The whole shelf

Loaded FHIR docs into Cursor; free tier is generous.

And dead easy to set up.

Built my own deep agent without LangChain, but their tutorial is a nice starting point.

Great integration into langchain ecosystem, native langsmith support. But is rather slow latency wise, therefore weak for interactive agents. Switching to omp as a coding agent for example here resulted in less scaffolding and faster runtimes of the coding agent.

For anyone dealing with OCR — optical context compression.

I use this — can recommend.

Nice abstraction for durable agents, and TS tool definitions with type safety make it easy for an agent to write its own tools. Downside — like a lot of Vercel's stuff — is the tight coupling to their infra.

Prompt-based extract gives much better content than default scraping, but costs more credits and is slower; the default pulls noisy headers/footers.

A lighter alternative to Firecrawl.

Open-source and free; does what we need, and we picked it over a similar paid tool.

Easy to setup, great to manage llm endpoints from different providers. Nice for first level tracing/observability (tracking costs and budget of deparments for example) or token cost optimisation via headroom as u set it up once and it works for the whole company.

Tackles long-term agent memory by distilling salient facts — reports ~26% better recall with big latency and token-cost savings.

Maps an agent's file edits onto an animated 3D codebase map — great eye-candy to hook an audience during a CAIO demo or presentation.

we used this instead of deepagents at a client. It is generally more efficient, faster, better compaction or context management. Generally superseeds deepagents. There are two ways to use it! Either use it as claude code as your coding harness, or call it via omp -rpc for the autonomouses mode It works with the claude subscription, but anthropics past showed that they ban people for that as its against their tos

Openhands works really well even with small local models

Worked better than Cursor in my tests.

Can recommend working spec-driven.

Open-source project that apparently supersedes CodeRabbit for code reviews.

Evaluated it as immature a while back, but they've been cooking — now has eval, observability and a gateway in the ecosystem.

Easy to deploy, easy to connect to confluence or other documentation databases. Solves a few low hanging fruits, like customer support assistant or to connect documentation to other agents via their mcp connection.

Free AI scraping tool — possibly a game changer.

Can recommend working spec-driven.

Considering it to lean on managed services; note it isn't widely known yet (Dennis hadn't come across it).

Extremely powerful — controls any desktop app via vision. Very much recommend testing it.