Apache 2.0 · local-first · BYOK judge

LLM observability for production apps and agents.

Cognifity builds observability tools for AI in production. Verdict — our first product — works with existing Verdict-compatible traces or captures supported LLM calls, then helps teams measure quality drift with calibrated PASS/FAIL rubrics.

Verdict v0 focuses on the LLM-call layer. Agent-run graphs and task-success metrics are on the roadmap.
Apache 2.0
Open-source SDK, evaluation, inspection, and dashboard tools
3 providers
Anthropic, OpenAI, and Google capture paths
BYOK
Judge scoring uses your provider key
Local
SQLite by default, Postgres when needed
Meet Verdict

Statistical drift detection for the LLM stack you actually run.

Most LLM observability starts with latency, tokens, and errors. Verdict adds quality monitoring: start with existing compatible traces or capture supported provider calls, group similar prompts, evaluate sampled responses with a rubric, and compare recent behavior against a reference window.

Use your traces—or capture them

Start with an existing Verdict-compatible SQLite or PostgreSQL store. If you need capture, verdict.init() wraps supported Anthropic, OpenAI, and Google SDK calls without putting a proxy in front of your traffic.

Statistical rigor, not guesswork

Fisher's exact for binary PASS/FAIL dimensions, Mann-Whitney U for continuous scores, Cliff's δ for effect-size gating, and Benjamini-Hochberg correction across dimensions.

Honest calibration workflow

Calibration scripts compare judge decisions against your own human labels. Public benchmarks are treated as sanity checks, not proof that a judge is calibrated for your workload.

See the full Verdict methodology
Why Cognifity

Engineered for teams that need evidence, not guesswork.

Production AI teams already have logs, latency charts, and token dashboards. Cognifity products focus on the harder operational question: did behavior change for the workloads your users actually run?

01

Statistical rigor by default

Fisher's exact for binary PASS/FAIL dimensions, Mann-Whitney U for continuous scores, Cliff's δ effect-size gating, and Benjamini-Hochberg correction across dimensions.

02

Honest about what works

The repo includes workflows for judge calibration, live capture checks, and injected drift tests. We say what v0 catches today and keep agent-run outcomes in the roadmap until they ship.

03

Open source, no lock-in

Verdict is Apache 2.0. Run it locally, inspect the code, keep your traces in your own storage, and bring your own provider key for judge-based scoring.

04

Provider adapters, not a proxy

Storage adapters for SQLite/Postgres/in-memory and provider adapters for Anthropic, OpenAI, and Google keep evaluation logic separate from provider SDKs.

What it looks like in production

Common drift modes that show up too late.

Silent model update

A provider, model, or prompt change shifts refusal behavior on support traffic.

Verdict captures the calls, tracks structural signals, and helps you compare recent windows against a baseline so a reviewer can inspect affected traces quickly.

Signal: refusal-rate and quality drift by workload
Pre-flight model swap

Your team wants to compare models without trusting a global average.

Use pairwise comparison and workload grouping to inspect where a candidate model helps, hurts, or needs more labels before you switch production traffic.

Risk: blind swap → workload-specific evidence
Cost runaway

A prompt change alters response length and token usage.

Captured traces include tokens, latency, estimated cost, finish reason, errors, and optional best-effort-sanitized content, so cost and quality investigations start from the same evidence.

Signal: token, latency, and response-length movement

See Verdict before you install it.

Watch the 72-second walkthrough, then install the latest compatible public-alpha packages directly from PyPI. If you need capture, add the maintained agent skill after the packages. No source checkout is required.