Forge Brain reads your code, telemetry, server logs, tickets, runbooks, and team chat — and serves grounded, cited answers to your field technicians, support engineers, and operators. Without ever exposing the sources.
Your field technician is in front of a broken device. Your support engineer is on a 2 AM call. Your contractor is debugging a customer's deployment. They all need the same thing: an answer that exists somewhere in your org.
It's in a Slack thread from last March. A Confluence page from 2022. A commit message someone forgot to write. The fault-code semantics that only live in the firmware repo. The fix Marc found before he left.
Today, the person who needs it opens five tabs. Pings two Slack channels. Pastes log snippets into ChatGPT on their phone — without a DPA, without an audit trail, without retention.
When they leave, the answer walks out with them.
A field tech opens their tablet. Asks one question. Watch what the brain does.
The brain doesn't just answer. It files the ticket, enriches it with prior incidents, and — if you want — triggers a cloud development session that opens a draft PR. Your tech reports once, in plain language. Everything after is automatic.
The brain reads code semantics + prior fixes + telemetry to scope the ticket, then dispatches a
cloud development session in your sandbox that opens a draft PR linked back to the original report.
Every step audited. Your tech's job is to say what's wrong. Everyone else's job got smaller.
Every source you run, ingested into one persistent, time-aware graph. Synthesized at query time — with citations.
| Capability | Forge Brain | Knowledge platforms |
Observability / APM |
Code AI assistants |
ITSM intelligence |
|---|---|---|---|---|---|
| Code semantics | ✓ | — | — | ✓ | — |
| Server logs | ✓ | — | ✓ | — | — |
| Device telemetry | ✓ | — | partial | — | — |
| Docs, wiki, chat threads | ✓ | ✓ | — | — | partial |
| Tickets, runbooks, prior fixes | ✓ | partial | — | — | ✓ |
| Synthesis across all of the above | ✓ | — | — | — | — |
| Citations on every claim | ✓ | partial | — | — | — |
| Time-aware (knows what's stale) | ✓ | — | — | — | — |
| Offline-capable (ships to device) | ✓ | — | — | — | — |
| Customer-VPC deployment | ✓ | rare | rare | — | rare |
| Audit-grade by default | ✓ | partial | partial | — | ✓ |
Knowledge platforms, ITSM intelligence, code-AI assistants — all cloud-only, all single-source, all conclusions without citations. Forge Brain inverts every one of those defaults.
Your team doesn't want another app to learn. You don't want another integration. Forge Brain ships as an SDK — embed it inside the tools your org already runs.
// Ask the brain anything across your org's // data — code, logs, telemetry, tickets, chat. import { brain } from "@forge/brain-sdk"; const answer = await brain.ask({ : "dylan@yourcompany.com", : "field-tech:east-region", : { : "NS-4823" }, : "Fault E47 — what does telemetry show and what's been tried before?" }); // → grounded · cited · audited · supersession-aware
We're partnering with three industrial leaders this year. You bring the first real use case. We bring the brain, the SDK, and the engineering team. Together we ship the answer your distributed team should have had years ago.