Skip to content
Over Unity

How this works

Most of it runs without us. Four things never will.

Intake, scoring, ranking, introductions, retention and backups all run on their own. That is not a boast, it is the only way a small operation can assess people properly: every hour a person spends on something a machine could do is an hour not spent checking whether somebody is any good.

The four judgements below stay with a person, and the software refuses to proceed without them. Those refusals are the product.

What we will not automate

Deciding whether a piece of evidence is real

The failure mode is not that a machine reads a reference badly. It is that anything a machine can check by pattern is something a machine can produce, and the register would fill with people who understood that before we did.

Assessing somebody a level

A level is a judgement about what somebody would be like to work with on a hard week. There is no artefact that contains that.

Writing the reason for a decline

The moment a decline reason is generated, it stops being a reason and becomes a paragraph. People can tell, and the ones who can tell are the ones we want.

Reading a shortlist before a hirer does

The scorer knows what it was told. Somebody has to check whether the brief left something out, which is the normal case rather than the exception.

If you are hiring

  1. 01

    You describe the problem

    Automated

    A short form. The problem, the disciplines you think you need, the shape of engagement, the budget. Ten minutes, not a discovery call.

    A first conversation whose only purpose is to collect facts a form could collect is a tax on your time that we charge you for later.

  2. 02

    The register is scored against it

    Automated

    Every admitted specialist is scored against the brief. Disciplines are weighted rather than tagged, so a shortlist can be ranked rather than filtered.

    Deterministic and reproducible. The same brief and the same register produce the same ranking, and the signals behind each score are stored with it so a shortlist can be explained months later.

    Enforced Unknown is a value, not a zero. A specialist who has not been assessed on a discipline scores as unknown rather than as bad, and the shortlist says so.

  3. 03

    A person reads the shortlist before you do

    A person

    Somebody checks the ranking against the brief and writes the reasoning you receive.

    The scorer is good at ordering candidates and blind to context it was not given. This is the step where somebody notices that your regulated-industry brief needs the person who has been through an audit, not the one with the higher score.

  4. 04

    You are introduced

    Both

    Direct introduction to the people you want to speak to. We are not in the middle of the conversation after that.

    An agency that relays messages is charging you for latency. Once you have met, you talk to them.

    Enforced An introduction cannot exist without a match behind it, and the lifecycle refuses transitions that make no sense.

  5. 05

    You engage them directly

    A person

    The contract is between you and the specialist. We do not hold or transfer payments and do not determine employment status.

    Sitting in the payment chain would make us a party to every dispute and would change what we are for. It also creates the incentive to keep an engagement running past the point it is useful.

If you are applying

  1. 01

    You apply with evidence

    Automated

    Up to six disciplines, at least one piece of evidence, a self-rating on each. Every problem with the form is reported at once rather than one at a time.

    Six is the most anybody can be assessed on properly. A form that reveals its objections one at a time is punishing you for its own design.

    Enforced An application without evidence is not assessed. Six disciplines is a hard limit.

  2. 02

    The application is triaged

    Automated

    Completeness, discipline mapping, and whether the evidence offered is the kind that can actually be checked.

    Mechanical. It decides what a reviewer looks at first, and it never decides an outcome.

  3. 03

    Evidence is checked

    A person

    Somebody looks at what you sent and records whether it was confirmed, could not be confirmed, or was refuted.

    This is the one that cannot be automated, and it is the only reason the register is worth anything. Anything that can be verified by pattern-matching a document can be faked by generating one.

    Enforced Nobody is admitted without at least one piece of verified evidence. The admit action refuses.

  4. 04

    A level is assessed and recorded separately

    A person

    A reviewer records their own view of your level against the evidence. Your self-rating is kept, unchanged, beside it.

    Two numbers that disagree are more informative than one number that has been reconciled. Where only your self-rating exists, a hirer sees it labelled as a claim, every time.

    Enforced No admission without an assessed level.

  5. 05

    You are admitted or declined, with a reason

    A person

    Either way you are told why, in writing, against the rubric version you were assessed under.

    A form rejection teaches you nothing and costs us nothing, which is exactly why everybody sends them.

    Enforced The software refuses a decline without a written reason of at least a sentence, and stamps the rubric version on every decision.

  6. 06

    It is looked at again in 12 months

    Both

    Admission carries a re-verification date. Evidence goes stale and people change what they work on.

    A register that never re-checks is a snapshot pretending to be current.

Where the time actually goes

StageCharged againstNote
Checking evidence and assessing a levelPer applicantThe bulk of it, and deliberately so. This is what is being bought.
Reading a shortlist and writing the reasoningPer briefFalls as the register grows, because the scoring gets more to work with.
Everything elseNear zeroIntake, triage, scoring, ranking, introductions, retention, backups. If a stage does not need judgement, a person should not be in it.

What we are not

  • Not a marketplace. There is no bidding, no proposals and no ratings. Ratings measure whether past clients were happy, which is a different thing from competence, and that difference is the whole proposition.
  • Not in the payment chain. We do not hold or transfer money and do not determine employment status. Sitting in the middle would give us an interest in keeping an engagement running past the point it is useful.
  • Not charging specialists. Not to join, not to be shortlisted, not to be introduced, not ever. Charging a work-seeker to find them work is prohibited under UK recruitment rules, and there is no column anywhere in our database for a practitioner fee. That is enforced by the schema, not by a sentence in the terms.
  • Not pretending the register is full. It is being assembled. The roadmap says exactly where it has got to, including what is blocked and on whom.
Describe a roleApply to the registerWhere this has got to

Over Unity makes introductions between hirers and independent specialists. It is not a party to any engagement, does not hold or transfer payments, and does not determine employment status. Specialists are never charged a fee.