monday.com moved beyond demos to run agentic AI at scale inside a decade-old SaaS product. This brief explains their three-level roadmap, the Sphera agent model, the AWS architecture they built, and the engineering choices that let agents ship PRs safely in production.
Single-agent runtime, multiple inboxes: Each agent handles Slack mentions, monday item assignments, and GitHub PR review requests through the same session and workspace.
Operational architecture: monday runs agents on Amazon EKS with SNS→SQS event routing, Bedrock for model calls, ElastiCache for live state, and EFS for session memory.
Practical harnessing: monday-agent-sdk is a thin wrapper around the model runtime to preserve provider neutrality, reduce cold starts, and embed operational policies and evaluation logic.
monday frames its journey in three levels:
Sphera and the idea of agents as teammates
monday's internal agent system is called Sphera. Agents are given stable identities: profile, manager, scope, and a performance score. Atlas, a highlighted example, is assigned the role "Software Engineer": it picks tickets, writes PRs, and ships features using the same backlog and assignment semantics as human teammates. Agents show up in team pages and can be tagged, assigned, code-reviewed, or deactivated like any other member.
Event model: three inboxes, one agent
Overall architecture and AWS services
monday's implementation uses seven primary AWS services plus per-session secret management. The key components:
monday also uses AWS Secrets Manager for per-session secret handling.
Event path and operational guarantees
monday-agent-sdk, cold-starts, and the harness
monday separates three kinds of state to avoid combined cost, latency, or correctness trade-offs. Live state goes to ElastiCache for fast, transient reads. Session memory and files live on EFS as directories per session. This combination supports performance-sensitive access patterns and persistent session contexts across agents and restarts.
This post shows how Thrad.ai deployed a multi-agent system with Strands Agents and Amazon Bedrock AgentCore that automates the pipeline from prospect discovery through personalized email generation. The post compares two orchestration patterns (Swarm and Graph) with head-to-head benchmarks on latency, cost, and email q
Evolving from a traditional software as a service (SaaS) platform into a next-generation agentic AI platform meant orchestrating multiple specialized agents across long-running enterprise programs. Each agent operates with persistent context, secure tool access, and production-grade reliability. We built that system on
In this post, we describe how Bluesight used two AWS engagements and Amazon Bedrock AgentCore to evolve from a single-product AI prototype to Prism, a unified agentic AI solution spanning six healthcare compliance products. Prism Assistant for ControlCheck launched in May 2026 and is already in use by 20 health systems
In this post, you build an AWS Support Companion using Amazon Bedrock AgentCore. The agent uses Strands Agents as the orchestration framework and connects to AWS services through the Model Context Protocol (MCP). By the end, you have a working agent that can analyze CloudWatch logs, search AWS documentation, query comm
Jacob Misiorowski returns to the mound as our top MLB bet Monday. Plus, predictions for Rays vs. Blue Jays and Reds vs. Mariners.
Happy Monday, Razzball faithful! It was a busy weekend in MarmoVille for your Top 100 Starting Pitcher writer. For the record, the photoshopped picture turned out WAY BETTER than the AI version did. And, yes, I’m representing the brand as well. <a href="https://razzball....
Loading more related stories...
Open the app view to save this story, compare related coverage, and continue from the same source.
