Go to Apify Store
User picture

Jeremy Laidi

filnith

Small tools for agent pipelines — each one tells you how far to trust its output.

ACTOR STATS

1 public Actor

2 total users

1 monthly user

>99% runs succeeded

What I build

Small, single-purpose tools for agent pipelines.

Each one returns a result and how reliable that result is. An agent cannot tell good output from bad — it will summarise, cite and act on either. So every tool here ships a verdict alongside the data, and the thresholds behind that verdict, so you can check the call instead of trusting it.

Three rules I hold to

  • A failure is never reported as an empty result. Could not read and nothing there are different facts, and confusing them is how a document silently loses half its content.
  • The thresholds behind every verdict are returned with the response.
  • Known limitations are written on the listing, not discovered by you in production.

Available now

  • Document → text with page, position and a reliability verdict — PDFs to text with the page and bounding box of every line, native/OCR decided per page, and a reliability verdict per page.

Based in the Basque Country, France.

If a tool ever returns a verdict it should not have, that is a bug I want to hear about more than a feature request.

Public Actors