01Quality pipeline
Four stages between your scope and your result
How an engagement is designed to run: from the scope you agree with us to the result you receive, with expert work and independent review in between.
Stage 1 of 4 · Intake
Scope the data need
- The domain and the kind of training or evaluation data
- The task types and questions in scope
- What a finished delivery looks like
Output of this stage
An agreed scope
02How an engagement works
Two sides of every engagement
How an engagement is designed to work: buyers scope the need and receive the result, vetted experts do the work, and Pensator runs everything in between.
Phases of an engagement
1 Scope
2 Match
3 Work & review
4 Deliver
For buyers
AI labs, data vendors, AI products- 1
Tell us the need
Share the training or evaluation data you need.
- 2
Agree the scope
The domain, the task types, and what a finished delivery looks like.
- 3
Nothing to manage
Experts do the work and it is reviewed — you don’t run a workforce.
- 4
Receive the result
Finished data or evaluation: answer sets, evaluations or test sets.
For experts
Vetted domain experts- 1
Vetted for the domain
Experts are vetted for the domain — for Indian tax, Chartered Accountants.
- 2
Matched to scoped work
Matched to the work the buyer has agreed with us.
- 3
Do the expert work
Write and check answers, evaluate model outputs, explain each error.
- 4
Work is reviewed
Quality-checked independently before anything is delivered.
AI labs, data vendors, AI products
- 1
Scope
Tell us the need
Share the training or evaluation data you need.
- 2
Match
Agree the scope
The domain, the task types, and what a finished delivery looks like.
- 3
Work & review
Nothing to manage
Experts do the work and it is reviewed — you don’t run a workforce.
- 4
Deliver
Receive the result
Finished data or evaluation: answer sets, evaluations or test sets.