Connect your tools. DevHub shows what is wrong and why, and tells each person what to do next.
A new engineering manager joins a company whose codebase, architecture and teams they don't know. They connect the tracker, the code host and CI. Within minutes they can see the architecture, who owns what, where sprints stall and which PRs nobody reviews.
Normally that takes months of onboarding, or a platform team building dashboards. DevHub measures time to understanding in minutes, not months.
These are screens from DevHub running in demo mode. Every page appears only when its source is connected.
Pace against time left, carryover and critical items per team, so you see which sprint is slipping before the review meeting.
A why engine for what changed, built on the links between tools: deploy, incident, flag change, PR, ticket, sprint, cost and team.
Architecture, module dependencies, ownership, teams, workflow stages, git and release flows are inferred from your data. Each inference shows where it came from, and an admin can correct it.
Read-only until you allow a write. Data scope tested on every endpoint. Vendor access only with consent. On-prem with your own AI.
DevHub reads with your own credentials and caches so it never hits rate limits. In-house tools connect through a generic connector, with no code.
GitLab, Bitbucket, Azure DevOps and Jenkins are on the roadmap.
AI code review
The AI pull request reviewer that comes with DevHub. It reviews better inside DevHub than on its own, because DevHub already knows the architecture, the owners, the tickets and the history.
Visit DiffwiseAnd you see everything they do.
Every write (a ticket, a PR, a comment, a message) needs an explicit grant and a confirmed preview, and is audited.
Per-team data scope, proven by tests on every endpoint.
SSO through OIDC, SAML or Google, managed accounts with MFA, directory sync with SCIM, owners and departments.
Support access is time-boxed, audited, and only with your consent.
Retention and data-rights tools are built in. Exports are always available, so your data is never locked in.
Run it in your network with your own AI provider, or switch AI off. Admins choose which data may be sent to AI.
One isolated instance per customer. Sign in with your identity provider; nothing to run.
One container image with a persistent data volume, on Docker, Compose or Kubernetes, with your own secrets and AI provider. Licenses are checked offline.
About 1,500 automated tests, a release check, live smoke checks, and layout and team-leak checks on every page.
Good tools exist in each category. DevHub connects what they keep apart.
DX, Jellyfish, LinearB, Swarmia
Delivery metrics and dashboards.
DevHub adds: Reads the code too, explains why, suggests the fix and opens the ticket or PR.
Backstage, Port, Cortex
A catalog you build and maintain by hand.
DevHub adds: Nothing to fill in: ownership, architecture, workflows and release flow are inferred from data.
CodeRabbit, Greptile, Sourcegraph
Deep on code.
DevHub adds: Connects code to sprints, releases, incidents, flags, cost and people.
Atlassian, GitHub, GitLab
Strong inside their own suite.
DevHub adds: Neutral: any tracker, code host, CI, cloud, billing or AI provider, including in-house tools.
Priced per active user per month. Each plan adds to the one before it.
Free or low cost. Dashboards, sprints, work items, pull requests, repositories, pipelines and search.
Adds AI, review and health, releases, planning, tests and alerts. Unlimited AI PR review.
Adds observability, feature flags, finance, agents and the why engine.
Adds SSO and SCIM, on-prem, your own AI provider, audit export and consented support access.
Thirty minutes, read-only, nothing installed in your repos. Your architecture, your stuck PRs, your bottleneck.