LOYour roles, not the occupation
Everything on this site describes occupations in general. A company is not a list of occupations — it is a specific roster, in a specific market, doing a specific mix of tasks. Mapping that roster against this evidence, task by task, is work, and it is the only thing here that costs money.
What stays free#
All of it. There is no task judgement, no evidence record and no source link behind a payment, and there will not be.
160 occupations broken into tasks, 968 verified records with their sources, the majors, the recent-changes timeline, the method page, and both ways of asking — typing and calling. If a paywall ever appears over a task judgement, this paragraph is the thing to hold us to.
A worked example#
A made-up company — 22 people, B2B software. The company is hypothetical; every number below is computed from the same task judgements and records the rest of this site runs on. You can run the same table on your own roles, free, on the business page.
| Role | Heads | Change method | Stays human | New duties | Evidence behind it |
|---|---|---|---|---|---|
| Sales / account manager | 4 | 2 | 1 | 1 | 4/6 tasks · 4 records |
| Customer service representative | 3 | 2 | 2 | 1 | 3/5 tasks · 5 records |
| Marketing specialist | 2 | 2 | 1 | 1 | 3/5 tasks · 3 records |
| Junior software developer | 6 | 1 | 2 | 1 | 4/5 tasks · 6 records |
| Product / UX designer | 2 | 1 | 2 | 1 | 5/5 tasks · 7 records |
| Accountant / Bookkeeper | 1 | 2 | 2 | 1 | 4/6 tasks · 7 records |
| HR / recruiter | 1 | 2 | 2 | 1 | 3/5 tasks · 5 records |
| Operations coordinator | 2 | 3 | 1 | 1 | 6/6 tasks · 5 records |
| Partnerships / channel manager | 1 | 2 | 2 | 1 | 4/6 tasks · 4 records |
Across this roster the evidence points at 9 duties that automation created and that nobody is assigned to by default — usually the most actionable part, because it is work appearing rather than work leaving.
Run this on your own roles — free · same data, same columns, including the empty ones
What is in the report#
- Your roster resolved to occupations — including the ones that do not resolve cleanly, which is itself a finding about how your company names work.
- Per role: which tasks should change method, which stay with a person, and which duties automation created. Each with its basis — verified record or platform inference — and what the judgement does not establish.
- The blind spots, named. Roles where we have no evidence, and roles where the evidence is one record from a source with a stake.
- One thing per role you can check inside your own company this week, without buying anything.
It is written against the same rules as the site, and the sources are all links you can open yourself. If you disagree with a judgement, you can go to the record and see what it rests on.
What it will not do#
These are the same refusals as the rest of the site, and they do not get relaxed because someone is paying:
- It will not tell you who to make redundant, and it will not rank your people. It is about tasks, and tasks are not headcount.
- It will not give you a readiness score, a risk percentage, or any single number for your company. A number is where people stop reading.
- It will not tell you what your product, your org chart or your business model should become. That needs per-company workflow and cost data we do not have, and we will not guess it for a fee.
- It will not recommend a vendor. Nobody pays to be in it, and nobody pays to be left out of it.
How to start#
This is new and it is not a checkout yet: there is no payment page, and the first few are being done by conversation so they get done properly before anything is priced. Send your role list — titles and rough headcount is enough, no names — and you will get back whether there is enough evidence to make a report worth buying. Sometimes the honest answer is no.
Analysis, not advice — the same terms as the rest of the site apply, and decisions about your company remain yours. Who makes this and who pays for it
Looking for something for one person rather than a whole company? Your first AI experiment at work