Sid Probstein

I am working on a bigger/badder integration now... 100% open source

Full details: github.com/swirlai/swirl-… and github.com/swirlai/swirl-…

Team SWIRL is delighted to announce the release of SWIRL Community 4.5!! This version includes: * Support for executing RAG with most LLMs and optional LLM instructions * New MCP server for easy integration with Claude or any MCP client * New Galaxy UI featuring search history & the return of dark mode * New admin console featuring bulk editing support for SearchProviders, built-in activity analytics, log viewer, JSON paste support and more! * Revised source selector now groups sources by tag and shows the tag selected instead of listing the underlying sources * Updated re-ranking model improves precision and reduces response time * Improved error handling, especially during AI insight generation For full details, see the link in the comments... congratulations @SWIRL_SEARCH for this milestone release!!

Instructions here: docs.swirlaiconnect.com/quick-start

Critical Chain Project Management en.wikipedia.org/wiki/Critical_…

THE RULE OF THREE Never let an AI agent try to fix the same bug more than three times. Attempt one is useful. Attempt two is plausible. Attempt three tells you something. Attempt four is where the agent renames three variables, invents a caching problem, and starts “cleaning up the tests.” No. After the third miss, stop the loop. Document what happened. Summarize the current state. Capture the exact error, files touched, theories tested, and what changed. Then critique the work. What did we assume? What did we miss? What did the agent confidently make worse? Then start a new task with clean context. Agents are great. But after three failed tries, you don’t have a coding assistant... you have a slot machine with a GitHub token.

MikeOSS continues its unstoppable acquisition spree. Larvey.ai now redirects to MikeOSS.com.

Many of them won't work unless you get API permission/access from the vendor first... and it's not always free.

SWIRL MCP Server for Community Edition: github.com/swirlai/swirl-…

Trust, but verify

BREAKING: SGA HAS WON MVP

Gordon Ramsay would be an extraordinary agentic coding lead. Think about it. His secret isn't talent. Every kitchen has talented chefs. His secret is that before anyone touches a pan, the standards are absolute, documented, and non-negotiable. The mise en place. The brigade system. The recipes. Every cook knows: you do not freelance in Gordon Ramsay's kitchen. Deviate? He finds out. Immediately. Loudly. "An inline panel?! We use MODALS in this kitchen! GET OUT!" This is exactly what your AI coding agent needs. Because here's the thing about agents: they are incredibly talented chefs who will happily cook whatever they feel like if you don't tell them otherwise. Leave them unsupervised across three sessions and you come back to a codebase with three different UI paradigms, two new dependency patterns, and a data model that made sense to nobody but the agent that invented it at 2am. Gordon doesn't leave chefs unsupervised. And he puts the standards in writing. Every service. Every shift. That's your prompt. Every task prompt needs three things: → What to build → The actual context (code, specs, contracts) → The standards: use existing patterns, ask before introducing anything new, no freelancing The last one is the Gordon Ramsay clause. It goes in every prompt. Not just the first one. Because the agent doesn't remember the last session. But Gordon would never let that be an excuse. "I don't CARE that it was a new context window. This is RAW." Be Gordon. Write the standards down. Put them in every prompt. Your codebase will thank you. Probably without screaming.

All tests pass!!

continueFromHereClaude

🚀 Major release of SWIRL Community coming this weekend !! Includes: ✅ RAG & embeddings support for most LLMs e.g. mxbai-embed-large via ollama ✅ Updated re-rank model ✅ New search UI with history ✅ New admin UI with basic activity dashboard If you want to integrate SWIRL Community with MikeOSS, wait for this version!!

The old source control discipline was designed around the human as the bottleneck. With coding agents, it's not. The natural unit isn't per-issue anymore ... it's per-day. One branch, one ticket, one PR. Commit when tests pass. New post → swirlaiconnect.com/blog/source-co…

Your AI gives confident answers. Confident isn't the same as correct. In enterprise environments, the right answer isn't just accurate ... it's the version your organization has actually approved. The document legal signed off on last Tuesday, not the draft someone left in SharePoint in 2023. The policy that's current, not the one from Q3. Current AI search stacks retrieve the most semantically similar content. They don't know what your organization has decided to believe. SWIRL 5 introduces the knowledge authority layer. Every result is ranked not just by relevance, but by source authority, version currency, and organizational endorsement. When two documents say different things, SWIRL surfaces which one your organization considers canonical - and why. Your AI stops being confidently wrong. That's what SWIRL 5 is built around. → swirlaiconnect.com #EnterpriseAI #KnowledgeManagement #AISearch #RAG #EnterpriseSearch #SWIRL5

i asked claude opus 4.7 to refactor a large codebase. 68 minutes, millions of tokens burned - it finished nothing worked. app completely broken

Last week I watched someone use an enterprise AI assistant during a meeting. It started well. A question came up about a customer escalation. The AI produced a polished answer in seconds. Everyone around the table nodded. Then someone asked the obvious question: “Where did that come from?” Silence. The user started clicking around. SharePoint. Teams. Outlook. Another SharePoint tab. Running searches. A few minutes later, nobody cared about the answer anymore. The room had shifted from “this AI is useful” to “can we trust this thing?” That moment captures the real problem with enterprise AI... the bottleneck is not generation. It’s verification. A lot of AI systems still work like this: RAG → hunt for the source. Generate first. Prove later. That works in demos, but internally, and especially with messy data, it breaks trust fast. People should not have to go on a scavenger hunt after the AI answers the question. The workflow should be: Search + RAG Search. Verify. Then ... Generate. And verify again, with deeplink citations. When SWIRL retrieves Microsoft 365 content, users can jump directly to the source document and see the contributing text highlighted. Not “according to your data.” Actual evidence. Some repositories still make deeplink citation difficult. We’re working on that too. But I’m convinced this is where enterprise AI is heading. The winners will not be the systems that sound smartest. They’ll be the systems users can verify instantly. If you’re thinking about enterprise AI architecture right now, this problem is worth paying attention to.

90% of companies be like