<?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: Jay Butle</title>
    <description>The latest articles on DEV Community by Jay Butle (@jaybutle).</description>
    <link>https://dev.to/jaybutle</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%2F4175728%2F75c075f1-0680-461c-87cd-9f346ced7c7c.jpg</url>
      <title>DEV Community: Jay Butle</title>
      <link>https://dev.to/jaybutle</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/jaybutle"/>
    <language>en</language>
    <item>
      <title>ProofPocket AI: Smart Document &amp; Payment Proof Manager</title>
      <dc:creator>Jay Butle</dc:creator>
      <pubDate>Sun, 18 Oct 2026 04:39:53 +0000</pubDate>
      <link>https://dev.to/jaybutle/proofpocket-ai-smart-document-payment-proof-manager-1ap9</link>
      <guid>https://dev.to/jaybutle/proofpocket-ai-smart-document-payment-proof-manager-1ap9</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week2-2026-10-12"&gt;Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;Freelancers can finish a project and still lose time sorting the paperwork around it: invoices, receipts, due dates, and the awkward question of who still owes what. I built &lt;strong&gt;ProofPocket&lt;/strong&gt;, a local-first paperwork workspace that turns PDFs and images into reviewable business records.&lt;/p&gt;

&lt;p&gt;ProofPocket extracts document text, asks a local Gemma model to suggest fields, and shows those suggestions for correction before anything is saved. Approved invoices feed into client balances, overdue tracking, and reports. The app calculates money and due status in code; the model does not approve invoices or decide financial totals.&lt;/p&gt;

&lt;p&gt;The intended customer is a freelancer, consultant, or small service studio with client paperwork spread across folders. A paid desktop product, setup help, or team workflow could be useful to them. That is a business hypothesis: I have not yet tested willingness to pay with real customers.&lt;/p&gt;

&lt;h2&gt;
  
  
  Demo
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Live preview:&lt;/strong&gt; [&lt;a href="https://pocketproof-psi.vercel.app/" rel="noopener noreferrer"&gt;https://pocketproof-psi.vercel.app/&lt;/a&gt;.]&lt;/p&gt;

&lt;p&gt;The hosted preview is a static, read-only tour with fictional sample records. It does not accept uploads or connect to an AI model. To try the full local workflow, follow the &lt;a href="https://github.com/jaybutle/pocketproof#macos-setup" rel="noopener noreferrer"&gt;macOS setup guide&lt;/a&gt; and &lt;a href="https://github.com/jaybutle/pocketproof#demo-script" rel="noopener noreferrer"&gt;demo walkthrough&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://github.com/jaybutle/pocketproof" rel="noopener noreferrer"&gt;ProofPocket on GitHub&lt;/a&gt; · &lt;a href="https://github.com/jaybutle/pocketproof#proofpocket-ai" rel="noopener noreferrer"&gt;README and architecture&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;p&gt;The interface uses React, TypeScript, and Vite. A local Express API stores approved records and document metadata in SQLite. PDF text is parsed with pdf-parse; images go through Tesseract.js OCR. The API sends bounded extracted text to Gemma through Ollama on the same device, validates the suggested fields with Zod, and presents them for human review.&lt;/p&gt;

&lt;p&gt;The app defaults to Gemma 3 1B through Ollama. I manually smoke-tested a synthetic invoice with Gemma 3 4B and checked the suggested fields against the source. That verified the local extraction adapter, but the full browser upload-to-review flow still needs a recorded walkthrough. Model output can be wrong, so suggestions remain editable and are not treated as verified accounting data.&lt;/p&gt;

&lt;p&gt;The rest of the workflow includes deterministic invoice and client balances, overdue calculations, a record-grounded assistant, and local CSV/JSON exports. See &lt;a href="https://github.com/jaybutle/pocketproof/blob/main/THIRD_PARTY_NOTICES.md" rel="noopener noreferrer"&gt;third-party notices and licenses&lt;/a&gt;.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;Invoices contain names, contact details, rates, and payment terms. With local inference, the document text goes to Ollama on the user's own device instead of a hosted model API, while approved records and source files stay in local storage. A freelancer can avoid per-document AI charges and can choose a compatible model through Ollama.&lt;/p&gt;

&lt;p&gt;That control has trade-offs: users provide the hardware, model setup takes effort, and smaller models can miss fields. The app validates suggestions and requires human review. The local database is not encrypted by ProofPocket, so this is an early productivity tool, not accounting or tax software.&lt;/p&gt;

&lt;h2&gt;
  
  
  My Agent Session
&lt;/h2&gt;


&lt;div class="ltag-agent-session"&gt;
  &lt;div class="agent-session-header"&gt;
    
    &lt;span class="agent-session-tool-icon-badge" title="Codex"&gt;
&lt;/span&gt;
    &lt;span class="agent-session-title"&gt;Preparing ProofPocket's Vercel Demo and Challenge Submission&lt;/span&gt;
  &lt;/div&gt;

  &lt;div class="agent-session-scroll"&gt;

      &lt;div class="agent-session-message agent-session-user"&gt;
        &lt;div class="agent-session-role-badge agent-session-role-user"&gt;
          You
        &lt;/div&gt;
        &lt;div class="agent-session-content"&gt;
                &lt;div class="agent-session-text"&gt;
                  &lt;p&gt;Deploy ProofPocket to Vercel. Its local version stores data in SQLite and files and uses local Ollama; prepare a read-only sample-data demo.&lt;/p&gt;

                &lt;/div&gt;
        &lt;/div&gt;
      &lt;/div&gt;

      &lt;div class="agent-session-message agent-session-assistant"&gt;
        &lt;div class="agent-session-role-badge agent-session-role-assistant"&gt;
          Agent
        &lt;/div&gt;
        &lt;div class="agent-session-content"&gt;
                &lt;div class="agent-session-text"&gt;
                  &lt;p&gt;I separated the hosted preview from the local app. The preview uses fictional records, has no API or upload path, and is configured as a static Vercel build.&lt;/p&gt;

                &lt;/div&gt;
        &lt;/div&gt;
      &lt;/div&gt;

      &lt;div class="agent-session-message agent-session-assistant"&gt;
        &lt;div class="agent-session-role-badge agent-session-role-assistant"&gt;
          Agent
        &lt;/div&gt;
        &lt;div class="agent-session-content"&gt;
              &lt;div class="agent-session-tool-call"&gt;
                
                  &lt;span class="agent-session-chevron"&gt;▸&lt;/span&gt;
                  &lt;span class="tool-name tool-name-execcommand"&gt;exec_command&lt;/span&gt;
                    &lt;span class="tool-input-preview"&gt;npm run build &amp;amp;&amp;amp; npm run build:demo; npx tsc --noEmit; inspect built demo bun...&lt;/span&gt;
                
                &lt;div class="agent-session-tool-detail"&gt;
                    &lt;div class="tool-section"&gt;
                      &lt;div class="tool-section-label"&gt;Input&lt;/div&gt;
                      &lt;pre class="agent-session-pre agent-session-terminal"&gt;&lt;code&gt;$ npm run build &amp;amp;&amp;amp; npm run build:demo; npx tsc --noEmit; inspect built demo bundle for /api references&lt;/code&gt;&lt;/pre&gt;
                    &lt;/div&gt;
                    &lt;div class="tool-section"&gt;
                      &lt;div class="tool-section-label"&gt;Output&lt;/div&gt;
                      &lt;pre class="agent-session-pre"&gt;&lt;code&gt;TypeScript check and both production builds passed. The demo bundle was separate, lazy-loads the chart, and contained no /api references.&lt;/code&gt;&lt;/pre&gt;
                    &lt;/div&gt;
                &lt;/div&gt;
              &lt;/div&gt;
        &lt;/div&gt;
      &lt;/div&gt;

      &lt;div class="agent-session-message agent-session-assistant"&gt;
        &lt;div class="agent-session-role-badge agent-session-role-assistant"&gt;
          Agent
        &lt;/div&gt;
        &lt;div class="agent-session-content"&gt;
              &lt;div class="agent-session-tool-call"&gt;
                
                  &lt;span class="agent-session-chevron"&gt;▸&lt;/span&gt;
                  &lt;span class="tool-name tool-name-execcommand"&gt;exec_command&lt;/span&gt;
                    &lt;span class="tool-input-preview"&gt;git push origin main&lt;/span&gt;
                
                &lt;div class="agent-session-tool-detail"&gt;
                    &lt;div class="tool-section"&gt;
                      &lt;div class="tool-section-label"&gt;Input&lt;/div&gt;
                      &lt;pre class="agent-session-pre agent-session-terminal"&gt;&lt;code&gt;$ git push origin main&lt;/code&gt;&lt;/pre&gt;
                    &lt;/div&gt;
                    &lt;div class="tool-section"&gt;
                      &lt;div class="tool-section-label"&gt;Output&lt;/div&gt;
                      &lt;pre class="agent-session-pre"&gt;&lt;code&gt;Pushed the Vercel demo configuration and submission link updates to https://github.com/jaybutle/pocketproof.&lt;/code&gt;&lt;/pre&gt;
                    &lt;/div&gt;
                &lt;/div&gt;
              &lt;/div&gt;
        &lt;/div&gt;
      &lt;/div&gt;

      &lt;div class="agent-session-message agent-session-user"&gt;
        &lt;div class="agent-session-role-badge agent-session-role-user"&gt;
          You
        &lt;/div&gt;
        &lt;div class="agent-session-content"&gt;
                &lt;div class="agent-session-text"&gt;
                  &lt;p&gt;Give all data to fill the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle template.&lt;/p&gt;

                &lt;/div&gt;
        &lt;/div&gt;
      &lt;/div&gt;

      &lt;div class="agent-session-message agent-session-assistant"&gt;
        &lt;div class="agent-session-role-badge agent-session-role-assistant"&gt;
          Agent
        &lt;/div&gt;
        &lt;div class="agent-session-content"&gt;
                &lt;div class="agent-session-text"&gt;
                  &lt;p&gt;Prepared a DEV submission describing Gemma through Ollama, local document processing, human review, deterministic financial calculations, business hypothesis, and the read-only sample demo. Kept the deployment URL explicit as a placeholder and did not invent first-user feedback.&lt;/p&gt;

                &lt;/div&gt;
        &lt;/div&gt;
      &lt;/div&gt;

      &lt;div class="agent-session-message agent-session-user"&gt;
        &lt;div class="agent-session-role-badge agent-session-role-user"&gt;
          You
        &lt;/div&gt;
        &lt;div class="agent-session-content"&gt;
                &lt;div class="agent-session-text"&gt;
                  &lt;p&gt;Give My Agent Session and My First User.&lt;/p&gt;

                &lt;/div&gt;
        &lt;/div&gt;
      &lt;/div&gt;
  &lt;/div&gt;

  &lt;div class="agent-session-footer"&gt;
    &lt;span class="agent-session-meta"&gt;
        7 of 7 messages
    &lt;/span&gt;
  &lt;/div&gt;
&lt;/div&gt;


&lt;h2&gt;
  
  
  My First User
&lt;/h2&gt;

&lt;p&gt;I don't have an external first user yet, so I can't report user feedback or payment. My next step is to ask a freelancer to try ProofPocket and learn which part of the paperwork workflow would save them the most time.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
    </item>
    <item>
      <title>TerraQuest AI: Turning “I Should Go Outside” Into a 15-Minute Adventure</title>
      <dc:creator>Jay Butle</dc:creator>
      <pubDate>Sun, 11 Oct 2026 15:38:44 +0000</pubDate>
      <link>https://dev.to/jaybutle/terraquest-ai-turning-i-should-go-outside-into-a-15-minute-adventure-11p0</link>
      <guid>https://dev.to/jaybutle/terraquest-ai-turning-i-should-go-outside-into-a-15-minute-adventure-11p0</guid>
      <description>&lt;p&gt;What I Built&lt;/p&gt;

&lt;p&gt;TerraQuest AI is a personalized outdoor-quest generator that turns a vague intention like “I should spend less time on my screen” into a small, specific activity you can start in minutes.&lt;/p&gt;

&lt;p&gt;Instead of handing you a generic list of things to do, TerraQuest asks a few quick questions and shapes a quest around your day:&lt;/p&gt;

&lt;p&gt;Activity type: for example, a mindful walk&lt;br&gt;
Available time: 5, 10, 15, 30 or 60 minutes&lt;br&gt;
Energy level: from easy and up&lt;br&gt;
Accessibility needs: seated options, low-energy days, flat routes&lt;br&gt;
Interests: what you’d like to notice, such as birds, moss or cloud shapes&lt;br&gt;
Eco-mission: an optional small act of care for nature, or a surprise&lt;/p&gt;

&lt;p&gt;Every quest comes with an accessible alternative, so weather, energy or terrain never leaves you without an option.&lt;/p&gt;

&lt;p&gt;How it gets people off the screen&lt;/p&gt;

&lt;p&gt;Discover: a daily “Today’s invitation” lowers the barrier to starting.&lt;br&gt;
My quests: saved quests you can return to and complete.&lt;br&gt;
Nature journal: a place to record observations and moments from outside.&lt;br&gt;
Passport: a local record of minutes outdoors, quests completed, activities explored and weekly activity, with a progress bar toward a 30-minute goal. Quiet milestone badges such as First Quest, Curious Observer and Seven Outdoor Breaks mark your progress without streak pressure.&lt;/p&gt;

&lt;p&gt;Who it’s for: students, remote workers and anyone who spends long hours on a screen and wants a gentle, low-pressure nudge outside. It’s built for all energy levels and abilities, which is why accessibility is an input to every quest rather than an afterthought.&lt;/p&gt;

&lt;p&gt;Privacy by design: no account, no precise location and no tracking. Your progress is stored locally on your device.&lt;/p&gt;

&lt;p&gt;Demo&lt;/p&gt;

&lt;p&gt;🎥 Video walkthrough:&lt;/p&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/IvheSH1XAEo" width="710" height="399"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;

&lt;p&gt;🔗 Live app: &lt;a href="https://terraquest-ai-iota.vercel.app" rel="noopener noreferrer"&gt;https://terraquest-ai-iota.vercel.app&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Code&lt;/p&gt;


&lt;div class="ltag-github-readme-tag"&gt;
  &lt;div class="readme-overview"&gt;
    &lt;h2&gt;
      &lt;img src="https://assets.dev.to/assets/github-logo-5a155e1f9a670af7944dd5e12375bc76ed542ea80224905ecaf878b9157cdefc.svg" alt="GitHub logo"&gt;
      &lt;a href="https://github.com/jaybutle" rel="noopener noreferrer"&gt;
        jaybutle
      &lt;/a&gt; / &lt;a href="https://github.com/jaybutle/terraquest-ai" rel="noopener noreferrer"&gt;
        terraquest-ai
      &lt;/a&gt;
    &lt;/h2&gt;
    &lt;h3&gt;
      
    &lt;/h3&gt;
  &lt;/div&gt;
  &lt;div class="ltag-github-body"&gt;
    
&lt;div id="readme" class="md"&gt;&lt;div class="markdown-heading"&gt;
&lt;h1 class="heading-element"&gt;TerraQuest AI&lt;/h1&gt;
&lt;/div&gt;
&lt;p&gt;&lt;strong&gt;Less scrolling. More exploring.&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;TerraQuest AI is a local-first nature companion for making small outdoor adventures feel easy to start. This first implementation focuses on the core experience: discovering activities, generating a practical quest, saving it in the browser, completing it in a low-distraction mode, and keeping a nature journal.&lt;/p&gt;
&lt;div class="markdown-heading"&gt;
&lt;h2 class="heading-element"&gt;Current core features&lt;/h2&gt;
&lt;/div&gt;
&lt;ul&gt;
&lt;li&gt;Responsive landing / Discover page with activity suggestions and real local impact counters.&lt;/li&gt;
&lt;li&gt;Quest creator with activity, duration, difficulty, and observation-interest inputs.&lt;/li&gt;
&lt;li&gt;Quest Council boundary: Gemma/Ollama generation, structured output validation, and clearly labeled curated fallback.&lt;/li&gt;
&lt;li&gt;Saved, in-progress, completed, searchable-by-tab quest views.&lt;/li&gt;
&lt;li&gt;Active quest mode with checklist and optional timer.&lt;/li&gt;
&lt;li&gt;Nature journal with add, edit, and delete flows.&lt;/li&gt;
&lt;li&gt;Browser &lt;code&gt;localStorage&lt;/code&gt; persistence for quests and journal entries.&lt;/li&gt;
&lt;li&gt;Nature Passport, Eco-Missions, accessibility preferences, photo attachments, export/reset controls.&lt;/li&gt;
&lt;li&gt;Express local API at &lt;code&gt;POST /api/quests/generate&lt;/code&gt; and &lt;code&gt;GET /api/health&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;Online/offline status indicator and truthful local-storage messaging.&lt;/li&gt;
&lt;li&gt;Reduced-motion-friendly transitions and keyboard-focus…&lt;/li&gt;
&lt;/ul&gt;&lt;/div&gt;
  &lt;/div&gt;
  &lt;div class="gh-btn-container"&gt;&lt;a class="gh-btn" href="https://github.com/jaybutle/terraquest-ai" rel="noopener noreferrer"&gt;View on GitHub&lt;/a&gt;&lt;/div&gt;
&lt;/div&gt;


&lt;p&gt;How I Built It&lt;/p&gt;

&lt;p&gt;TerraQuest AI uses Gemma, an open-weight model, served through Ollama to generate quests from the user’s preferences. The model runs locally on my Mac, so I could build and experiment with open-weight AI without depending on a closed, paid AI API for inference.&lt;/p&gt;

&lt;p&gt;How the AI fits in&lt;/p&gt;

&lt;p&gt;The quest form collects the user’s activity type, eco-mission, available time, energy level, accessibility needs and interests.&lt;br&gt;
These inputs are combined into a structured prompt and sent to Gemma through Ollama.&lt;br&gt;
Gemma generates a quest: a title, a short invitation, simple steps, something to notice, and an accessible alternative.&lt;br&gt;
Completed quests, minutes outdoors and journal entries are stored locally and feed the Passport and its badges.&lt;/p&gt;

&lt;p&gt;Design decisions&lt;/p&gt;

&lt;p&gt;Small and specific: a good quest is something you can start right now, not a plan you’ll put off.&lt;br&gt;
Calm, nature-inspired design: the interface should feel like an invitation, not a productivity tool.&lt;br&gt;
Gentle progress: quiet milestones instead of guilt-inducing streaks.&lt;br&gt;
Accessibility built in: low-energy and seated alternatives are part of quest generation from the start.&lt;br&gt;
Local-first: user data stays on the device.&lt;br&gt;
Why Does Open Innovation Matter?&lt;/p&gt;

&lt;p&gt;TerraQuest is built on the belief that getting outside should be accessible, private and low-pressure. Open innovation made that possible in ways a closed API wouldn’t have:&lt;/p&gt;

&lt;p&gt;Privacy: with a locally run model, a user’s preferences, energy levels and accessibility needs never have to leave their machine.&lt;br&gt;
No paywall or API dependency: I could build, test and iterate freely, without usage costs, rate limits or deprecations from a single vendor.&lt;br&gt;
Control and customization: open weights let me inspect, prompt and tune the model so quests respect real-world constraints like limited time, low energy or mobility needs.&lt;br&gt;
Community potential: anyone can fork the project, swap in a different model, or adapt it for schools, parks and community groups.&lt;/p&gt;

&lt;p&gt;A closed API could have generated quests too, but it couldn’t have offered the same privacy, independence and room for others to build on it.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
      <category>programming</category>
      <category>webdev</category>
    </item>
  </channel>
</rss>
