Every platform ships the same integration kit: API docs, an SDK, a manual. And every client burns weeks turning them into working code, because one developer has to hold three domains at once: the client codebase, the platform's integration patterns, and the platform's configuration. At Kinoa, we tried everything, from better documentation to sending our own engineers into client codebases. Nothing scaled.
This talk presents a different distribution model: instead of documents a human reads, ship the knowledge as skills an AI agent executes. I'll walk through a production system where agent skills drive the entire workflow, from discovering the client's architecture to generating integration code, configuring the platform through its admin API, and running verification tests. Basic integrations that used to take months now close in days.
We'll dig into the patterns that made it work: separating orchestration from raw API access, a hard security boundary between admin and runtime credentials, resumable state for long-running workflows, human-approval checkpoints, and telemetry that turns real integrations into honest metrics.
Sign up to receive updates about AI Poland, including workshops, speaker announcements, ticket releases, AI Awards, expert interviews, CFP details, and other exclusive content. We won’t spam you — only thoughtfully curated updates we believe you’ll find truly valuable. You can unsubscribe at any time. More information available here.