monday.com made this year’s Elevate NYC wonderfully easy to judge for once, promising everyone attending the event a chance to build their very own working AI agent.
Sounds fun. Although whoever ends up managing those agents back at the office might have a few questions. On October 8, Gartner named monday.com a Leader in collaborative work management for the fourth consecutive year, so clearly, the company’s still making a good impression. It also scored highest in three Critical Capabilities use cases, including Work Planning and Execution. As CPTO Daniel Lereya said in the announcement:
“Our strength has always been the core of work management: helping teams plan, track, and deliver collaboratively with clarity.”
Still, now it’s time to see what the agents can actually do for project teams. We covered monday.com’s AI Work Platform shift in May. Elevate is where we find out how far it’s come.
The opening keynote starts at 11 a.m. Eastern on October 28, after this morning edition. That makes today’s assignment a little simpler: set the test now, then check monday.com’s answers when they’re available.
We’re setting three tests before the keynote: what customers can actually use, whether anyone’s getting measurable results from agents in production, and what happens to the bill once those agents start working every day.
Because building an agent in a workshop is one thing. Trusting it to turn a signed client deal into a properly staffed delivery project is another.
What Can monday.com’s AI Agents Do, and Who Can Use Them?
The most gripping question for Elevate NYC is how far monday.com has come since introducing its agent builder last year. In March, it opened the platform to external agents that can organize projects and update workflows. We covered that launch at the time.
There’s already useful functionality to build on, too. Dependency and Risk Mapper traces critical paths and flags threatened deadlines. Status Reporter gathers board updates without somebody chasing everyone on Friday afternoon. Project Intake organizes incoming requests, while Approvals Agent routes sign-offs and follows up when decisions get stuck.
There are resource-management templates, as well, like the Resource Rebalancing Agent that can identify overloaded teams and propose changes, although that’s different from automatically reorganizing an entire project portfolio.
Catching a problem isn’t the same as fixing the schedule, though. If a design approval slips, can an agent work out which teams are affected, recommend a sensible adjustment, and get the right person’s approval?
The agent builder lets users describe jobs in plain English, connect boards and configure scheduled or triggered work without writing code. For now, though. monday.com currently lists native agents as available on its AI platform, with availability in other products still marked “coming soon.” Creating or managing agents also requires administrator permissions.
Tools on Demand, which builds integrations from natural-language requests, remains in beta. Bring Your Own Agent is also in beta and requires administrator-enabled permissions.
So can a project manager recreate an Elevate demonstration on the subscription they already pay for? An agent built in a configured conference workshop isn’t automatically available back at the office.
Otherwise, that’s a rather expensive souvenir.
Can monday.com Agents Really Make Better Project Staffing Decisions?
Monday’s Enterprise Portfolio Risk Insights already reviews connected project boards daily, while AI Smart Assignment recommends staffing based on skills and availability. A human confirms assignments, and neither feature currently consumes AI credits. So what would a metered agent add?
Currently, monday.com’s marketing talks about identifying portfolio risks in real time, but its support guide says Risk Insights refreshes the list daily. Maybe those descriptions cover different processes. Either way, if a developer becomes unavailable halfway through Tuesday, when does the PMO learn that Thursday’s deadline is in trouble?
On October 22, monday dev highlighted Capacity Planner for workload, velocity, and sprint availability. Its guidance covers planned sprints on Pro and Enterprise, rather than autonomous staffing across an entire portfolio. The same update brought VS Code and GitLab integrations, connecting development tasks more closely with project records.
We’ve already explored why project capacity matters more than another completed task, and how dependencies complicate AI project management. Now Elevate needs to show whether an agent can create a delivery project after a deal closes, find an available manager, and arrange the handoff without disrupting existing commitments.
As Christopher Richardson of PMI wrote on October 22:
“AI can accelerate the work, but it cannot own the work.”
There’s a reliability question, too. monday.com reported EU automation delays on October 19, caused by queue congestion and resolved that morning. One incident proves little, but how would a project manager detect and recover a missed handoff?
How Much Do monday.com AI Agents Really Cost to Run?
It’s still not a hundred percent certain what a working project agent will cost. monday.com lists AI credits at $0.01 apiece with annual billing, or $0.0125 with monthly billing. Under its May purchasing model, new AI Work Platform customers buy credits alongside seats, with minimum monthly allocations ranging from 1,000 on Basic to 3,000 on Pro. Promotions may offset those charges.
Now for the expensive bit. monday.com estimates 50–150 credits for an intermediate agent task and 150–250 for a complex one. A single run can involve several tasks, while model choice, retries, and corrections can change consumption.
Imagine ten project agents completing one 100-credit task each working day. Across 20 days, that’s 20,000 credits, worth $200 with annual billing or $250 monthly. That isn’t necessarily an extra invoice, since purchased allocations and discounts affect the bill.
There’s also the question of what you’re paying for. Portfolio Risk Insights and AI Smart Assignment currently consume no AI credits. Why pay an agent to repeat work your subscription already covers?
External agents make the calculation tougher. Under the beta Bring Your Own Agent model, outside reasoning incurs the other provider’s charges, ordinary board updates count against API quotas, and monday AI actions consume credits. Connecting the agent itself currently costs nothing extra.
monday.com documents alerts at 80% and 100% usage, but buyers should ask what happens to essential project workflows when allocations run out.
How Does monday.com Compare With Asana, Atlassian and Other AI Project Tools?
monday.com has plenty of competition when it comes to getting project work assigned and keeping task boards up to date. Some rivals are already tackling the less glamorous problems that make project managers want to throw their laptops out of the window.
Asana’s October 22 roundup highlighted automation across tasks and portfolios, including round-robin assignment for incoming requests. Handy when the marketing team gets fifteen new jobs on Monday morning, although splitting requests evenly doesn’t necessarily account for who’s already overloaded.
