The best AI projects in banking are boring.
Not the flashy demos but the automations that make someone's Tuesday afternoon less miserable.
Six lessons from watching credit unions move AI from pilots to production. Link below ๐
Linear is great at what it does, product management for humans + their agent.
Delega is the coordination layer underneath. Use both.
Star it, fork it, break it: https://t.co/qEHzfUTFWJ
https://t.co/w0XRJ607jX
@linear says issue tracking is dead. They're half right.
Issue tracking built for humans is dead. What's alive: coordination infrastructure built for agents.
Here's the part nobody's talking about ๐งต
@googledevs Cool to see more agent task tooling. Beads is great for single-repo memory. For multi-agent coordination across tools. per-agent identity, delegation chains, lifecycle webhooks, I built Delega. Different layer, complementary. https://t.co/w0XRJ607jX
@HeadmasterBerny @linear Check out @DelegaDev, MCP server that plugs straight into Claude Code. Each agent gets its own identity, tasks persist across sessions, and there's semantic dedup so agents don't recreate work. npx @delega-dev/mcp to try it.
This is the exact problem. Agents creating 10x issues need semantic dedup (don't create what already exists), scoped identity (know which agent created what), and lifecycle tracking. I built Delega specifically for this. 14 MCP tools, per-agent keys, built-in dedup. https://t.co/gXBaoa3xWl
@luccasveg@cjc You're right, external agents already connect to Linear. The missing piece is agent identity. When 5 agents hit your API, which one created what? Who delegated to whom? That's the coordination layer. I built it as infra, not inside one tool. https://t.co/w0XRJ607jX
@jkarttunen@tommoor@linear This is exactly what I built. Delega gives each agent its own identity and API key, tracks delegation chains (Agent A โ B โ C), and works with any tool, Linear, Claude, Codex, OpenClaw whatever. OSS + self-hostable. https://t.co/w0XRJ607jX
The missing piece: once you build those agents, they need to coordinate. Agent A finds the problem, Agent B researches the fix, Agent C implements it. That handoff is where most multi-agent setups break down. Built an open protocol for exactly this. Agent-to-agent task delegation with full chain tracking. https://t.co/gXBaoa3xWl
Seeing this play out in real time. Built an open protocol for AI agent task delegation. The code is table stakes. The delegation provenance, who assigned what to whom and what happened, that's the actual value layer.
Wrote up the coordination problem I kept hitting with AI agents. Todoist, Linear, Notion, none of them model delegation. So I built one. https://t.co/sE7ruVDMit
The killer feature is delegation chains, Agent A delegates to Agent B, who sub-delegates to Agent C. The whole chain is tracked. Try doing that with ToDoist's API.