The method behind every build

The audit agent: measurement, not guesswork.

Before I automate anything, a lightweight agent runs for a week on your team's machines and finds where the hours actually go — which workflows repeat, which tools the work bounces between, what it costs. You get a ranked Automation Opportunity Report; I get a build plan. The audit is the method. Building the automation agents is the service.

Sees app and window names, so the report is specific enough to act on · everything stays on your computers · only the final summary comes to me, and only with your approval · your team opts in and sees exactly what's recorded · a stricter no-names mode is available · removed in seconds.

Automation Opportunity Report

1 week

Ranked by recoverable hours — what to automate first.

#1
QuickBooks ↔ Excel data transfer208 switches/wk · 85% automatable · ~192 hrs/yr back
#2
Recurring report assembly4.1 hrs/wk · 75% automatable · ~149 hrs/yr back
#3
Invoice & document processing3.1 hrs/wk · 80% automatable · ~120 hrs/yr back
≈ 461 hours/year recoverable · data stays on your computers · the report is yours
How it works

One week, four steps, zero content captured.

Consent & install

Before anything runs, everyone on your team sees a plain-English list of what's recorded and what never is — and agrees to it. Then I install it with you on a short call, machine by machine. Nothing ever goes on quietly, and removing it takes seconds.

It watches the work, not the people

For one working week it notes which apps are open, for how long, and in what order; what kinds of files move where; and which kinds of websites the browser time goes to. Time away from the desk doesn't count against anyone, and names are replaced with codes. It measures the work — not the people.

It aggregates on the device

The minute-by-minute detail never leaves the computer it was recorded on. The agent adds everything up right there on the machine, and only a short summary is ever shared — through one door, which opens only when a person approves it.

You get the ranked report

The Automation Opportunity Report: your team's real workflows, ranked by recoverable hours and dollars, with the evidence behind each. That ranked list is the build plan for the actual service — the automation agents.

The privacy promise

Enforced in code, not promised in prose.

The privacy promise isn't a policy page — it's built into the software itself. The instant anything is recorded, a filter throws away everything except the short approved list, and if anything off-limits ever shows up, the agent shuts itself down rather than save it. Nothing leaves the machine without a person saying yes. It's the same standard I build to at a major US bank.

Collects — stays on the device

· App + the real window/tab title + activity category
· Active duration & idle time
· Process name + start/stop (never command-line arguments)
· File action + name + type + size bucket + location category (never the full path)
· In-browser SaaS category + the real tab title

Real titles are what make the report specific enough to act on. Want stricter? One setting switches to a no-names mode (metadata-only) that drops every title and keeps just categories.

Never collects (any mode)

· Keystrokes, screenshots, clipboard
· File or message content
· Passwords, secrets, tokens
· Full URLs (query strings) or hostnames
· Full file paths
· Command-line arguments
· Personal identity — people appear only as codes, never names

Storage & controls

Everything — window names included — stays on each computer and is summarized right there. Only the final report is shared, and only after a person approves it. Your team opts in · everyone sees exactly what's recorded · data deletes itself after 30 days (sooner if you ask) · removed in seconds · and the purpose is fixed: find work worth automating — never to monitor or score employees.

Onboarding

Install & enroll — one command per machine.

I install it with you on a short call, and removing it is just as quick. There's nothing hidden inside: the agent is plain, readable code, so your IT person — or any technical friend you trust — can look through every line of what it does before it runs.

macOS

curl -fsSL https://nulldrift.tech/install.sh | bash

Shows the manifest, asks for consent, installs a per-user background service, and triggers the one-time "Automation" permission click that lets it read Safari/Chrome tab titles (kept on the device, never the URL). Requires Python 3.10+.

Windows

irm https://nulldrift.tech/install.ps1 | iex

Shows the manifest, asks for consent, registers a least-privilege Scheduled Task, and sets up the optional browser extension (loaded unpacked — it categorizes in-page and only ever hands over a category). Requires Python 3.10+.

Where the data goes

By default: nowhere. The agent writes its daily aggregate to a local datastore on the machine. For multi-machine pilots, devices enroll with a per-client key and upload summaries to a private, client-scoped bucket — encrypted, TLS-only, auto-expiring, and revocable per device. Uploads stay off until you and I turn them on together.

Installs are issued per engagement — you'll use the install code from your NullDrift audit email.

Or for IT folks — one line, either OS. The installer downloads a checksum-verified bundle from this domain, shows you the full privacy manifest, and asks for consent before anything runs. Uninstall is one command. Engagements are still white-glove — start with the free audit →

Ready to see your real number?

One week of measurement. A ranked report in hours and dollars. Then I automate the top of the list.