Looking Glass Evals

1 min read Original article ↗

NewEvals dashboard + Chrome extension

10x your BPO throughput and cut operating costs with AI-assisted data extraction.

Scale Ops

Automate repetitive data entry

Free operators from copy-paste work. Focus judgment on edge cases where it matters.

Built-in Evals

Accuracy you can trust

Real-time correctness metrics, trends, and failing fields keep quality auditable.

Drop-in

Works with your forms

Scan any form in the browser. Bring your own data source via a tiny endpoint.

How it works

1) Scan

Extract the form schema

Operators click Scan in the Chrome extension. Fields and types are captured from the page.

2) Retrieve

Fetch authoritative answers

Backend returns ground truth for the requested entity and schema via a simple endpoint.

3) Fill & Verify

Human-in-the-loop

Suggested answers populate the page. Operators confirm, correct, or mark unknown—then submit evals.

Tailored for your team

Operators

Speed and accuracy

Review model suggestions side-by-side with source context. Submit judgments in a click.

Engineers

Integrate in minutes

Expose a small endpoint that returns ground truth. Get instant evals and feedback loops.

Ready to see accuracy in action?

Open the dashboard and explore recent evals.

Open Evals