SKTCH: Preparing startup assessments so reviewers have more time to judge
Reading applications, cross-checking claims, and applying internal screening criteria before a human makes the decision.
The work behind the project.
For each new program, the SKTCH team had to read hundreds of applications and external materials. Coxwave worked on automating preliminary review and fact-checking so the team could devote more attention to the selection decision itself.
Application materials arrived in different document formats. Much of the limited review period was spent finding and reading information before a reviewer could assess the opportunity.
Because this information supported startup selection, an AI-generated summary alone was not enough. The team needed claims to be checked against external information and assessments to reflect its own review standards.
From knowledge to execution.
- 01
Extract the key information
Read the different kinds of submitted documents and identify the information needed for preliminary assessment.
- 02
Cross-check the claims
Compare submitted information with external data to check whether the claims are supported.
- 03
Apply internal review criteria
Reproduce the organization’s assessment standards in the preliminary review. The selection decision remains with the team.
Clear boundaries. Informed decisions.
Submitted claims are cross-checked against external information, and preliminary assessments follow the organization’s own criteria. The Engineer prepares the review; the team retains the selection decision.
Project scope and outcome.
The intended division of work was explicit: the Engineer handles preliminary examination and fact-checking, while staff make the selection decision. The project combined document understanding, external verification, and the organization’s assessment criteria.
