The tools that displaced you can make you an owner

Reuters reported that Meta is installing tracking software on US-based employee computers to capture mouse movements, clicks, and keystrokes for AI model training. The tool is called Model Capability Initiative. Reuters said it runs on work-related apps and websites and may take occasional screen snapshots for context. Meta told Reuters the data is for model training, not performance reviews, and said safeguards protect sensitive content.

The report gives the narrow fact pattern. The larger fact is simpler: AI is an agency multiplier.

An institution will use it to multiply institutional agency. It will watch how work gets done, capture the steps, train agents on the pattern, and move more action into systems the institution owns. The worker may keep a job, lose a job, manage the agent, train the agent, review the agent, or become the trace that taught the agent.

The surveillance critique is true, but it is not enough. Stopping there leaves the worker in the same position: inside someone else’s system, arguing over how much of their workday can be harvested.

The harder question is what the worker can do with the same class of tools.

Agency Belongs To Whoever Owns The Work

A company wants agents that can use everyday software because everyday software is where the business lives. Dropdown menus, keyboard shortcuts, CRM fields, spreadsheets, ticket queues, finance portals, email threads, calendar invites, form packets, and status changes are not small details. They are the motor movements of modern office work.

When the company captures those movements, it is trying to make the work legible to machines it controls.

A worker needs the opposite move. Make your judgment legible to a business you can control.

Not every laid-off product manager, analyst, lawyer, recruiter, accountant, designer, marketer, or operations lead should become a solo founder tomorrow. The old gap between “I know how the work should happen” and “I can run a business around that judgment” is shrinking.

The tools that threaten the employee inside a large company can help that same person operate outside one.

The Worker Has More Than Clicks

Mouse traces are thin. Keystrokes are thin. Screen snapshots are thin. They show action, but not the whole reason for the action.

A capable worker knows what the customer actually meant, which exception matters, and which vendor cannot be trusted. They know the record that will satisfy the reviewer, the promise that should not be made, the follow-up that will save the account, the missing document that will break the case next week, and the Slack thread where the real decision happened.

Judgment is the asset.

Inside a large company, the asset often gets trapped in meetings, tickets, inboxes, dashboards, and private memory. The worker supplies judgment, but the institution owns the customer relationship, the workflow, the records, the distribution, and the compounding benefit.

Agents change the shape of that bargain. They can help a person turn judgment into a smaller operating company: research the niche, draft the offer, find prospects, write the follow-up, prepare the packet, reconcile the record, schedule the next step, remember the promise, and surface exceptions before the customer has to chase.

Call it what it is: clerical load, coordination load, and memory load moving from a person alone into tools the person can direct.

The Owner/Operator Path Is Concrete

Start with a worker who understands a painful office process because they lived inside it.

In a dental practice, they have watched prior authorizations fail for reasons no generic dashboard names. On a contracting team, they have seen margin disappear between estimate and closeout. Grant reports, maintenance requests, insurance paperwork, franchise callbacks, and recruiting loops all have the same hidden layer: the person closest to the work knows where the process breaks and which promise the customer actually cares about.

The agency-multiplier path is narrower than “launch an AI company.” Turn the known pain into a service promise, then use agents to research the buyer, competitor, rule, and workflow. Let the tools draft outreach, follow-up, packets, scripts, checklists, and status updates. Keep operating memory where it belongs: who promised what, which evidence is missing, where the case sits, which exception is open, and what deadline is next.

Paperwork gets less brutal without becoming invisible. Customer promises stay tied to records instead of vibes. Human judgment still owns the call when a case turns on risk, context, trust, or timing.

That is owner/operator work.

The point is not to imitate Meta at smaller scale. Harvesting workers is not the model. Neither is pretending every click is wisdom or building a fake autonomous company that cannot stand behind its promises.

The point is to give a skilled person enough operating reach to sell a real outcome.

Hadto’s Side Of The Line

Hadto takes this side of the line: turn employees and domain experts into people who can govern work, not only perform it.

An owner-ready system has to preserve more than activity. It has to name the customer promise, the record, the rule, the source, the next owner, the exception path, the approval point, and the final state. Otherwise the worker has only a faster assistant. Faster assistants can still leave all the agency with the platform, employer, or buyer.

Ownership requires the work to become inspectable.

For claims cleanup, each case needs source documents, rule notes, reviewer status, customer communication, and a next action. Estimate follow-up needs every lead, appointment, quote, objection, message, and booked job tied together. A back-office rescue service needs the system to remember deadlines, evidence, approvals, exceptions, and promises better than one tired person can.

The displaced intellectual worker should not hear only that the old job is at risk. The risk is real, and it hurts. They should also hear the more useful fact: the same tools lowering the cost of coordination inside the institution can lower the cost of coordination outside it.

None of this makes displacement fair. Institutional extraction remains harmful. Opening a chat box does not guarantee ownership.

It does mean resistance cannot be the whole strategy.

Institutions will use agents to concentrate agency. Workers need tools, records, workflows, and ownership paths that multiply their own agency before every operating advantage is locked inside someone else’s company.

The tools that displaced you can make you an owner, but only if they are attached to judgment, promises, records, and customers you can actually govern.

---

Source evidence used in this note: Reuters, Meta to start capturing employee mouse movements, keystrokes for AI training data, April 21, 2026, for the Model Capability Initiative name, US employee computer tracking, mouse movement, click, keystroke, work-app, occasional screen-snapshot, model-training, safeguard, and non-performance-review claims. Reviewed 2026-05-21. Hadto interpretation: the public labor story is treated as evidence about agency concentration, not as a repeat of the earlier surveillance-trust critique.