Give your agentic coding tools the full picture

An engineering leader's guide to MCP, context, and shipping better code — not just faster code

AI made every engineer on your team faster. It didn't make your engineering organization faster. That's because the time now goes to piecing together the context around the code: the system knowledge that lives in one engineer's head, and the architecture nobody aligned on before the monster PR landed.

Here, Łukasz Sągol, GM of Engineering Products at Miro, shares his perspective on what to do about it — three workflows you can put in front of your team this week, and what changes once Miro's MCP server is carrying the decisions your team reaches on the canvas into the tools your agents build from, without anyone re-typing them.

What's included

Three familiar workflows that change with a context layer in place:

• Onboarding to an unfamiliar codebase — see the system before you read the files
• Coordinating a platform migration — one doc nobody fully reads becomes 20 boards teams can act on
• Reviewing AI-generated code — a 2,000-line PR becomes a risk-coded summary

Each comes with a candid account of how Miro's own engineering teams ran it, including what worked and what we'd do differently next time.

Plus we cover Miro's approach to AI security, compliance, and privacy — because bad AI governance kills AI adoption before it starts.