev.to/consolatacleah/data-data-data-bgj</link>
      <guid>https://dev.to/consolatacleah/data-data-data-bgj</guid>
      <description>&lt;p&gt;Data is like the lifeblood of any organization or individual. It helps us make informed decisions and understand the world better. But, not all data is created equal. Just like blood types, some data is O-positive and some is AB-negative. In this blog, we'll take a lighthearted look at some cool examples of databases as well as the unique data qualities.&lt;/p&gt;

&lt;p&gt;But let's not take data too seriously, it's time to have some fun and look at the different characteristics of data and some of the exciting databases that exist in the world today!&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Structure&lt;/strong&gt;&lt;br&gt;
Data can come in different structures, including numerical, categorical, ordinal, and binary. Numerical data consists of numbers that can be quantified, such as age, height, and weight. Categorical data is a set of categories that describe the data, such as gender, occupation, and nationality. Ordinal data is a type of categorical data that has an order, such as movie ratings from best to worst. Finally, binary data is a type of data that only has two categories, such as yes/no, true/false, or male/female.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Volume&lt;/strong&gt;&lt;br&gt;
Data can come in different sizes, ranging from small to big. Small data is usually manageable by humans and can be processed using basic tools like Microsoft Excel. Big data, on the other hand, is too large to be processed by traditional tools and requires special software and hardware to manage.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Velocity&lt;/strong&gt;&lt;br&gt;
Data can be generated at different speeds, ranging from slow to fast. Slow data is generated over a long period of time and can be analyzed in a more relaxed manner. Fast data, on the other hand, is generated in real-time and requires immediate attention and analysis.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Variety&lt;/strong&gt;&lt;br&gt;
Data can come in different forms, including text, images, audio, and video. Text data is the most common form of data and consists of written information. Image data consists of visual information, such as photos and illustrations. Audio data consists of sound recordings, such as music and speech. Video data consists of visual and audio information, such as movies and television shows.&lt;/p&gt;

&lt;h2&gt;
  
  
  Now, let's take a look at some exciting databases that exist in the world today:
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;IMDb (Internet Movie Database)&lt;/strong&gt;&lt;br&gt;
IMDb is a database of information related to movies, TV shows, and celebrities. It includes information about actors, directors, writers, production companies, and more. With over 6.5 million titles and more than 100 million registered users, IMDb is the ultimate destination for movie and TV fans.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Wikipedia&lt;/strong&gt;&lt;br&gt;
Wikipedia is a free, open-source, multilingual, online encyclopedia that is based on a collaborative model. With more than six million articles in English alone, Wikipedia is one of the largest and most comprehensive databases of information on the internet.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;GenBank&lt;/strong&gt;&lt;br&gt;
GenBank is a database of genetic information that includes DNA sequences, protein sequences, and genetic maps. With more than 200 million sequences, GenBank is a valuable resource for researchers in the field of genetics and biotechnology.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;World Bank Open Data&lt;/strong&gt;&lt;br&gt;
The World Bank Open Data is a database of information related to the global economy and development. It includes data on finance, trade, health, education, and more. With more than 3,000 indicators and data from over 210 countries, the World Bank Open Data is a valuable resource for policymakers, academics, and the general public.&lt;/p&gt;

&lt;p&gt;In conclusion, data is a crucial aspect of our lives, and it's important to understand its different characteristics and the databases that exist to make the most of it. So let's embrace data and have some fun exploring the world of information!&lt;/p&gt;

</description>
      <category>analyticaladventure</category>
      <category>datascience</category>
      <category>machinelearning</category>
      <category>webdev</category>
    </item>
    <item>
      <title>Analytical Adventure</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Wed, 01 Feb 2023 18:03:26 +0000</pubDate>
      <link>https://dev.to/consolatacleah/analytical-adventure-4j3b</link>
      <guid>https://dev.to/consolatacleah/analytical-adventure-4j3b</guid>
      <description>&lt;p&gt;"Once upon a time, in a land far, far away (ok, maybe just a few cubicles over), I embarked on a wild and wondrous journey into the world of data. It was a land filled with endless possibilities, where every number and statistic held the key to unlocking the secrets of the universe (or at least, the secrets of your business).&lt;/p&gt;

&lt;p&gt;But as with any great adventure, there were challenges to overcome. The data was vast, overwhelming, and sometimes, just plain confusing. But fear not, my friends! For with the power of datascience, even the most daunting of data problems can be conquered.&lt;/p&gt;

&lt;p&gt;So grab your calculators and your courage, and join me on this analytical adventure as we delve into the world of data and discover the solutions it holds!"&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Data Science&lt;/strong&gt; is an interdisciplinary field that involves using scientific methods, algorithms, and systems to extract knowledge and insights from structured and unstructured data. &lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Machine Learning&lt;/strong&gt; is a subset of Artificial Intelligence that provides systems the ability to automatically learn and improve from experience without being explicitly programmed.&lt;/p&gt;

&lt;p&gt;In data science, data is collected, cleaned, and analyzed to gain insights and make predictions. Machine learning algorithms are then used to build models that can automatically make predictions or decisions based on input data. These models can be used in various applications such as natural language processing, image and speech recognition, and prediction of outcomes in fields such as finance, healthcare, and marketing.&lt;/p&gt;

&lt;p&gt;Exciting advancements are constantly being made in the field, including the development of deep learning algorithms and the increasing availability of big data, leading to improved accuracy and efficiency in prediction and decision making.&lt;/p&gt;

&lt;p&gt;In our next blog we'll get to cover the characteristics of data sources. &lt;/p&gt;

</description>
      <category>watercooler</category>
    </item>
    <item>
      <title>Don't just sit there! Start Power Pages</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Fri, 18 Nov 2022 15:10:35 +0000</pubDate>
      <link>https://dev.to/consolatacleah/dont-just-sit-there-start-power-pages-2a9n</link>
      <guid>https://dev.to/consolatacleah/dont-just-sit-there-start-power-pages-2a9n</guid>
      <description>&lt;p&gt;Hey there, glad to have you here. Ever wondered how you’d create your own website as this one to fuel your business in just a few days/ hours without having to worry about security. Worry not as Power Pages just like redbull gives you wings to fly!!&lt;/p&gt;

&lt;p&gt;This year we get to introduce the newest baby in the power platform family. For businesses already utilizing the Microsoft 365 suite and also new businesses looking to adopt the Microsoft suite.&lt;br&gt;
&lt;strong&gt;Power pages&lt;/strong&gt; is a secure platform for creating, hosting and administering modern external facing business websites using data stored in Microsoft Dataverse.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2F003nsonwzyqp304lwtdm.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2F003nsonwzyqp304lwtdm.png" alt="Powerpages" width="520" height="503"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  Power Pages Capabilities
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Simplified authoring experience for makers: This has been made possible by using available templates, design studio providing styling, data and setup workspace for site management.&lt;/li&gt;
&lt;li&gt;Responsive rendering as power pages is based on Bootstrap hence supporting responsive and mobile friendly applications.&lt;/li&gt;
&lt;li&gt;Advanced custom development for pro developers to extend power pages functionality using visual studio.&lt;/li&gt;
&lt;li&gt;Integration to other Microsoft Power Platform components like dataverse, PowerBi, Power Apps, Automate and Power Virtual Agents.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Pricing
&lt;/h2&gt;

&lt;p&gt;Power pages was previously purchased as Power Apps portals but is now a stand-alone product with two license types:&lt;/p&gt;

&lt;h4&gt;
  
  
  Pre-paid subscription plans
&lt;/h4&gt;

&lt;ul&gt;
&lt;li&gt;Power Pages authenticated users per website (for 100 users/site/month) is priced at $200 with two additional volume tiers. Each subscription plan includes 2GBs of Dataverse Database storage and 16GBs of Dataverse File storage that accrues to the tenant.&lt;/li&gt;
&lt;li&gt;Power Pages anonymous users per website (for 500 users/site/month) is priced at $75 with two additional volume tiers. Each subscription plan includes 500MBs of Dataverse Database storage and 4GBs of Dataverse File storage that accrues to the tenant.&lt;/li&gt;
&lt;/ul&gt;

&lt;h4&gt;
  
  
  Pay-as-you-go
&lt;/h4&gt;

&lt;ul&gt;
&lt;li&gt;Power Pages authenticated user per website is priced at $4 per user.&lt;/li&gt;
&lt;li&gt;Power Pages anonymous users per website is priced at $.30 per user.
Authenticated and anonymous users are consumed/metered only when a user accesses a website in a calendar month, making it easy to deploy websites to broad audiences without assigning licenses to individual users each month.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Sign up for a new trial here to explore further: &lt;a href="https://powerpages.microsoft.com/" rel="noopener noreferrer"&gt;PowerPages&lt;/a&gt;&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;                   Happy Learning!!
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;

</description>
      <category>powerfuldevs</category>
      <category>powerapps</category>
    </item>
    <item>
      <title>Class Diagrams Relationships</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Sun, 23 May 2021 01:46:18 +0000</pubDate>
      <link>https://dev.to/consolatacleah/class-diagrams-relationships-532n</link>
      <guid>https://dev.to/consolatacleah/class-diagrams-relationships-532n</guid>
      <description>&lt;p&gt;Class diagrams are the main building block in object-oriented modeling. They are used to show the different objects in a system, their attributes, their operations and the relationships among them.&lt;/p&gt;

&lt;p&gt;Classes are interrelated to each other in specific ways. In particular, relationships in class diagrams include different types of logical connections. The following are such types of logical connections that are possible in Unified Modelling Language:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Association&lt;/li&gt;
&lt;li&gt;Composition&lt;/li&gt;
&lt;li&gt;Aggregation&lt;/li&gt;
&lt;li&gt;Inheritance&lt;/li&gt;
&lt;li&gt;Generalization&lt;/li&gt;
&lt;li&gt;Specialization&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;An example to understand the relationships &lt;br&gt;
consider a person with his/her gadgets (phones, Laptops etc )&lt;/p&gt;

&lt;p&gt;From here we have 2 classes class person and class gadget&lt;br&gt;
when a person charges a gadget that creates an &lt;strong&gt;Association&lt;/strong&gt;  between the two classes&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Composition&lt;/strong&gt; implies a relationship where the child cannot exist independent of the parent.&lt;br&gt;
Battery is part of a phone &lt;br&gt;
A phone can't exist without a battery hence that is an example of  composition&lt;/p&gt;

&lt;p&gt;A battery and a phone can however exist independently . This is an example of &lt;strong&gt;Aggregation&lt;/strong&gt;&lt;br&gt;
Aggregation implies a relationship where the child can exist independently of the parent&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Generalization&lt;/strong&gt; is a mechanism for combining similar classes of objects into a single, more general class.&lt;br&gt;
Phones and Laptops are examples of Gadgets .&lt;br&gt;
class phones can be a subclass of class Gadgets&lt;br&gt;
This is an example of Generalization as phones inherit attributes from the super class Gadgets&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Specialization&lt;/strong&gt; is the reverse process of Generalization means creating new sub-classes from an existing class hence in the example we could have class phones being created from class Gadgets&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Representation of the relationships&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fpdi36zgy2g4l2gp9ch7p.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fpdi36zgy2g4l2gp9ch7p.png" alt="Alt Text" width="300" height="200"&gt;&lt;/a&gt;&lt;/p&gt;

</description>
    </item>
    <item>
      <title>Nevertheless, Consolata Gicheru Coded in 2021!</title>
      <dc:creator>consolata gicheru</dc:creator>
      <pubDate>Sun, 21 Mar 2021 04:00:40 +0000</pubDate>
      <link>https://dev.to/consolatacleah/nevertheless-consolata-gicheru-coded-in-2021-44e8</link>
      <guid>https://dev.to/consolatacleah/nevertheless-consolata-gicheru-coded-in-2021-44e8</guid>
      <description>&lt;p&gt;&lt;strong&gt;My most recent achievement&lt;/strong&gt; was Speaking at the Global Power Platform Bootcamp in Nairobi.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Advocating for myself looks like&lt;/strong&gt; believing in myself and the skills that I have and pushing to learn more each day and become a better person.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;My biggest goal is&lt;/strong&gt; to be an expert in my field and maybe one day be a Power Platform advocate hence grow my community and empower them to do more.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;My advice for allies to support underrepresented folks who code is&lt;/strong&gt; Kindly give them a platform to showcase their skills and grow their careers.I'd also like them to keep this in mind &lt;strong&gt;A candle loses nothing by lighting another candle&lt;/strong&gt;&lt;/p&gt;

</description>
      <category>wecoded</category>
    </item>
  </channel>
</rss>
