We’ve spent most of October watching software vendors give AI agents increasingly impressive job descriptions. Apparently, they can handle just about everything a project manager hates doing now. We’re still waiting to hear who’s cleaning up when they make a mess of the schedule.
That’s what makes this week’s work management news so interesting.
Microsoft and Alphabet’s October 28 earnings and monday.com’s more than 40 Elevate demos give us plenty to examine. PMI’s Global Summit added something I’d argue is more valuable: evidence of projects actually getting finished. And new releases from Linear, Notion, Jira and Wrike have given project teams more than a keynote to mull over.
Back in July, Microsoft CEO Satya Nadella said the goal was to
“Turn tokens into business results.”
Is that actually happening? It’s about time we found out.
TL;DR:
-
Microsoft Copilot earnings: On October 28, Microsoft reported nearly 50% quarterly growth in daily Copilot users and an eightfold annual increase in customers with over 50,000 seats. It gave no new overall seat count or project-delivery measure.
-
monday.com Elevate announcements: monday Frontier will combine managed services with forward-deployed engineers. The Done Awards also highlighted Clearwater’s 20-agent project handoff system, with concrete activity figures but no independently verified improvement in on-time project delivery.
-
Project-management news: Wrike’s October 26 update stops simple agent actions using AI Elite allowances; Linear improved developer handoffs, while Notion’s November 2 credit transition and Jira’s new coding-agent link put costs and execution in focus.
-
Work management renewals 2027: AI runs are another expense to account for. I’d check what a correctly completed workflow costs, including rework, before signing.
What Did Microsoft’s October Earnings Reveal About Copilot Adoption?
Microsoft’s got quite a sales story to tell. In July, it reported more than 30 million paid Microsoft 365 Copilot seats, with net additions more than doubling quarter over quarter. Satya Nadella also said enterprises deploying Copilot to most of their information workers had increased nearly 75%. He called it
“a signal of how central Copilot has become to their operations.”
On October 28, Nadella said daily Copilot users had risen nearly 50% quarter over quarter and more than fivefold over the year. Customers with more than 50,000 Copilot seats increased over eightfold year over year, and customers consuming usage-based AI credits grew more than 75% during the quarter. Nadella put the business shift plainly:
“Our business model is expanding from per-seat subscriptions to include consumption.”
That’s massive momentum. Still, I’d be wary of treating licenses as a proxy for completed projects. Microsoft 365 Copilot seats also tell us nothing by themselves about Microsoft Planner adoption. A PMO facing missed deadlines needs evidence that Copilot catches slipping dependencies or frees up capacity. Bigger deployment figures won’t fix a badly staffed project plan.
Alphabet’s October 28 results raise a similar question. Google Cloud reported more than 100 million seats sold for knowledge work across its AI services, and said nearly 90% of the Fortune 100 use Gemini Enterprise. That isn’t a standalone Workspace or Gemini Enterprise paid-seat total, and neither figure tells us whether project deadlines are improving.
Google’s Workspace connectors can retrieve project information from Asana and monday.com. Useful, certainly. But finding a delayed task is rather different from getting it back on schedule. We’ve already covered the integration opportunity, though.
What matters now is paid adoption, sustained use, measurable changes in project delivery and cost per completed workflow.
monday.com had quite a lot to prove at Elevate. In May, the company recast itself as an AI Work Platform, promising agents that could execute tasks using existing workspace permissions.
In September, co-CEO Roy Mann said,
“AI is opening up an entirely new way of working, where people and agents can accomplish more together.”
We still need to know how much more work customers are actually getting done. Preferably before another project misses its deadline.
The clearest new move from Elevate was monday Frontier, combining managed services and forward-deployed engineers to help enterprise customers deploy AI. At an October 28 investor session, co-CEO Eran Zinman said enterprise customers were adopting AI agents at roughly three times the rate of SMBs. That’s adoption, rather than a verified PMO delivery improvement.
Elevate offered a more concrete project example than an agenda promise. monday.com’s Done Awards highlighted Clearwater’s 20-agent Project Intelligence Network. According to monday.com, the team has 508 live projects and logged more than 3,071 agent actions in its first two months, with an estimated 10,000-plus hours returned annually. Those are customer and vendor-reported figures, not independently audited reductions in project delays.
At Elevate, monday.com’s Ryan Nigro emphasized giving agents minimum necessary permissions. Sensible when agents edit project records.
Monday left some things unanswered, but Frontier at least addresses the difficult part of getting agents into real workflows. I’ll be interested to see how these updates affect project managers on a day-to-day basis.
Which Work Management AI Update Will Change the Bill?
Notion’s October 19 overhaul raises a more useful question for project managers: how much will it cost to keep an AI agent working?
Its agents now have browser and computer access and can work for up to 24 hours. Notion says Auto matched premium models on more than 20 tasks at about 90% lower average cost. That’s its own test, not proven project savings.
Existing Business and Enterprise workspaces move to Notion’s new AI credit system on November 2. Typical monthly allowances are 1,500 Standard and 500 Select credits per member. Balances cannot be pooled or rolled over, and allowances vary by plan.
By default, a shared Custom Agent spends its owner’s Select credits. Administrators can fund an agent with workspace Notion credits instead. Without sufficient credits, the agent pauses until allowances reset or more credits become available.
My renewal test? Whose allocation does a long-running agent use, when does it stop, and what do failed runs cost? I’d want the bill before approving more credits.
Can AI Stop One Staffing Decision From Derailing Three Projects?
There’s an annoying habit in project planning. An urgent project gets the strongest person available, and suddenly three other deadlines become somebody else’s problem. I’d love an allocation tool that could tell a manager exactly what they’re about to break.
Wrike’s Chief Product Officer, Alexey Korotich, made a good point:
“If effort estimates are missing or skills information is outdated, AI might suggest someone who looks like a good fit on paper but doesn’t actually have the capacity to get the work done, putting the entire project at risk.”
I think that deserves more attention than another promise of faster status reports. An AI agent might find the perfect technical match and completely miss the fact that the person’s already committed to two major releases.
Wrike says it’s developing AI resource allocation capabilities that consider skills, availability, and existing workloads, with managers retaining control over final assignments. Sensible, although I’d want to see how accurately those recommendations reflect what’s happening across an entire portfolio.
My test? Reassign one developer and show me which other deadlines are likely to move. Then let the project manager decide whether the disruption is worth it.
There’s an annoying habit in software projects. Someone marks a change ready for testing, everybody assumes the handoff happened, and the QA plan is still sitting in someone’s head two days later. I’d rather have the trigger attached to the work itself.
Linear’s October 22 update gives Loops more ways to do that. Events in Slack and GitHub can trigger an automated workflow. Linear’s example is a pull request marked “ready for testing” kicking off a draft QA plan, rather than someone remembering to ask for one.
Teams can save Loops as shared templates and select reasoning levels to manage cost. Linear’s agent can also read and update Figma designs through MCP.
Useful, certainly. Still, generating a QA plan isn’t the same as getting the tests run. I’d want to know who reviews the output, where the resulting tasks land, and what happens if the GitHub trigger fires twice.
My test? Start with a real pull request, create the QA plan, and show me who owns the work by Friday. Then let the project manager decide whether the automation has saved anybody a chase.
Talking Point: Do Earnings Change Work Management Software Renewals?
I’d be pretty concerned if a strong earnings call were enough to convince a PMO to approve another year of AI spending. Great sales numbers tell us customers are buying. Finance still has to deal with the bill when agents waste credits repeating jobs, and project managers are left fixing the results.
Iron Mountain’s October 19 survey found 86% reporting AI project delays or cancellations over data concerns, although this isn’t a PM software measure. If your project records are a mess, paying an agent to rewrite a portfolio forecast sounds like an expensive way to spread the confusion.
Iron Mountain’s Varun Gupta warned:
“Enterprises need to know that the information powering their AI is accessible, current and governed, with clear ownership and permissions.”
I’d agree.
These figures concern AI projects generally, rather than measured improvements from project-management software. Still, I’d want some pretty convincing evidence before signing another contract.
In Other News: Project Results, New Releases and What’s Next
A few things to watch:
1. PMI’s October Awards: What Does Successful Project Delivery Look Like? PMI named Saudi Arabia’s Ministry of Human Resources and Social Development World PMO of the Year. Its enterprise PMO oversees more than 370 initiatives and 1,800 projects; PMI reports 98% strategy achievement and 94% project-delivery performance. MTN Ghana won the African regional award, with project completion reportedly rising from 40% to 80%. None of those results should automatically be credited to AI.
2. Wrike’s October 26 Agent Update: Simple actions, including fixed status changes and copying fields, can now run without consuming AI Elite allowances. Agents already in use must be reopened and saved; requests involving interpretation or conditions still use AI. For project managers, that’s a practical example of why usage charges deserve a closer look.
3. Jira’s Factory Droid Integration: Atlassian’s October 27 release lets engineers invoke a coding agent from Jira through Rovo Chat. It writes results back to the issue; Jira Cloud with Rovo is required. I’d still want proof it saves developers time.
I’d be particularly interested in anything that helps managers catch a resource clash before it turns into another missed deadline.
Looking ahead, Atlassian reports FY27 Q1 earnings on November 5. After October’s Rovo and Jira announcements, we’ll be watching for evidence of actual customer adoption.
Microsoft Ignite follows November 17–20, and monday.com brings Elevate to London on November 24. Both should give project leaders more to consider before committing to next year’s software budgets.
Which brings us to your 2027 renewals. What survived the budget meeting, and what got cut? We’d love to hear what finally settled the argument.
Final Thoughts: October’s Promises Still Need to Deliver
We’ve heard plenty this month about AI agents taking responsibility for project work. Microsoft’s earnings and monday.com’s Elevate announcements give us a chance to see how much of that ambition customers are actually benefiting from.
Personally, I’m more interested in whether a project manager can reassign someone without wrecking three other deadlines, or catch a problem before it costs another week.
That’s what I’d want to see in next year’s software proposals. Give me evidence of better project delivery, sensible costs, and managers who can explain the decisions being made.
The impressive AI demos can come afterward.
FAQs
What Did Microsoft Disclose About Copilot Usage in October 2026?
On October 28, Microsoft reported nearly 50% quarterly growth in daily Copilot users and more than eightfold annual growth in customers with over 50,000 seats. It did not provide a new absolute paid-seat total or a measure of projects delivered faster through Copilot. I’d want those delivery outcomes before taking the adoption figures into a renewal meeting.
monday.com lists native agents on its AI platform, with availability in other products still marked “coming soon,” and administrators are required to create and manage agents. Elevate also introduced monday Frontier, an enterprise implementation initiative rather than a newly available agent plan. Its detailed eligibility and pricing weren’t disclosed . A demo doesn’t prove a project manager can reproduce anything.
What Did PMI’s October Awards Reveal About Project Delivery?
Saudi Arabia’s Ministry of Human Resources and Social Development won World PMO of the Year. PMI says its enterprise PMO oversees more than 1,800 projects and achieved 94% project-delivery performance. MTN Ghana won the African regional award. Useful delivery evidence, but not proof that AI caused the improvements.
What Should PMOs Compare When Renewing AI Work Management Tools for 2027?
Give each vendor an awkward, recurring job. Say, rebuilding a portfolio report after three projects miss their deadlines. What does the finished report cost, including corrections? We’d want that answer before agreeing to another year’s AI charges.
