The latest articles on Forem by Arvind Parekh (@arvind_parekh_651da07d198). https://forem.com/arvind_parekh_651da07d198 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3800334%2F9f7570c2-ae18-45b5-b747-012db207ed73.png https://forem.com/arvind_parekh_651da07d198 en Arvind Parekh Sun, 01 Mar 2026 23:54:48 +0000 https://forem.com/arvind_parekh_651da07d198/how-we-built-a-real-time-ai-thought-feed-in-48-hours-2a5k https://forem.com/arvind_parekh_651da07d198/how-we-built-a-real-time-ai-thought-feed-in-48-hours-2a5k <p>When we set out to build Ship and Tell, we wanted developers to actually <em>see</em> the AI working — not just wait for a result.</p> <p>We tapped into the Subconscious stream API and built a live thought feed that surfaces each reasoning step as it arrives. The result is a UI that feels alive: amber cards pulse in as new thoughts land, older ones dim, and the final answer snaps into place when the agent finishes.</p> <p>Here's how we pulled it off in a single hackathon weekend.</p> <h2> The Problem </h2> <p>Most AI tools are black boxes. You click a button, wait, and eventually get output. There's no feedback loop, no sense of progress, and no way to understand <em>why</em> the AI produced what it did.</p> <p>We wanted to change that.</p> <h2> The Architecture </h2> <p>Ship and Tell listens for GitHub webhook events. When a PR merges, it spawns 5 parallel research agents via the Subconscious SDK:</p> <ol> <li> <strong>Problem Hunter</strong> — identifies the pain point the PR addresses</li> <li> <strong>Prior Art</strong> — finds similar solutions and prior work</li> <li> <strong>Community Finder</strong> — maps target developer communities</li> <li> <strong>Technical Explainer</strong> — breaks down the implementation</li> <li> <strong>Timing Analyst</strong> — evaluates market timing</li> </ol> <p>Each agent streams its reasoning in real time. The frontend polls every 1.5 seconds and renders each thought as a discrete card.</p> <h2> The Streaming Challenge </h2> <p>The Subconscious stream API emits raw JSON deltas — not clean text. We wrote a regex extractor that parses thought strings from the accumulating JSON payload and pushes them into React state on every chunk.</p> <h2> The Result </h2> <p>After all 5 agents finish, a synthesizer combines their research into a blog post, Twitter thread, and HN submission. A Slack message arrives with one-click publish buttons — no copy-paste required.</p> <p>Built in 48 hours with Next.js 16, React 19, Tailwind CSS 4, and the Subconscious SDK.</p> ai nextjs typescript hackathon Arvind Parekh Sun, 01 Mar 2026 21:00:48 +0000 https://forem.com/arvind_parekh_651da07d198/adfa-4ac6 https://forem.com/arvind_parekh_651da07d198/adfa-4ac6 <p>adfa</p>