AI SRE
Investigate production issues, find the cause and open a pull request.
See the AI SRE
Know what happened.
Fix it. Prove it.
Observability data lake and an AI SRE agent, to catch issues and continuously improve production.
Free up to 10k events a day. No credit card.
-
The alert lands in Slack with the failing request
-
Request and response bodies, next to the trace
-
The AI SRE finds the cause and opens the fix
-
What the customer saw
monoscope watches, fixes and improves production,
so your engineers can focus on building.
Open source
AGPL 3.0
AGPL 3.0
20x
faster MTTR at Blockradar
faster MTTR at Blockradar
780+
OpenTelemetry integrations
OpenTelemetry integrations
“Easy onboarding and they added an integration just for our use case, thanks again! We didn’t have insights into our API load before and this helps very much.”
Sebastian Zwach, Founder of Neorent GmbH
“We had a major incident, and our tech support could see via APItoolkit which third-party integration partner was responsible, and could take action without needing the engineering team’s help.”
Joshua Chinemezum, CEO of Platnova
Why the full request matters
A 200 OK can hide 41 stuck orders.
A supplier drops tracking_id but still returns 200 OK. Without the full response, an AI investigation guesses wrong.
Missing payloads
POST /v1/shipments · 182 ms
- Status
- 200 OK
- Response body
- { unavailable }
AI: “The supplier returned 200. Check your code.”
Sampled requests
Only a sample of 41 failures retained.
AI: “Only a few errors. Probably noise.”
Short retention
Earlier history lostLast 15 days
AI: “No previous occurrence found.”
Response comparison · api.acme-logistics.com · before / after 07:52
HTTP/1.1 200 OK{ "shipment_id": "shp_7Hk2m9", "status": "acknowledged", "tracking_id": "AX-99812-NG",missing since 07:52 "eta": "2026-10-09"}
With monoscope
“Acme removed tracking_id. All 41 failures started at 07:52.”
monoscope opened the pull request
PR #491 · Fall back when tracking_id is absent
The fix is ready for your review.
Routines: AI checks that run on a schedule, then act on what they find.
Every 5 minutes
“Read the new logs and spans. If anything looks wrong, find the cause.”
Then posts the cause in #oncall
-
OpenTelemetry
- Errors
- Slack
After every deploy
“Compare errors, latency and response shapes with the previous version.”
Then replies in the deploy thread
- GitHub
-
OpenTelemetry
- Slack
Every night
“Find the slowest endpoints and queries, and make them faster.”
Then opens a PR with the optimization
-
OpenTelemetry
- GitHub
Weekdays at 09:00
“Summarize what broke overnight, what’s still open and what’s getting worse.”
Then sends a briefing to Slack
- Errors
- Linear
- Slack
01 · Know what happened
Catches what breaks
Stop hearing about it from customers. monoscope watches errors, logs, traces and the shape of every request, and tells you first.
-
Hear it before your customers do
The alert reaches Slack, PagerDuty or email with the exact request that failed, not a support ticket.
-
Catches failures that still return 200
Missing fields and changed types in any API you serve or call, caught in the response.
-
Flags what’s new, without rules
New errors, endpoints, request shapes and log patterns, flagged the moment they appear.
Requests, logs and traces
Payloads, telemetry and everything else that explains the failure.
Explore requests and traces
Metrics and dashboards
Track latency, traffic and errors on dashboards you keep as code.
View dashboards
Session replay
Watch the customer’s session and jump to the request behind the error.
See session replay
02 · Fix it
Finds the cause and opens the fix
Reads the failing request, its trace and recent deploys, then opens a pull request with the fix.
-
Opens the fix on its own
Errors, crashes and regressions get a pull request, with the request that proves it.
-
Answers in Slack
Ask why checkout is slower than yesterday. It answers from logs, traces and deploys.
-
Speeds up slow endpoints
Finds the slowest endpoints and queries, and opens PRs that make them faster.
Works with your coding agents
Ask your agent what broke. It reads production through monoscope.
The monoscope CLI
One binary that gives your terminal, your CI and your agents the same view of production, with stable JSON output for scripts.
curl https://monoscope.tech/install.sh | sh
monoscope logs search 'body has "tracking_id"' --since 6h
Agent skills
Skills for investigating incidents, on-call triage, managing monitors and dashboards, and adding instrumentation, in Claude Code or Codex.
Claude Code
claude plugin marketplace add monoscope-tech/skills
claude plugin install monoscope-skills@monoscope-skillsCodex
npx skills add monoscope-tech/skills -a codex
A hosted MCP server
About 50 tools for Claude Desktop, Cursor or your own agents, using the same API key, with nothing to install.
https://api.monoscope.tech/api/v1/mcporders-service — claude
03 · Prove it
Keeps the proof, and puts it on every PR
Every request and response is kept, unsampled. Before a pull request merges, monoscope checks it against that record: does it fix the issue, or break something that worked?
github.com/acme/orders-service/pull/512
duckdb · or the same SQL from any Postgres client
INSTALL delta; LOAD delta;
SELECT timestamp,
attributes___http___request___method AS method,
attributes___url___full AS url,
attributes___http___response___status_code AS status
FROM delta_scan('s3://your-bucket/timefusion/otel_logs_and_spans')
WHERE project_id = 'your-project'
AND date = DATE '2026-09-27'
AND attributes___server___address = 'api.supplier.com'
ORDER BY timestamp;
-
Keeps every request and response
Every call in and out, with bodies, trace and replay. One customer took it to court.
-
Puts proof on every pull request
Shows if a PR fixes the issue or breaks something, from real traffic and past incidents.
-
Keeps it in your bucket
Open Delta Lake and Parquet in your S3 bucket, queryable with DuckDB or SQL.
Security and your data
Your data, your bucket, code you can read.
- GDPR
- We’re happy to sign a data processing agreement (DPA) whenever you need one.
- EU data region
- If you’d prefer, we can store your data in the EU.
- SSO
- Single sign-on and advanced authentication come with Enterprise.
- Redaction
- The SDK strips out the headers and body fields you choose, such as
$.password, before any data leaves your app. - Data ownership
- With your own S3 bucket, your bucket is the database, so you can query it directly with DuckDB, Spark or SQL.
- Open source
- The whole platform is open source on GitHub under AGPL-3.0, so your team can read the code that handles your data.
Side by side
Every tool has an AI now.
Only monoscope reads the full request and response.
 | Datadog | Sentry | AI SRE layersPolylane, Superlog | |
|---|---|---|---|---|
| Request and response bodies, in and out | Yes, every call, never sampled | Not by default; traces are sampled | No | Only what your telemetry already has |
| Breaking changes in third-party APIs | Yes | Not in public docs | No | No |
| AI sees the full request | Yes | Not by default | No | Only if your telemetry has it |
| AI investigates and opens fix PRs | Yes, built in | Bits AI, priced separately | Seer, a paid add-on | Yes, on top of Datadog or Sentry |
| Pricing |
|
|
|
|
| Open source or your own S3 | Yes, AGPL; S3 from $199/mo | No | Source-available (FSL) | No |
Sources and assumptions
List prices from public pricing pages, September 2026 (Datadog host fee: Infrastructure $15 + APM $31; replays: RUM $0.15 + Session Replay $2.50; monoscope Cloud: $29 a month includes 20M events, then $1 per million; metrics included up to 20M datapoints). — means we couldn’t confirm it. Draft: verify every cell before publishing.
Pricing
Pay per million events, not per host, per GB and per index
Cloud from $29/month. Requests, logs and spans at one price, with no host fees.
20M
Adjust workload · 10 hosts · 400 series · 2k replays
Estimated monthly cost · same workload · 30-day retention
monoscope
Datadog
Splunk Observability
Sentry
Assumptions & price sources · September 2026
Estimates from public list prices, September 2026, annual billing where offered, 30-day retention on every vendor, 2 kB average event, one metric datapoint per series per minute.
Datadog: Infrastructure $15 and APM $31 per host; logs $0.10 per GB ingested plus $2.50 per million indexed; custom metrics $5 per 100 series beyond 100 per host; RUM $0.15 plus Session Replay $2.50 per 1,000 sessions.
Splunk: Observability Cloud App & Infrastructure $60 per host; Splunk Cloud log ingest at about $12.50 per GB/day per month; RUM $14 per 10,000 sessions; custom metrics not modelled.
Sentry: Team $26; spans $1.60 per million beyond 5M; errors at $0.29 per 1,000 assuming 0.5% of events are errors; replays $3.75 per 1,000; no metrics product.
monoscope Cloud: $29 includes 20M events, 20M metric datapoints and 2,000 replay sessions; then $1 per million events, $1 per 10 million metric datapoints and $1 per 1,000 sessions. Everything else is included.
Free forever · 10k events/day, no credit card
Optional: your own S3 bucket, +$199/month
Self-host for free
04 · Get started
Two ways to start
Let your coding agent set it up, or point any OpenTelemetry exporter at monoscope. 780+ integrations, free up to 10k events a day.
Recommended · about 2 minutes
With your coding agent
Ask your agent to instrument the project. It covers HTTP, SQL, queues and jobs, and checks that every route reports in.
Terminal
npx skills add monoscope-tech/skills
curl https://monoscope.tech/install.sh | shAny OpenTelemetry SDK or collector
By hand, in any language
Set four environment variables. Our SDKs for Node.js, Python, Go, PHP, Java, .NET and Elixir also capture bodies.
.env
OTEL_EXPORTER_OTLP_ENDPOINT="http://otelcol.monoscope.tech:4317"
OTEL_EXPORTER_OTLP_PROTOCOL="grpc"
OTEL_SERVICE_NAME="orders-service"
OTEL_RESOURCE_ATTRIBUTES="x-api-key=YOUR_API_KEY"
Already on Datadog or Sentry?
Run monoscope next to it for a week or two, compare which errors and API changes each one catches, then move your services over. An engineer from our team can help.
What changed for teams on monoscope
View all reviews20x
faster MTTR
70%
reduction in manual troubleshooting
3x
faster system recovery
5x
faster incident debug times
9x
less customer support calls
500ms
shaved off improved endpoints
We had a major upgrade planned for our iOS/Android app... After adding monoscope to our laravel app, we could see every new route and request. It gave me and the dev team complete confidence that our users weren’t being affected by our changes.
... became the smoothest and LEAST STRESSFUL rollouts we have ever done. Really an amazing tool and now a permanent part of our workflow. Thank you!!
Lazarus Morrison
Founder of Community Fluency
The best observability tool we use today at Woodcore, monoscope notifies us about any slight change that happens on the system.
Most especially, for the features we utilise today on monoscope, would cost us a lot more elsewhere.
Samuel Joseph
CEO of Woodcore
I fell for monoscope because it integrated effortlessly with my application and provided valuable API insights. Whenever I needed help, the team was always ready to listen and resolve any issues. That’s why we pay for their service.
Odohi David
CTO of Grovepay
