Home/Blog/monday.com Agents Powered by Make: When to Build an Agent on Your Boards, and When a Plain Automation Is Enough
AI AutomationNovember 10, 2026· Team Nuevexa

monday.com Agents Powered by Make: When to Build an Agent on Your Boards, and When a Plain Automation Is Enough

monday.com now lets Make power agents your team can chat with, mention or assign. Here is when an agent earns its place, when a board rule is enough and what stays human.

monday.com Agents Powered by Make: When to Build an Agent on Your Boards, and When a Plain Automation Is Enough - Nuevexa

Most monday.com accounts in a growing service business carry a quiet layer of rules. When a status changes to Done, notify the account lead. When a date arrives, move the item to a new group. These board automations do one job each, and they do it the same way every time. For a lot of work, that is exactly what you want.

Now there is a second option sitting on the same boards. monday.com's Bring Your Own Agent feature lets you connect an agent built elsewhere, and Make is one of the routes in. Your team can chat with the agent, @mention it on an item or assign it work like a colleague, while a Make scenario does the work behind the scenes. monday.com put agents at the centre of its Elevate '26 updates, with the promise that you decide where you stay in the loop and what AI can do on its own.

So the real question for an ops lead or founder is not whether you can put an agent on your boards. It is which jobs actually need one, which are better left as a plain rule or a Make scenario, and what should stay with a person. This guide covers what changed, how the Make setup works and a simple way to decide.

What does Bring Your Own Agent mean for monday.com users?

According to monday.com's support article on external agents, an external agent becomes its own entity in your account. People can chat with it, mention it on items, assign it tasks and trigger it from board events or a schedule. Its instructions, skills and tools stay on the platform where it was built. monday.com acts as the governance layer and the record of what it did.

Three details matter for a small team:

  • It is still in beta. The feature is labelled as beta, and the Make modules behind it are marked beta too.
  • It starts with no access. External agents follow an allowlist model. A new agent cannot see any boards or docs, including public ones, until someone grants access. Nothing is inherited from the person who connected it.
  • Its actions carry its own name. Every action is logged under the agent's identity in the activity log, and admins can deactivate an external agent at any time from the Agent directory.

That last point is what separates an agent from simply pointing an AI chat window at your account. The same article notes that an MCP connection runs with the full permissions of the person using it and logs actions under that person.

What can a board automation already handle on its own?

A board automation is a rule. monday.com's getting started guide describes every automation as a trigger, a condition and one or more actions. When a status changes, only if priority is high, notify the team lead.

Rules are good at work where the right answer is already known:

  • Moving items between groups when a status changes
  • Assigning an owner when a new request arrives
  • Sending a reminder when a due date is close
  • Creating a subitem checklist when a project starts

monday.com is also testing a new automation builder, currently in beta for a closed group of users, with an AI smart condition and a Generate with AI block for light judgement inside a rule, such as summarising or translating a text column. Once that reaches your account, rules will cover even more ground.

If you can write the logic as "when this happens, and this is true, do that", a board automation is usually the right home. It is predictable, it is easy for anyone on the team to read and it does not need a separate tool to maintain. Each run does use actions from your account's automation quota, as monday.com explains in its actions and limits article, so it is worth keeping an eye on busy boards.

When does a plain Make scenario make more sense?

Some rules need to reach outside monday.com, or run several steps in order. A new deal is won in your CRM, so a project board is created, a folder is set up and the client gets a welcome email. That is still fixed logic, but it spans several tools.

This is where a plain Make scenario fits. The monday.com app in Make can watch board items, create and update items, post updates and work with files, groups and subscribers. A board change can start the scenario instantly through a webhook, and Make handles the steps in other systems.

A few practical points from Make's documentation:

  • Check your module version. Version 1 of the monday.com app in Make is now legacy. Existing scenarios should be upgraded to Version 2 to keep running.
  • Know whose token it uses. A standard monday.com connection in Make uses a personal API token, so the scenario acts with that person's access.
  • Plan for failure. Make can store incomplete executions so a failed run can be retried or resolved by hand, but this is off by default.

A Make scenario is still deterministic. It will do the same thing every time, which is a strength. It just cannot read a messy request and work out what someone meant.

When is a monday.com agent actually worth building?

An agent earns its place when the input is unstructured and the next step depends on reading it. A client update that could be a change request, a complaint or a simple question. A long item thread someone needs summarised before a call. A request to "pull together everything open for this client" across several boards.

monday.com's developer guide for Make agents describes three ways to trigger one:

  • Chat. Someone messages the agent directly and it keeps the conversation history for follow-up questions.
  • Mention. Someone @mentions the agent in an update, and it receives the item's context.
  • Assign. Someone assigns the agent on an item, for example through a column on the item.

Good early jobs for a service business tend to be read-heavy. Summarise an item's history before a client call. Draft a reply to a client update for an account lead to review. Check a new request against the board and suggest which group and owner it belongs to. In each case the agent saves time on reading and drafting, and a person still makes the call.

If you are weighing an agent against an approval step more broadly, our guide to what to connect, approve and keep off-limits for always-on agents covers the same thinking outside monday.com.

How do you set up a monday.com agent with Make?

The setup follows the developer guide and Make's monday.com app documentation:

  1. Create the agent in monday.com. Go to Agents, then Bring your own agent, then Make. Name the agent and copy its API key. monday.com shows the key once, so store it safely.
  2. Open the Monday Agent template in Make. The scenario arrives with a trigger, a router that splits chat, mention and assign events, a Make AI Agent on the chat branch and a reply step.
  3. Connect the trigger. In Watch Agent Events, create a monday.com Agent connection with the agent's API key and save. This creates the webhook and attaches it to your agent.
  4. Replace the placeholders. Out of the box, the mention and assign branches reply with a placeholder message saying the agent is not yet configured. Swap them for your own logic.
  5. Give the agent board access. Add the monday.com MCP server to the Make AI Agent so it can read and act on boards, then grant the agent access only to the boards it needs.
  6. Write tight instructions. The guide suggests pointing the agent at a specific board and including its ID. Make's AI agent best practices also recommend clear tool names, descriptions and guardrails.
  7. Test each trigger. Chat with the agent, mention it on a test item and assign it once, then check the run in Make and the reply in monday.com.

Two gotchas are worth knowing early. A monday.com agent supports one webhook at a time, so creating a new one for the same agent silently replaces the old one. And if the agent cannot see a board, some Make modules return an empty list rather than an access error, which can look like "no results" when it is really a permissions gap.

Board automation, Make scenario or agent: how do you decide?

Use this as a first pass. Most teams will end up with all three, each doing the job it suits.

If the job looks like thisUseWhat stays with a person
A fixed rule on one board: status, date, owner, notificationBoard automationDeciding what the rule should be and reviewing it when the process changes
Fixed steps across several tools, triggered by a board changePlain Make scenarioHandling failed runs and approving anything that sends money or messages
Reading a messy request, summarising a thread or drafting a replyAgent, read and draft onlySending the reply and making the decision
Updating items or routing work based on what a request saysAgent with write access to named boardsSpot checks on the activity log and anything high-value or unusual
Anything touching payments, contracts, access or client commitmentsNot an agent on its ownThe whole decision, with the agent at most preparing a draft

A simple rule of thumb: if you can write the logic down, use a rule or a scenario. If the job needs reading and judgement, consider an agent. If getting it wrong would cost money, damage a client relationship or change who has access to what, keep a person in the loop every time.

What should stay with a person?

monday.com's own framing is that AI does the work and you make the call. Its native agents add new write tools as inactive until someone switches them on, according to the agent builder guide. That is a sensible default to copy when the agent's logic lives in Make.

In practice, keep these with a person:

  • Client-facing messages. Let the agent draft. Let a person send.
  • Money and commitments. Quotes, invoices, refunds and anything that promises a date or a deliverable.
  • Deletions and access. Archiving or deleting items, adding board members and changing permissions.
  • Exceptions. Anything the agent flags as unclear should go to a named owner, not back into the queue.

It also helps to decide who owns the agent. Someone should review its activity log, update its instructions when the process changes and pause it if it starts doing something unexpected.

The takeaway: start with the rule, add judgement where it pays

Agents on monday.com boards are a real step forward, and the Make route gives you full control over what the agent does behind the scenes. But most of the work on a typical board does not need judgement. It needs a clear rule that runs every time.

Start by listing the jobs your team repeats each week. Put the predictable ones into board automations or Make scenarios first. Then pick one read-heavy job, such as summarising an item before a client call, and trial an agent there with access to a single board. Widen its access only once the activity log shows it doing the right thing.

If you would rather not build and maintain this yourselves, Nuevexa works as a done-for-you implementation partner and is an authorised monday.com partner. We set up the automations, Make scenarios and agents inside the tools you already pay for, alongside our wider workflow automation work.

To see which of your monday.com workflows suit a rule and which might suit an agent, take the automation scan.

Ready to automate?

Book a free strategy call and leave with a clear roadmap for your first automation build.

Book a call

Tags

AI AgentsWorkflow AutomationAI Automation
TN

Team Nuevexa

Building intelligent systems that replace manual work.

The 90-second scan

Ninety seconds to see where your leads are leaking.

We build and run the lead capture and follow-up system for accounting, advisory, legal and trades firms with 10 to 200 staff. Fixed price. Your accounts. Live in about three weeks.

Take the 90-second scan

Frequently Asked Questions

If manual work is your ceiling, let's remove it.

We reply within 1 business day.