Snapshot
Continue to About ↓- What it does
- LangWatch is an LLM operations platform for engineering teams building and running AI agents, combining tracing and observability, simulation-based agent testing, automated evaluation and a model gateway with budget controls.
More detail
- It is sold as managed cloud or self-hosted software, metered by billable events and core seats.
- Best for
- AI platform and engineering teams with agents already in or nearing production that need regression testing, automated scoring and production trace visibility across Python, TypeScript or OpenTelemetry instrumentation.
More detail
- Also fits data-sovereignty-sensitive organisations that need self-hosted, VPC or air-gapped deployment.
- Pricing
- Metered SaaS plus seats, not a flat subscription.
More detail
- Developer tier is €0 with 50k events/month, 14-day data access, 2 users and limits of 3 scenarios, 3 simulations and 3 custom evals.
- Growth is €29 per core seat per month with 200k events included, then €5 per 100k events, and 30-day retention includ
- Main limitation
- Teams wanting a no-code, fully managed tool with no engineering effort, and buyers without someone to own SDK/CLI instrumentation or, for self-hosting, Kubernetes operations.
More detail
- Not a fit for teams that need confirmed SOC 2 certification, since only ISO 27001 and GDPR are asserted.
- Record
- Living and monitored
- Reviewed
- Evidence
- Public web research
- Machine access
- Structured profile ↗
About
- LangWatch is an LLM operations platform for engineering teams building and running AI agents, combining tracing and observability, simulation-based agent testing, automated evaluation and a model gateway with budget controls.
- It is sold as managed cloud or self-hosted software, metered by billable events and core seats.
What it helps with
- Agents that work in demos break in production, and manual testing only covers a fraction of the paths an agent can take.
- Teams lack end-to-end visibility into each agent step, token spend and latency in live traffic.
- Response quality and policy adherence are scored inconsistently or not at all across releases.
- Safety and security gaps such as jailbreaks, policy breaks and unsafe tool calls surface only after users find them.
- Model calls across providers are hard to route, budget and attribute per person, team or customer.
- Prompt changes ship without versioning, A/B testing or review history.
- Spend by engineering coding agents such as Claude Code and Codex is untracked per engineer and repository.
Capabilities
A practical view of what is available and where limitations apply.
| Offering | Available | Detail |
|---|---|---|
| OpenTelemetry-native tracing | Yes | Python, TypeScript and Go SDKs with GenAI semantic conventions; framework-agnostic via OTel. |
| Simulated-user agent testing | Yes | Multi-turn text and voice simulations with judge scoring and pass/fail at any turn. |
| Automated evaluation | Yes | LLM-as-judge, code and workflow scoring, offline and online, plus built-in RAGAS, hallucination, toxicity and PII evaluators. |
| Red-teaming | Yes | Adversarial simulations including Crescendo escalation, refusal detection and backtracking. |
| AI Gateway | Yes | One endpoint for multiple model providers using the customer's own keys, with budgets per person, team and customer. |
| Prompt management | Yes | Versioning, A/B testing and GitHub sync. |
| Self-hosted deployment | Yes | Helm, Docker Compose, in-VPC and air-gapped; runs the same software as cloud. |
| Coding-agent cost tracking | Yes | Tokens, cost and cache per engineer and repository for Claude Code, Codex, Copilot and opencode. |
| SOC 2 certification | Unknown | ISO 27001 and GDPR are claimed; no first-party page asserts a completed SOC 2 certification. |
| Independent third-party reviews | Unknown | No credible independent review or press coverage found in the evidence pack. |
Bot trust
How confidently AI systems can verify this profile from public evidence.
—of 100
ChatGPT—/100
Claude—/100
This profile has not completed a Bot Score assessment yet.
Evidence
Where this record came from, when it was reviewed, and how confidently it can be used.
- Verification
- Scribble researched these facts from the linked public sources.
- The owner has not manually confirmed every field.
- Last reviewed
- Primary source
- langwatch.ai
Technical profile details
Record ID
05f406b7-a28a-4e9a-9f45-efb37ab9d9f4Canonical URLhttps://profiles.scribble.network/listing/langwatch-05f406b7