Triage every new defect against everything you've ever solved.
Precedent turns your team's years of resolved tickets and code fixes into a living memory — so a new escalation, or even a raw log dropped in before a ticket exists, gets matched against everything you've already solved, before an engineer opens the file.
A senior engineer's afternoon, every single day.
A customer hits a failure. A ticket lands with a log bundle attached. Someone senior spends hours grepping proprietary log formats, searching old tickets by memory, and asking around for whoever "just knows" what that error means.
The answer almost always already exists — in a ticket from three years ago, in the commit that fixed it, in an engineer's head. It's just not searchable. So the same root causes get re-diagnosed from scratch, MTTR stays high, and the knowledge walks out the door the day that engineer changes teams.
Precedent makes that knowledge permanent, searchable, and automatic.
From raw logs to cited triage, automatically.
Connects to what you already have
Precedent ingests your ticketing history (Jira, Salesforce, or equivalent) and your git history — tickets, comments, resolutions, attachments, and every commit that ever referenced one. No new workflow to adopt.
Understands your actual log formats
Not just generic text search — Precedent parses the log and event formats your systems really produce, including proprietary and binary formats other tools can't touch. Structured events, error signatures, and stack-trace fingerprints get extracted automatically.
Builds institutional memory, not just an index
Every ticket, fix, and failure signature gets indexed together — text and structure, semantic and exact. The link between a ticket and the commit that closed it is preserved.
Triages before a human looks at it
The moment a new ticket lands — or even before one exists, from a raw log dropped in during internal testing — Precedent posts a cited summary: what the logs show, the closest historical precedents, whether this looks like a regression or a known issue, where the fix likely lives, and who's fixed something like it before.
Gets smarter with every resolution
Every new fix feeds back into the memory. The system's coverage compounds — it never stops learning your systems' failure history.
Built for support and engineering alike
For support & sustaining engineering
See the closest historical ticket — how it was triaged, what the fix looked like, and what to tell the customer. Stop re-diagnosing issues your team has already solved.
For engineering
See the closest historical fix — the actual commit that resolved this failure pattern last time, and the files it touched. Go from root-cause hypothesis to fix location in the time it takes to read a summary.
Every claim is cited back to a source ticket, commit, or log line. If the evidence is thin, Precedent says so — a triage tool that guesses is worse than one that admits it doesn't know.
Built for proprietary logs, not clean cloud telemetry
It learns your formats, not just your text
Generic log tools and AI assistants work on plain text. Precedent's parsing framework onboards proprietary and binary log formats as fast as writing a spec, not building a bespoke parser each time.
It catches issues before a ticket exists
Point Precedent at a raw log from an internal test run or field return, with no ticket filed yet, and get the same triage. Precedent has something to search from day one.
Your logs never leave your network
Deployed inside your VPC. No log data, no proprietary format, no customer information is sent anywhere outside your environment.
It's model-agnostic, by design
Built on a thin, swappable LLM layer — it improves as the underlying models improve and gets cheaper as inference prices drop, without changing how your team works with it.
It doesn't stop being maintained
This isn't a one-off internal hackathon tool that quietly rots when its builder moves on. Evaluation and outcome-tracking discipline is part of the product, not an afterthought.
Teams drowning in proprietary log bundles
- VPs and Directors of Support / Sustaining Engineering running 15–50 person escalation teams who need to cut MTTR and stop burning senior headcount on repeat issues
- VPs of Engineering and staff engineers who want root-cause-to-fix in minutes, not a re-investigation from scratch
- Teams in networking, storage, embedded, telecom, industrial, and medical device companies — anywhere field issues arrive as proprietary log bundles, not clean cloud telemetry
Not for: teams looking for a cloud observability platform. Precedent is offline, attachment-based, proprietary-format field-issue forensics — a different problem than tools like Datadog or Dynatrace solve.
Where this stands today
Precedent is in active R&D — Stage 1 proved the retrieval + triage approach on Apache Kafka's public issue history, and we're iterating on accuracy before any real company's data is touched.
We're looking for a small number of design partners to pilot the next stage with real ticket history and log formats.
Push, not pull.
Precedent is built to sit inside your existing ticket pipeline — posting triage summaries directly to tickets, answering follow-up questions inside Jira, or through your team's own AI assistant. Always on, not a chat window someone has to remember to open.