Observability,
finally rebuilt around AI.
Squasher unifies errors, logs, traces, and uptime into one observability platform — with AI agents woven into every surface to detect, diagnose, and resolve incidents.
Monitors
12
Most "AI features" are bolted on. We rebuilt the platform around the agent.
Squasher is not Sentry with a chat sidebar. The AI is in the ingest pipeline, in the alert routing, in the incident timeline, and in the fix loop — because the data shape was designed for it.
AI in the ingest path.
Every event is fingerprinted, grouped, and enriched by an LLM on the way in. Stack traces dedupe against your codebase, not a string hash. Logs cluster semantically. The signal is clean before you see it.
AI on the alert surface.
When an alert fires, an agent reads the stack trace, cross-references the last 30 days of deploys, and posts the root cause — with file, line, and likely commit — before the page reaches a human.
AI in the fix loop.
The agent opens a tight pull request with a regression test. On confirmed sev-1 regressions, it rolls back automatically. Every action is cited and reversible.
Every exception, grouped by an LLM on the way in.
Stack traces are fingerprinted against your codebase, not a string hash. So three different 'TypeError: undefined' from three different files become three issues, not one.
- 13 first-party SDKs
- Semantic grouping, not string hashing
- Group by deploy + auto-resolve
One ingest endpoint. Native OTLP.
No exporters, no Vector, no Fluent Bit. The AI uses the trace graph to walk between services when diagnosing — so you can finally ask 'why is checkout slow?' and get an answer.
- Native OTLP ingest
- Log clustering, not log search
- Trace-aware AI diagnosis
Thirteen monitor kinds. Six regions. One bill.
HTTP, heartbeat, DNS, SSL, TCP, SMTP, port — checked every 30 seconds from iad, sfo, fra, lhr, syd, gru. Public and private status pages stay in sync with the incident.
- 30-second cadence, 6 regions
- Status pages auto-sync with incidents
- SSL + DNS expiry alerts included
The agent takes the page first.
Schedules, rotations, overrides, escalations — designed for human responders. The AI takes every page first; humans only get woken when judgement is needed.
- AI takes the page first
- Native iOS + Android apps
- Schedule overrides with one tap
On-call~ 412 customers affected
From alert to mitigation, in one timeline.
When something breaks, an incident opens. The AI investigates, the team gets a Slack ping with full context, the fix lands as a PR. The whole arc lives in one place.
- Auto-correlate alerts → incident
- PagerDuty, Slack, Discord, statuspage
- Every action audited + reversible
checkout-svc · NullPointerException spike
gh pr create · #4221 with null-check + regression testBuilt by an on-call rotation that isn't sleeping anymore.
Behind every Squasher surface is a specialized agent. They share the same data model, hand off cleanly, and only escalate to a human when judgement is required.
InvestigatorReads the stack trace, finds the offending commit.
Stream watcherTails logs and traces, surfaces the anomaly.
MechanicOpens the pull request and the regression test.
First responderRolls back the deploy in under a minute.
Give this to your agent.
Install Squasher's agent skills and hosted MCP once, then hand the onboarding prompt to your coding agent. It can add the SDK, drains, or OpenTelemetry path that fits your repo.
# Install Squasher skills + hosted MCP for your agent CLI
bash <(curl -fsSL https://squasher.ai/install.sh) --agents -y
# Then give this prompt to your coding agent:
# Use $squasher-onboard to install Squasher in this repo
# and verify one real runtime event.
# Skills index:
# https://skills.squasher.ai/Priced by event volume. Not by per-seat tax.
Every plan ships the full observability stack. AI triage is included on Team and up — no nickel-and-diming the incident response.
Pro
Production for small teams
Start free- AI triage on every incident
- Source maps + release tracking
- Replay-linked issue triage
- All log drains
Team
Replace Sentry, replay, and status tooling
Start free- Everything in Pro
- Auto-fix pull requests
- GitHub integration
- Status pages + maintenance windows
Scale
High-volume teams + procurement
Talk to us- Everything in Team
- SSO + SCIM + audit log
- Dedicated VPC option
- Solutions engineering
Switch to AI-native observability.
Install the SDK in 90 seconds. The agent diagnoses your first real incident the same day.