What you get
Codified investigations
- Your most frequent incident types written as runbooks Claude follows step by step
- Reasoning steps, known pitfalls and confirmation gates in each one
- Connections to your existing tools: shell, cloud APIs, metrics, logs and chat
Deterministic parsers
- Tested Go or Python code that parses, aligns and classifies your telemetry
- The same input gives the same answer, every time
- Reports a reviewer can check line by line
Accountability
- An audit log of every command, with recall for later incidents
- Read-only by default; changes need an engineer's confirmation
- Draft write-ups an engineer reviews before posting
How the build runs
Discover week 1
Walk through recent incidents with your on-call engineers. Pick the incident classes that repeat most and cost most. Agree the scope in writing.
Codify weeks 2–3
Write the runbooks and connect them to your tools. Each one is run against a real past incident to check that it reaches the right answer.
Compute weeks 3–5
Build the parsers and analyzers your runbooks call, with tests built from your own data.
Hand over week 6
Documentation, a training session for the rotation, and tests in your CI. You own all of the code.
Typical timeline for an initial scope of a few incident classes. The written scope from week 1 is what you pay for.
After the build: the retainer
Systems change, and automation that isn't maintained goes stale like any runbook. The monthly retainer keeps it current:
- Monthly review of new incidents, turning lessons into runbook pitfalls and new parser features
- Updates when your infrastructure, tools or alerts change
- New runbooks as new incident classes appear, within the agreed hours
- Async support for the on-call rotation
Price
| Price | Terms | |
|---|---|---|
| Build: 3 incident classes | From $36,000 | Fixed price for the written scope; 50% at kickoff, 50% at handover |
| Each additional incident class | From $10,000 | Added to the scope in writing before work starts |
| Retainer: Standard | $2,500/mo | 10 hours a month, billed monthly in advance |
| Retainer: Extended | $5,000/mo | 20 hours a month, billed monthly in advance |
The final price is set by the scope agreed in week 1. If you've had a readiness assessment in the last 60 days, its fee is credited.
Where it runs
Everything runs inside your environment, with your credentials and your own model access. Your telemetry doesn't pass through HASHJOIN's systems, and nothing depends on a HASHJOIN service staying up. The runbooks are built on Claude Code; the parsers are ordinary code you can read, test and change.
Start with the assessment
Not sure what to automate first? The readiness assessment answers that, and its fee is credited toward a build started within 60 days.