monday.com says AI agents lifted PR throughput by more than 50%
AWS ML Blog · rss · 2026-07-22
AWS ML Blog explains how monday.com runs production AI agents at scale on Amazon Bedrock, backed by its own internal usage data.
What changed in practice
- Nearly 9 in 10 builders now use AI coding tools monthly, up from roughly half a year earlier.
- Per-engineer PR throughput has increased by more than 50%.
- monday treats agents as teammates with identities, managers, scopes, and performance scores rather than as a sidecar tool.
Architecture highlights
- One agent session can be triggered from three inboxes: Slack mentions, monday item assignments, and GitHub PR review requests.
- The system uses SNS → SQS → EKS, with Bedrock for model calls and a thin monday-agent-sdk wrapper around the Claude Agent SDK.
- State is split by purpose: ElastiCache for live state, EFS for sessions and memory, and S3 for durable transcripts, artifacts, and evals.
- Bedrock helps with inference profiles, auditability, cross-region failover, and PrivateLink/VPC isolation.
Why it matters
The post emphasizes that greenfield demos are easy; the hard part is retrofitting agents into a decade-old SaaS codebase with real customers, compliance, retries, back-pressure, and resume-after-crash behavior.
More from coding & agent
- GPT-6 Astra beats Factorio with enemies in 44 in-game hours at ~$4,500 API cost — liminal_bardo · 2026-09-11
- Investment Analyst Asks How to Build a Claude-Based Diligence Agent Stack — Careless_Tie2286 · 2026-09-11
- Treating agents like 50 First Dates: a 3-layer context system so every conversation doesn't start from zero — evielync · 2026-09-11
- Running the Firefox MCP on Android via Termux, ngrok, and mcp-proxy — Nervous-Strain7544 · 2026-09-11
- SmolVM open-sources persistent computer infrastructure for agents that outlive chat sessions — aniketmaurya · 2026-09-11
- ARRM targets silent economic regressions in AI agents that functional tests miss — Beautiful_Belt_601 · 2026-09-11