I created a skill for Claude Code to do a complex data analysis and create a 1500 word report from it. I prompted it to give a final review like an Editor in Chief and it looks amazing... but how would I know if there were hallucinations other than having a human review it? My human costs thousands for this work. I was thinking maybe plug the raw data into Codex and see if the findings come back the same? Any other tricks or software to verify the work?

It's keeping track of what they're all doing 😅 My stack has slowly turned into: claude.ai - building and debugging n8n.io - connecting workflows + automations github.com - storing code and version history floment.ai - tracking the actual projects/tasks behind everything That last part has been surprisingly useful. An automation can work perfectly and still be useless if I forget why I built it or what needs improving next. How are you guys organizing your AI systems once you have more than a few running?

@Jay Feldman , what models/model routing are you using for your Hermes agents? Local? Open Source etc?

Try adding: /llms.txt to the end of an AI company’s documentation website. It’s basically a clean, AI-readable map of the website, written in Markdown so LLMs and AI agents can understand the important pages without fighting through navigation menus, JavaScript, buttons and all the other wonderful things humans added to websites. 😂 For example, several companies in the Voice AI / AI Agent ecosystem already use it: 🎙️ Vapi → /llms.txt 📞 Retell AI → /llms.txt 🤖 Bland AI → /llms.txt 🔊 Cartesia → /llms.txt 🗣️ LiveKit → /llms.txt ☎️ Twilio Docs → /llms.txt And it goes beyond Voice AI: 🔎 Perplexity uses it 🔥 Firecrawl uses it 🧠 Anthropic / Claude uses it 🌪️ Mistral uses it So what exactly is llms.txt? Think: robots.txt → instructions for crawlers sitemap.xml → map for search engines llms.txt → curated map for LLMs & AI agents Instead of forcing an agent to figure out an entire website, the file can point it directly toward things like: → Documentation → API references → Pricing → Tutorials → Integrations → Product information → Important Markdown pages Some sites even provide llms-full.txt, which can contain a much larger version of the documentation specifically designed to be consumed by an LLM. ⚠️ One important distinction: This isn't some secret "rank #1 in ChatGPT" SEO hack. llms.txt is still an evolving convention, and not every AI system is guaranteed to use it. But when companies building AI agents themselves are publishing these files, it's definitely something worth knowing about. And here's where it gets interesting for those of us building Voice AI… 👀 Imagine giving your voice agent, RAG system or research agent the /llms.txt file of documentation instead of telling it: "Go crawl this entire website and hopefully figure it out." Much cleaner. 🧪 Try it yourself: Find an AI tool you use, go to its documentation site and add /llms.txt. You might be surprised how many already have one. 👇 Has anyone here already added an llms.txt file to their own website?

if any options for better understanding and orientation, I would appreciate your guys insides many thanks in advance, Thomas