Skip to main content
Flowstream

Monday.com

What’s new in monday.com in 2026: agents, MCP and AI credits

monday.com relaunched as an AI Work Platform in 2026. Here are the updates that matter for teams running their operations on boards, and what to do about them.

· 2 min read · Flowstream

A laptop showing code on screen in a dark room lit in blue
Photo: Nemuel Sereti on Pexels

2026 has been monday.com’s biggest year of change in a long time. In May, the company repositioned itself from a work management platform to an AI Work Platform, and most of what followed builds on that. Here is what changed, in plain terms, and what we think teams should do about it.

monday agents and AI connectors

The May relaunch introduced monday agents: AI agents that non-technical users can set up to work on live board data, under human supervision. It also added one-click connectors to Claude, ChatGPT and Microsoft 365 Copilot, and an AI Platform Gateway for working with more than one model.

What it means for you: agents are most useful where a board already holds clean, structured data. If your boards are a mix of subtasks, free-text updates and duplicated items, an agent will mostly automate the confusion. Structure first, then agents.

Sidekick becomes the main way into AI

Sidekick, monday’s context-aware assistant, left beta and became the main entry point for AI in the product. It reads context from boards, docs and people, connects to email and Slack, and through 2026 gained the ability to create dashboards from a described need and to work with sub-items, connected and mirrored columns, and folders.

AI Workflows, MCP and human approval

Two summer updates matter for anyone automating on monday.com:

  • monday MCP: monday’s hosted MCP server lets external assistants work with your workspace data, and agents can now create, read and disable automations without the interface.
  • AI Workflows: multi-step automations gained an MCP block that can act on third-party apps with a public MCP server, and a human-in-the-loop approval block that routes the workflow based on someone’s response.

What it means for you: the approval block is the one we would use first. It lets AI draft or decide, while a person signs off on anything that matters.

Seats plus AI credits

With its first-quarter results, monday.com moved to a pricing model that combines a price per seat with a monthly pool of AI credits that grows with the plan. Admins also gained per-user AI credit limits and a usage dashboard. Check the current pricing page before you plan a rollout, because AI-heavy workflows now have a running cost of their own.

CRM and admin updates

monday CRM added validation rules per lifecycle stage, triggers on emails and activities, deals linked to quotes and invoices, and AI that fills CRM fields after calls. For larger accounts, managed templates now scale to far more instances, with view sync and data validation across them.

What we would do now

  1. Clean the structure before adding agents. One item per real thing, statuses that mean one thing, and a central board that the rest of the workspace syncs into.
  2. Start with one supervised workflow. Pick a repetitive step, add AI, and put an approval block in front of anything customer-facing.
  3. Watch your AI credits. Set per-user limits early, so a single enthusiastic automation doesn’t use the whole pool.

If you want help deciding where these features fit your operation, that is exactly what our discovery call is for.

Book a Discovery CallFREE← All articles

More articles

Ready to connect your systems?

Start with a free discovery call. We’ll map how your business runs today and where better systems would help most.

Book a Discovery CallFREEContact Us