Begin with a manual baseline
Use a small set of images you own or have permission to process. Record the expected sources and the cases that should return nothing. A manual review tells you which outputs your application must preserve before you automate requests.
Evaluate the integration contract
- Check allowed use, retention and deletion requirements.
- Understand job status, rate limits and error responses.
- Preserve source URLs alongside scores.
- Budget for retries and human review, not only successful requests.
For a photographer checking ten portfolio images once, a manual workflow may be easier to audit. For a rights team reviewing a permissioned catalogue regularly, an API may reduce repetitive input work while leaving the judgement with reviewers.
Keep test mode explicit
Facehunch's example connection uses demo mode and a restricted test index, separately from the main workspace search. A successful upload-and-polling sequence verifies the integration flow, not production coverage. Do not extrapolate quality or costs from a demonstration without reviewing the provider's current documentation and conducting an authorised evaluation.
Continue with tool selection worksheet.
Written by Facehunch with AI assistance and checked against the cited sources. Our editorial policy.