Professional-domain evaluation, managed as evidence.
betaPlan defines the professional scope, prepares and assesses capacity where a qualification route exists, manages assigned work, separates independent review, and preserves a record a buyer can inspect.
Four kinds of work
Professional-domain AI work can depend on safety, regulation, source quality and judgement that a generic label does not capture. The work therefore begins with a task definition and the professional competence it requires.
Capacity would be commissioned for that scope only after it is supportable. Completion of learning is not qualification, qualification does not guarantee assignment, and no person is assigned merely because they joined a course. Where qualification is used, a human decision and its rule version are recorded.
Labelling and structuring
Specialist material is structured against a defined schema, access boundary and evidence standard. The required professional scope is agreed before capacity is commissioned.
Output evaluation
A scoped reviewer evaluates an AI output against observable criteria and records the reason, evidence and escalation rather than returning an unexplained label.
Domain red-teaming
Tests target failures that matter in the domain, including fabricated sources, unsafe recommendations, privacy breaches and missed escalation. Regulated claims require the relevant professional review.
Agent workflow review
A step-by-step review identifies evidence gaps, unsafe hand-offs and points that require human judgement. It does not declare automation safe without the required domain evidence.
What a procurement team asks now
- Who did the work, and were they paid properly?
- The project contract defines worker category, access, compensation and task records. Personal identity is disclosed only under the agreed lawful arrangement; buyer evidence can use pseudonymous identifiers.
- How do you know they were competent?
- The required competence, assessment route and professional ratification state are declared for the project. Learning completion alone is never evidence of qualification, and critical failures cannot be averaged away.
- What jurisdiction is my data in?
- Jurisdiction, hosting, subprocessors, retention and access controls are confirmed in the project agreement. This public product description is not a substitute for that data-processing review.
- Can I check any of that without trusting you?
- Yes. In the delivery model, an approved delivery carries a sealed record you verify offline against a published key. The public checker demonstrates this on a synthetic record, not on a client delivery.
Request a sample
Describe the professional domain, decision risk, data constraints and evidence you need. We can then determine whether an evidence-backed pilot is supportable and what professional ratification or capacity must come first.