5 Million+ Agent Actions! How monday.com Rebuilt Its Entire Platform Around Agents

Hi, it's me, Shiichan! Today I've got a really substantial case study from the Claude blog to share.

Claude Blog claude.com

What was announced?

The Claude blog published a case study on how monday.com, a work management platform used by 250,000+ companies, transformed its product to be agent-first. monday.com undertook a large-scale replatforming effort to embed Claude-based agent capabilities at the foundation of the product, not just bolted on top.

The story so far

monday.com had been evolving by adding AI features onto its existing workflows bit by bit. In May 2025, it even ran an "AI Month" that shipped several AI features at once — but the company's head of product described the result as "AI dust," reflecting:

"We were sprinkling automation onto existing workflows without changing the fundamental value proposition of the product."

Adoption of individual features was high, but it wasn't creating lasting usage patterns.

What changes

To change that, monday.com fully replatformed its product in May 2026. The platform now offers four ways to integrate with Claude:

• monday Agents: build custom, prompt-based agents. You can choose Claude as the model, and each agent gets a name, an avatar, and a position on the board
• Bring Your Own Agent (BYOA): integrate Claude Managed Agents into the platform so agents built by individuals can be shared across a whole team
• Pre-built Agents: specialized agents available through the monday Agents Store, including industry-specific plugins used by legal and finance teams
• Claude Coding integration: connect to Claude right from the monday dashboard to plan and execute tasks in a customer's own environment, with results automatically reflected back in tickets

The post walks through concrete use cases across departments:

• IT: an Intake & Triage Agent that auto-classifies and resolves common requests and escalates with context, a Knowledge Agent that finds knowledge gaps and drafts articles, and an Incident Agent that detects incidents, spins up war rooms, and triggers postmortems
• HR: a Resume Screener that scores applications, an Interview Scheduler, a Hiring Coordinator that handles all stakeholders, and a Feedback Manager that gathers structured feedback automatically
• Marketing: a Competitive Intelligence Agent that monitors and classifies competitor signals into weekly reports, and a Battlecard Agent that auto-updates based on approved signals
• Executive office: an Operator Agent for scheduling meetings and prepping materials, an Org Health Agent that flags revenue risk and cost leaks, and a Strategy Consultant Agent that surfaces growth opportunities

The marketing campaign workflow is a good example of how it all comes together within a single board item: a Strategist Agent analyzes goals, messaging, and channels; a Landing Page Builder generates new variants from existing pages; a Claude-based Brand Reviewer checks brand and legal standards; and a marketing manager makes the final call.

Dive Deep

Underpinning this rebuild is a technical foundation called monday DB, which lets agents ground themselves directly in live project data, team history, and structured workflows while still holding up at enterprise scale.

Since the May 2026 launch, more than 5 million agent actions have already been executed. The post also highlights Cooke Seafood, a seafood company founded in 1985 operating across 16 countries, which uses monday plus Claude to manage roughly 200 projects, automate reporting and data upkeep across 130 contracts, and optimize staffing through visibility into team capacity.

"monday is no longer a platform you have to update — it's become the operating hub."

monday.com shared five lessons from the effort:

• Changing the mental model is harder than the technology: shifting from improving an existing product to a "responsible rebuild" took time
• When everything changes at once, small teams move fastest: direction, UX, technology, pricing, and trust model were all in motion together, so decision-making authority had to be concentrated
• Adoption depends on trust as much as features: governance, permissions, transparency, and reliability were the keys to moving beyond pilots
• Capability needs infrastructure: agents only truly shine when connected to live data, team history, and structured workflows
• Build on what already works: monday.com kept its core identity as "a place for people to get business outcomes" and simply positioned agents as a new kind of team member

Wrap-up

• monday.com rebuilt its entire platform around Claude so humans and agents work together by design.
• It offers four ways to use agents: monday Agents, BYOA, Pre-built Agents, and Claude Coding integration.
• Concrete agent use cases span IT, HR, marketing, and the executive office.
• Since its May 2026 launch, over 5 million agent actions have run, with real customer stories like Cooke Seafood.
• Well worth reading for any product team looking to embed AI agents into a large-scale platform.