AI agents for COOs leading high performing orgs
Already have an invite? Sign in
Have your team removing waste, with a number to show for it
Untoil finds the organizational waste across your company and prices every piece of it in annual hours. Two things fall out of that:
Your operations talent removes waste instead of chasing it
The interviews, the process audits, the guessing about where the time goes: finding waste is the part that eats an operations team's quarter. Untoil's agents run that hunt continuously, so your people start every week with the target list already made and spend their time on the fixes that move the number.
You get one metric for toil, and proof when it drops
Untoil totals the known waste into a single figure: the annual hours toil consumes at your company. Every fix your team ships moves it down, which gives operations something it rarely has: a return the rest of the business can see.
Every company is buying AI agents; almost none can say where to point them. Operations leaders can name their headcount, spend, and SLAs, but not the twenty processes quietly consuming the most hours, because that work was never instrumented. It lives in threads, spreadsheets, and workarounds nobody logged. The cost of not knowing is easy to compute and rarely computed:
- 12,500
- Hours a year burned by one 30 minute daily task across 100 people, at 250 working days.
- 6 roles
- The same figure as headcount, dividing by 2,080 hours per full-time equivalent.
- Zero
- How many of those hours appear in a standard operations dashboard, because nothing was ever tracking them.
What is an AI agent for a COO?
An AI agent for a COO is software that observes how a company actually operates and reports back on it without being asked. Untoil is one: its agents read the channels, documents, and threads a company already produces, identify the repetitive manual work inside them, estimate the annual hours each one consumes, and rank the result by cost so the operations leader knows what to fix next.
The category splits in two. Most tools sold as AI agents execute a task you have already scoped, such as answering a ticket or moving data between systems. A smaller group, Untoil included, works the other end of the problem: deciding which process is worth automating in the first place.
What Untoil's agents do for an operations team
Find the work nobody files a ticket for
Untoil's agents read the channels, threads, and shared documents your company already produces, then flag the repetitive operational work inside them. Nobody has to fill in a form, so the expensive problems surface even when the team has stopped complaining about them.
#operations
UntoilUntoil AI is scanning for toilSam · 9:14
shipping the Q3 deck now 🎉
Ana · 9:31
spent all morning re-typing vendor invoices into the portal again 😩
Toil found · est. 340 hrs/yrJo · 9:42
lunch orders in the thread 🌮
Put an annual hour count on each one
Every issue is scored for how long it takes, how often it happens, and how many people it hits. Untoil turns that into annual hours and a full-time-equivalent figure, so an operational complaint arrives on your desk already denominated in the unit you plan headcount in.
Someone vents
Devin · #engineeringhonestly every friday i lose an hour copying everyone's updates out of slack and jira into a doc i'm pretty sure nobody even reads 🙄
UntoilUntoil AI is estimating this toilUntoil makes it valuable
Weekly status reports are copy-paste hell
- Frequency
- Every week
- People affected
- 25
- Time each
- ~30 min
Let the affected teams confirm the scope
Anyone living the same problem can co-sign it, add what the estimate missed, or correct the frequency. The number stops being one manager's guess and becomes something the people doing the work have signed their name to.
Rank the whole backlog by what it costs
Everything rolls into one list ordered by annual cost. That is the artifact most operations teams are missing: not a dashboard of what happened, but a ranked queue of what to fix next and what fixing it is worth.
How is this different from process mining, surveys, and consultants?
All four try to answer the same question. They differ in where they look, whether the answer comes with a number, and how quickly it goes stale.
| Approach | What it finds | Prices it? | How current |
|---|---|---|---|
| Employee surveys | What people remember to mention in the window you asked | No, answers stay qualitative | Quarterly at best |
| Process mining | Steps inside instrumented systems like your ERP or CRM | Yes, for the systems it is connected to | Continuous, after a long integration project |
| Operations consultants | Whatever the engagement scope covers | Yes, in a deliverable | Once, then it goes stale |
| Untoil | Repetitive work described anywhere your teams already talk or write | Yes, annual hours and FTE per issue | Continuous, from the tools you already use |
Process mining remains the stronger tool once a process already runs inside an instrumented system end to end. Untoil is aimed at everything that never made it into one.
What does one repetitive task cost per year?
Very. A team of 100 losing just 20 minutes a day is like having 4.0 full-time employees doing nothing but toil, all year.
Team Members
Min per day each
8,333
Hours lost per year 🤦
What operational toil looks like when Untoil finds it
Three examples, priced the way every issue arrives.
Filing expense reports takes half a day
Every trip means re-typing receipts into the old expense portal, chasing approvals over email, and re-submitting the whole thing when one field is wrong.
Weekly status reports are copy-paste hell
Every Friday the whole team copies updates out of Slack and Jira into a doc by hand, reformats it, and pastes it into an email nobody reads.
Onboarding a new hire is 100 manual steps
Creating accounts, requesting access, and setting up tools is a checklist we redo by hand for every single hire across five different systems.
Where do the agents look?
Untoil reads the tools your company already works in. Most operational toil is never written down anywhere else: it lives in Slack threads, Google Drive files, and Microsoft Teams channels.
- SlackChannels and threads
- Google DriveDocs and files
- Google ChatSpaces
- Microsoft TeamsChannels and chats
- OneDriveShared files
- WebsiteYour company site
Connecting those tools gives Untoil the richest context about the friction happening inside your company. We use that context to find work that is stuck, duplicated, or waiting on someone, then surface it so your team can see it, price it, and decide what to fix.
Without integrations, Untoil only knows what people remember to log. With Slack, Google, and Microsoft connected, we can help catch the expensive problems even when nobody files a thing.
Questions COOs ask
- What can an AI agent actually do for a COO today?
- It can read where work is discussed and surface the repetitive parts, estimate what each one costs per year, and keep that list current without anyone maintaining it. Untoil does this rather than executing the work itself, because the harder problem for most operations leaders is knowing which process to point automation at, not running the automation.
- How is this different from RPA or process mining?
- Process mining and RPA both start from systems that are already instrumented, so they see the steps inside your ERP or CRM and miss everything that happens in Slack threads, spreadsheets, and email. Untoil starts from how people describe their work, which is where undocumented toil actually lives, and it needs no system integration project to begin.
- How do you measure ROI on an AI agent for operations?
- Compare the annual hours recovered against the cost of the fix. Untoil gives every issue an annual hours figure and a full-time-equivalent number before anyone commits engineering time, so the business case exists at the moment you prioritize rather than after the project ships.
- Does Untoil read private messages?
- No. Autonomous scans only read public channels and spaces the workspace has invited Untoil into. Direct messages and private group chats are never scanned. A person can always post something manually from a private conversation, but the agents will not go looking there.
- How long before we have a ranked list?
- Connecting a workspace takes a few minutes and the first scan runs immediately after. Most teams have a ranked, priced list of operational toil the same day, because Untoil reads history rather than waiting to collect new data.
- Which tools does Untoil connect to?
- Slack, Google Drive, Google Chat, Microsoft Teams, OneDrive, and your public company website. Teams can also post toil by email, through a browser extension, or from an AI assistant connected over MCP.
Find out what your operation actually costs
Connect one workspace and Untoil returns a ranked, priced list of the repetitive work inside it. Untoil is invite-only while we onboard teams by hand.
Already have an invite? Sign in
Last updated August 26, 2026