You find out your agent failed when a customer screenshots it. Glassray tells you first.

Agents don't throw errors. Most failures are silent. Glassray monitors your agent traces to find where they broke down and gives you the fix.

Join the alphaSee an example
trusted by teams at
Case studySuper44 replaced the agent monitoring it built in-house with Glassray and increased agent reliability 2.2x2.2xagent reliability55custom failure patterns736deviation investigations“The statefulness of Glassray is the biggest win: it reduces the noise and helps us figure out what's important over time.”Henry George, Co-Founder at Super44Henry George, Co-Founder at Super44Case studyArchitect caught 15x more agent failures and cut investigation time by 19 hours a month with Glassray15xmore failures caught5xmore fixes verified~19hsaved per month“The quality of the answers we give is an extension to our business. It's absolutely essential and core.”Luke Ramsden, CPTO at ArchitectLuke Ramsden, CPTO at Architecttrumpet& more
currently monitoring
147,589traces / month
150failure modes
the problem

Agents don't throw errors.
They just act.

Engineers are shipping value to users fast, but evals are very hard and painful to build and maintain. Teams don't have a way to know what issues their agent has before a customer catches them unless they spend a lot of time looking at traces.

01

Building evals takes weeks. Most become stale in weeks.

A proper eval set takes weeks of engineering time, goes stale within a sprint, and still only covers what you thought to test. So most teams ship on instinct and hope.

observationWe had ground truths for the happy path. But this week we've changed the schema and need to rebuild them.
02

Hundreds of wrong answers. Zero alerts.

Agents don't throw exceptions. They just act, confidently and incorrectly, across hundreds of conversations before anyone notices.

observationThe agent had been issuing wrong refunds for two days. An engineer caught it while debugging something else.
03

Every change ships into uncertainty.

Agents are probabilistic. When you fix one thing, you have no way of knowing if something else quietly broke. There is no coverage to run against, no signal until a customer tells you.

observationPushed a fix Monday. Found out Thursday it broke the other flow. No way to catch it earlier.
customer example · Super44

The briefing read perfectly.
The forecast was invented.

Super44 briefs local merchants on their own business every morning: an agent running up to 20 tools compiles sales, staffing, reviews, and the weather that feeds the revenue estimate. One morning the weather tool returned empty, and the agent wrote the briefing anyway, reusing yesterday's forecast as if it were confirmed.

No error, no complaint, nothing for an eval to fail. Glassray compared the reply against the tool outputs in the trace, found a forecast with no data behind it, and flagged the run with the evidence attached. The pattern stays monitored, so a made-up number never gets to become a habit the merchant plans around.

“Do merchants trust that when the agent tells them something about their data, it's true? If it says ‘you've sold more Aperol Spritzes’ and they think, I don't believe you, our value diminishes.”
Henry George
Henry GeorgeCo-Founder at Super44
Read the case study
Super44app.super44.ai
Tuesday briefing

Good morning! Today's sunny and mild, so expect the terrace to fill up. Projected revenue: €18,400.

Ask anything about your business...
What the merchant sees
Trace viewmorning briefing
  • ok
    Point-of-sale lookupsales data returned
  • ok
    Reviews + staffingreturned normally
  • problem
    Weather toolreturned empty, no data for today
  • problem
    Composerwrote briefing using yesterday's forecast
  • ok
    Final output‘today’s sunny and mild’ · fluent, confident
What Glassray sees
More examples
The solution

Install Glassray.
Every trace improves your agent, forever.

Glassray wraps your agent to watch production behavior, evaluates it against your code's intent, proposes code-level fixes, and validates them before you ship.

~/project
glassray · sources
$ npm install @glassray/sdk
added 1 package in 0.8s
reading agent intent from codebase
// agents/support.ts
import { glassray } from '@glassray/sdk'
const agent = glassray.wrap(myAgent)
LangSmith
connected
LangfuseLangfuse
connected
OpenTelemetryOpenTelemetry
connected
4,821 traces ingested
01

Install

Connect Glassray to your traces. It reads the code to learn what each agent should do.

02

Judge every step

An AI judge evaluates every reasoning step in a production trace, not just the final output, and flags silent failures and intent mismatches against your code.

glassray.ai / org_atlas/traces/trace_8a2f1cjudge · running
support-agent · run #418claude-sonnet-4-6 · 5 steps
01user_message
intent understood
02llm_reason
reasoning valid
03tool_call
correct tool
04tool_result
data retrieved
05respondhallucination
4 ✓ · 1 ✗ · T+11.4sreplay
glassray · judge logT+11.4s
step 01 user_messagepass

intent classified · refund · order #4421 · confidence 0.94

step 02 llm_reasonpass

chose to retrieve policy before responding · matches code intent

step 03 tool_callpass

get_refund_policy("4421") · correct tool · args valid

step 04 tool_resultpass

retrieved · 0 docs (policy unset for this order type)

step 05 respondsilent failure

policy fabricated · cited a 30-day window not in retrieved docs

proposed fix · agents/support/refund.ts:42
+ if(!docs.length) escalate()

// guard the empty-docs path before respond()

replayed1,247 tracesaccuracy87.4% → 94.2%delta+6.8% ▲
judge · every step · not just the final output
agents/support/respond.ts+3 −1
28 async function handleMessage(ctx: Context) {
29 const query = ctx.message.text;
30 const prompt = buildPrompt(ctx);
31− return respond(await llm.generate(prompt));
31+ const docs = await kb.search(query);
32+ if (!docs.length) return escalate();
33+ return respond(await llm.generate(prompt, docs));
34 }
glassray proposes·validated on 1,247 traces
03

Propose the fix

On every deviation, Glassray proactively suggests a code-level change. It lands in your repo as a diff with the failing trace attached.

04

Validate, then ship

Glassray replays the proposed fix in a sandbox against your trace history and surfaces an accuracy delta before merge.

fix/support-agent-escalation+3 −1
glassraycommented just nowsafe to merge

Replayed fix against 1,247 production traces. Accuracy improved from 87.4% to 94.2% with 0 regressions detected.

+6.8% accuracy·0 regressions·1,247 traces replayed
See more:Trace breakdown →Replay report →Diff →
validate · then ship · replay delta posted to the pr
Example

See the silent failure.
Ship the validated fix.

A support agent was asked why users were being logged out across devices. The knowledge base had no document covering it. Instead of saying it didn't know, the agent fabricated a policy. Standard evals scored it highly. Glassray flagged it.

trace_8a2f1c
1 issue
TraceAI fabricates logout policy user cancels subscription
1user_message0.1s“I keep getting logged out whenever I switch between my laptop and desktop.”
2llm_reason0.8s
3tool_call0.3s
4tool_result
5tool_call1.2s
6tool_resultretrieval miss{ results: [], matched: 0 }
7llm_reasoninference gap1.1s
8respondhallucinationThe logout behavior is expected – our policy limits sessions to one device.
9log_event0.1s
10complete
Glassray analysishigh

search_knowledge_base returned 0 results. No policy document covers multi-device sessions. Instead of escalating, the agent inferred from similar SaaS tools and responded with a fabricated one-device limit which does not exist. The user cancelled their subscription.

// step 8 · confidence 0.97 · 23 similar traces · 4 led to churn
agents/support/respond.ts+3 −1
31−return respond(await llm.generate(prompt));
31+const docs = await kb.search(query);
32+if (!docs.length) return escalate();
33+return respond(await llm.generate(prompt, docs));
replay sandbox1,247 traces
accuracy87.4%94.2%
delta+6.8%
regressions0
Ship fix & close loop
“
The statefulness of Glassray is the biggest win: it helps developer experience, reduces the noise, and helps us figure out what's important over time.
Henry George
Henry GeorgeCo-Founder at Super44
Read the case study
pricing

Observability is generously free.
Pay only when we catch mistakes your agent made.

We follow a "price should be a no-brainer" strategy. If it's not, tell us why.

Traces & observability
$0free tier

Every trace, in one place. Plug in your stack and see what your agents are doing.

  • SDK (native), OpenTelemetry, Langfuse, LangSmith, PostHog, Braintrust & these integrations
  • 150k traces / month, then $1 per 10k
  • 1-year retention
  • 1 project
  • Trace explorer & search
  • MCP, API & CLI access
  • Unlimited seats
  • A better UI than the others
Agent monitoring
$499/ month

Glassray watches production for you and flags the runs that quietly went wrong.

  • Everything in free
  • Silent-failure detection on every trace
  • Built-in failure modes + custom signals
  • 100 active confirmed deviations, then $5 each
  • Suggested code-level fixes
  • Destinations: push deviations to Slack / webhooks
  • 5 projects
Join the alpha
Enterprise
Custom

A custom solution for teams running agents at scale.

  • Everything in Agent monitoring
  • Model post-training on your traces
  • Custom RL environments
  • SSO/SAML login
  • Audit logs
  • Dedicated customer success manager
Talk to a founder
pricing calculatorEstimate your cost
Why Glassray

Other tools log agent activity
for humans to review. Glassray fixes it.

Observability platforms, eval frameworks, LLM-as-judge models: every existing tool gives engineers better dashboards for reading traces. Glassray replaces the human in the loop. It judges every step, proposes the fix, and validates it before you ship.

capability
observability + evals
llm-as-judge
glassray
Trace ingestionCapture every step
yes
partial
yes
Step-level intent judgmentCatch silent failures inside the trace
no
yes
native
Code-level fix proposalsProactively suggests a code-level change
no
no
yes
Works without ground truth labelsNo labeling pipeline needed
no
partial
yes
Framework agnosticWorks across every stack
partial
partial
yes
Product-level contextLooks beyond the agent at the full user session
no
no
yes
No new pipeline

Glassray plugs into the stack you already have.

LLM SDKs
OpenAI SDKOpenAI SDK
InstructorInstructor
Anthropic SDKAnthropic SDK
Azure OpenAIAzure OpenAI
Google GenAIGoogle GenAI
Vertex AIVertex AI
AWS BedrockAWS Bedrock
CohereCohere
LiteLLMLiteLLM
GroqGroq
MistralMistral
OllamaOllama
WatsonxWatsonx
Together AITogether AI
Aleph AlphaAleph Alpha
HuggingFaceHuggingFace
ReplicateReplicate
SageMakerSageMaker
Ruby LLMRuby LLM
Frameworks & Observability
OpenAI AgentsOpenAI Agents
Claude Agent SDKClaude Agent SDK
LangChainLangChain
LangGraphLangGraph
LangflowLangflow
LlamaIndexLlamaIndex
CrewAICrewAI
AutoGenAutoGen
DSPyDSPy
Pydantic AIPydantic AI
Google ADKGoogle ADK
SmolagentsSmolagents
Strands AgentsStrands Agents
GuardrailsGuardrails
HaystackHaystack
AgnoAgno
MCPMCP
BeeAIBeeAI
PipecatPipecat
MastraMastra
SuperagentSuperagent
BAMLBAML
Vercel AIVercel AI
LangfuseLangfuse
HeliconeHelicone
Arize PhoenixArize Phoenix
BraintrustBraintrust
OpenTelemetryOpenTelemetry
Alpha

Make your agents
self-improve.

Limited alpha spots. Join now and we'll reach out when your stack is ready to plug in.

Join the alphaTalk to a founder