Share announcements and updates about company news, upcoming events, or teammates who deserve some kudos. ⭐
267 members
Thursday resource: Anthropic's "Building effective agents" If you're adding agents or multi-step AI to your product, this is one of the clearest free write-ups on when to reach for them and when not to. It was written by Erik S. and Barry Zhang at Anthropic (Dec 2024). No login or payment needed. The page itself notes that much of the tooling has changed since then, but the design patterns still hold up. Who it helps: founders, PMs and engineers deciding how much autonomy to give an LLM feature. Problem it solves: teams jumping straight to complex agent frameworks when a simpler setup would be cheaper and easier to debug. Two practical takeaways
Start with the simplest thing that works. The authors separate "workflows" (LLMs and tools run through predefined code paths) from "agents" (the model directs its own steps). They recommend adding complexity only when it demonstrably improves outcomes, and note that for many apps, a well-optimized single LLM call with retrieval and examples is enough.
Treat tool design like UX. They suggest putting as much effort into the agent-computer interface as into a human one: clear tool docs, testing, and making mistakes hard. One example: changing a tool to always require absolute file paths.
It also walks through five reusable patterns: prompt chaining, routing, parallelization, orchestrator-workers, and evaluator-optimizer. Read it: anthropic.com/engineering/building-effective-agents Which of those patterns is running in your product today? Reply with one and what it does for you.
Sent using Claude
Tuesday tech news: two updates worth a look 1. Notion rebuilds its AI agents and pricing (Oct 19) What changed: Notion says its agents can now work on a single task for more than 24 hours, up from about five minutes. The release adds computer and browser use, a "Review for me" approval mode, and an updated Notion MCP. Business and Enterprise seats now get 1,500 Standard credits (Notion picks the model) and 500 Select credits (you pick the model or run Custom Agents) per month. The new credit system is live for new customers; existing customers move over on Nov 2. BYOK is in alpha for Enterprise only, and an AI Command Center is listed as coming soon. Why it matters: If your team runs workflows on Notion AI, check how the two credit pools map to your usage before Nov 2. Notion's claim that its "Auto" router matches frontier models at about 90% lower cost comes from its own testing, so it's worth comparing on your own tasks. Source: notion.com/blog/notion-ai-more-for-less 2. PostgreSQL 19 release pushed to around Nov 19 (reported Oct 16) What changed: The Register reports that 8 to 10 significant features were pulled from PostgreSQL 19 and moved to version 20, including SQL Property Graph Queries (SQL/PGQ). Committer Peter Eisentraut attributed the reverts partly to LLM-assisted code review surfacing many edge-case issues late in the cycle. Why it matters: If you were planning a Postgres 19 upgrade or counting on SQL/PGQ, adjust your timeline. It's also an early real-world example of AI code review raising the quality bar on a major open-source project. Source: theregister.com/databases/…/5303471 Are you using AI code review on your team yet? Reply with whether it's catching real bugs or mostly noise.
Sent using Claude
Monday question: how do you know an AI feature is actually working? A lot of us have shipped something AI-powered this year: a copilot, a summarizer, an agent that drafts or triages. Launch day is the easy part. The harder part is the first week or two, when you have to decide whether it's earning its place in the product. So I'm curious: what's the one signal you check first after shipping an AI feature? A few examples of what people look at:
Repeat usage (do people come back to it on day 2 and day 7?)
Edit or override rate (how often users change or reject the output)
Support tickets or complaints that mention it
Cost per use versus what the feature is worth to the customer
If you've had a feature that looked great in demos but flopped in real usage, what was the signal that told you? Reply in the thread, short answers welcome.
Hi everyone! 👋 We’re looking for link exchange collaboration opportunities with relevant websites to build mutually beneficial SEO partnerships. If you’re interested, DM me your website, and we can explore relevant contextual link placement opportunities. Note: We’re interested in mutual link exchanges only, not paid link placements. Looking forward to collaborating!
Hi all, Today we launched the waitlist for Agentrail, an intelligence layer that shows companies how AI agents use their product and where they get stuck. If you have a minute, a comment or reshare would really help us reach the right people: LinkedIn: lnkd.in/p/… X: x.com/SidhantGoyal8/status/2109519776678887568?s=20 Thank you for the support
Hey everyone 👋 I'm Paulo, co-founder of Comet ☄️ (getcometapi.com). We help teams run all their AI agents under one roof, including multiple accounts, Slack, Gmail and their own tools, so the agents actually get work done instead of living in separate tabs. I'm also co-founder of Wellnite, I've spent many years on the operator side of SaaS, AI and healthcare. I'm here to swap notes with other founders on agent workflows, distribution, and what's actually working with AI in production. If you're building with agents, or drowning in too many AI seats and accounts, I'd love to chat. Always happy to help where I can, too. Thanks to NachoNacho for the invite!
i’m curious, do you guys send DM’s with calendar links? https://www.linkedin.com/posts/davidgoecke_calendar-link-spam-in-the-dms-is-real-and-share-7494147640629161985-OgNo
