|
Tracefold — Reliability testing for AI agents
Tracefold is a reliability testing platform for AI agents: statistical regression runs, semantic trace diffs, and audit-ready evidence packs. Built in Europe.
|
84 % |
|
Laminar - Open-source observability for AI agents
Open-source platform to trace, evaluate, and debug AI agents. Laminar surfaces agent failures, and helps you fix them.
|
82 % |
|
RunAgain: The self-improving loop for AI agents
RunAgain traces every agent run in production, scores what changed, and turns every issue into a test. Works with the Vercel AI SDK, Claude Agent SDK, MCP an...
|
82 % |
|
Instryx — Agent Evaluation & Observability for AI Teams
Instryx evaluates AI agents across 100+ dimensions, traces every step, and pinpoints why an agent failed — so teams catch regressions before customers do.
|
82 % |
|
Find recurring agent failures and turn them into evals · Perseval
Perseval is a local-first observability tool that finds recurring failures across agent traces, shows the exact evidence, and turns verified patterns into re...
|
82 % |
|
AI Agent Regression Testing for Tool-Using Agents | Konsista
Catch AI agent regressions before release. Repeat mocked tool-use scenarios to expose skipped approvals, wrong targets, missing tools, and duplicate actions.
|
81 % |
|
Invarium AI — The behavioral testing layer
Your agents pass evals. They look production-ready. Then they skip steps, hallucinate sources, and ignore guardrails — silently.
|
81 % |
|
faultline — catch the moment your agent silently does the wrong thing
Deterministic silent-failure testing for AI agents. Catch agents that silently drop tool arguments, fabricate tool results, or act on empty tool responses wi...
|
81 % |
|
ReliAgent — know if your agent's last move actually worked
ReliAgent checks every AI agent tool call for hallucination, repetition loops, schema violations, confidence collapse, and four more failure modes — one POST...
|
81 % |
|
TraceDojo — Agent testing
Run controlled failures against your agents. Inspect every action and compare the fix.
|
81 % |
|
Vantage — know what's exploitable, and what was never tested
Ten specialist AI agents test your authorized estate in parallel — web, API, network, Active Directory, cloud, SAP, mobile, LLM apps, source and supply chain...
|
80 % |
|
Autobots — Turn Production AI Agent Failures Into Validated Fixes
Find the first thing that went wrong across your AI agent’s model, context, memory, tools, policies, and runtime. Generate regression tests and replay-valida...
|
80 % |
|
EvalView – pytest for AI Agents | Open Source Agent Testing Framework
EvalView is an open-source testing framework that brings pytest-style workflows to AI agents. Generate tests, track regressions, and keep tool-calling agents...
|
80 % |
|
Chanly
The agent platform that shows you everything. Visual, open, controllable, transparent — built for SMBs and AI startups.
|
80 % |
|
Isolate — Pinpoint which component improved your AI agent
Isolate tells AI developers which part of their agent is working. Run controlled experiments to attribute performance changes to specific components.
|
80 % |
|
Vorlo — Know exactly why your AI agent failed
Vorlo is the debugger for AI agents — it tells you WHY your agent failed and hands you the exact fix, in 30 seconds. Works with LangChain (Python & JS). Ever...
|
80 % |
|
Trace8 — Every Alert Investigated. Not Triaged.
Trace8 is an AI-native SIEM replacement that runs a full investigation on every alert through eight specialized agents — with zero-ingestion federation acros...
|
80 % |
|
ClearViews — Know what your agent's users actually wanted
ClearViews turns your agent's existing conversation logs into a structured behavioral record — what users wanted, where the agent silently fails, and what br...
|
80 % |
|
ClosedLoop AI — Agents that get better every run
ClosedLoop AI closes the feedback loop for AI agents. Every execution becomes training signal: observe, evaluate, improve, redeploy — automatically.
|
80 % |
|
7runs — trust scores for AI agents
7runs measures AI agent reliability: every test case runs 7 times and reports pass rate with Wilson 95% confidence intervals, flakiness detection, double-jud...
|
80 % |
Detected social media networks