Roadmap
What we're building next
Built for the people who decide about money. Software development is an expense on your books, and you want to invest it wisely. Everything below serves one question — what did that spend return, and what would a different decision return — with the uncertainty stated. What's live now first, then what we are working on, what comes next, and what comes later. We ship weekly — see the changelog.
Available today — full product experience
GitHub + GitLab
Commits, PRs, reviews, issues, CI. The full end-to-end product experience — dashboards, reports, and AI chat wired through.
Jira
Issues, sprints, boards, workflows. Canonical model landed and live for plan-vs-reality and sprint work.
Linear
Issues, cycles, projects, teams. Fully wired end-to-end — not a future connector.
LocalGit
On-prem agent for pre-push local analysis and blast radius. Source code never leaves your network.
Custom dashboards
Compose a dashboard from widgets over existing read endpoints and skills — table, line, bar or single number, positioned on a grid. Persisted per customer, with save, share and schedule.
Audit log SIEM export
Stream API and MCP audit events to your SIEM. GET /api/v1/audit-log/export returns JSONL with a cursor for scrapers, plus webhook registration for push delivery.
The analytic catalogue as a platform
388 calibrated decision functions, each callable from chat, the API (POST /api/v1/capabilities/{id}/run) and the MCP server, discoverable by the question you are trying to decide rather than by name.
AI Spend Governance
Per-user and per-organisation monthly caps enforced at call time, a live meter, an absolute ceiling that fails closed. You decide what the AI layer may cost before it costs it.
GitHub Copilot Metrics API + Cursor Admin API
Authoritative AI-assistant usage in the canonical model, joined to licence cost — so AI ROI is measured, not inferred.
SAML 2.0 SP login + EU residency tier
SAML ACS with just-in-time provisioning completes the SSO story beside SCIM 2.0. EU residency is enforced at provisioning time with KMS-encrypted backups.
MCP stdio server + Azure DevOps connector
A real MCP server with read and write tools for Claude, Cursor and every other MCP client; Azure DevOps at parity with GitHub, GitLab, Jira and Linear.
see.gitrev.io
A public, read-only tenant on real-shaped data, reset nightly. See what the answers look like before connecting anything.
Data ingestion available — dashboards rolling out
The canonical ingestion pipeline (extract → transform → 3NF) is live for these vendors today. Connect them and the data flows into the model; dedicated pickers and dashboards are rolling out.
Bitbucket Cloud
Commits, PRs, branches. Canonical ingestion live; in-app picker and dashboards rolling out.
Slack + Microsoft Teams
Channels, users, message metadata (PII-stripped). Canonical chat ingestion live for communication-pattern signals.
PagerDuty + Opsgenie
Incidents, services, on-call schedules. Canonical incident ingestion live — feeds DORA-4 and on-call burden.
incident.io + Rootly
First-class incident lifecycle ingestion for teams standardized on these platforms.
Sentry, Datadog, New Relic, Honeycomb
Observability signals ingested canonically, ready to correlate with deploys and contributors.
Jenkins, Buildkite, CircleCI, Argo CD
CI/CD ingestion beyond GitHub Actions / GitLab CI — build times, failure rates, deploy frequency.
GitHub Actions + GitLab CI
Pipeline runs and deploy signals ingested from the CI providers you already run.
BambooHR + Personio
HR-system ingestion for tenure, org structure, and headcount context.
Zulip
Threaded-chat ingestion (PII-stripped) as an alternative communication datasource.
Google Calendar
Meeting-load and focus-time signals ingested for context-switching analysis.
CSV upload (BYOD)
Bring custom data without a connector. Upload a zip of CSVs; each maps to a table in your own byod_raw schema. Idempotent on the hash of the uploaded bytes.
Generic inbound webhook (BYOD)
Accept arbitrary JSON on a per-stream endpoint with its own bearer token and rate limit. Create and list streams over the API; query alongside ingested data.
Every answer is a decision surface
The 388 functions already answer; the work is making each answer land as something a budget holder can act on. Forecast calibration and causal-robustness evidence panels, stop / continue / scale triage boards, scenario and policy trees, reconciliation waterfalls and ownership explorers ship for the first families of functions. Next: the same treatment across the whole catalogue, so no result arrives as a bare table.
Suggestions that lead to the right function
The empty chat screen suggests questions resolved against what you have connected; a suggestion that leads nowhere is recorded as such. Next: suggestions ranked by the money at stake in your data — the initiative most over budget, the team whose cost per outcome moved most — rather than by what is merely popular.
A receipt with every answer
Each agent-analytic result carries a Decision & Evidence Receipt: which data, which method, which assumptions, which limitations. Next: receipts you can hand to a board or an auditor as-is.
Cost per outcome, everywhere
Labour cost, contractor-vs-FTE split and AI licence spend are already in the canonical model. Next: every velocity, quality and risk view carries a cost axis by default, so 'faster' and 'better' are always shown next to 'for how much'.
Better maths, honestly labelled
Point numbers give way to intervals: p50 / p75 / p90 on every forecast, calibration tracked against what actually happened, and the interval widened when the model has been wrong. Bayesian pooling where the data is thin; change-point detection so a trend that has already turned is not extrapolated.
Capital-allocation questions, answered directly
What does one more engineer buy, in this team, this quarter? What is the cost of delaying this initiative a month? Which of these five projects should stop? Real-options valuation, cost-of-delay and portfolio constraint explorers over the canonical data — the questions a CFO asks, with the uncertainty stated.
Product analytics as datasources: Google Analytics, Kissmetrics, PostHog
Engineering work is attributed to cost today; next it is attributed to value. Product-analytics usage — sessions, feature adoption, activation and retention by user — joins the canonical model alongside commits, PRs and initiatives, so the question becomes: which of the work we paid for did users actually use, and what did it change.
Counterfactual budgeting
The What-If Simulator already runs Monte Carlo over reassignment, AI-tool rollout and framework migration. Next: budget scenarios — headcount, contractors, tooling — compared on expected outcome and downside, with synthetic-control estimates of what a past decision actually changed.
New visualisations for the money view
Budget versus outcome over time; dependency cascades that show what a slipped initiative takes down with it; a what-changed-and-why narrative generated from the metric dependency graph; and an executive one-pager and PowerPoint export refreshed for all of the above.
DORA as an input, not the goal
Deployment frequency, lead time, change-failure rate and time to restore are computed and kept. They are four of the several hundred inputs the platform reasons over, and the roadmap does not have a DORA phase — the destination is an analytic platform that can say what engineering spend returned, which DORA alone cannot.
Knowledge graph + bus factor UI
File × contributor expertise reduced to single points of failure, with cross-train recommendations and reachability heatmaps — the people-risk side of the investment picture.
Release Risk PR-comment integration
Per-PR risk score and blast radius surface as a comment on every PR; cross-team impact notifications go to affected team leads.
Ingestion-vendor dashboards
The vendors with canonical ingestion get dedicated in-app pickers and dashboards, one by one.
Slack bot — Block Kit charts + user mapping
Native install and a signature-verified slash command already ship. Still to come: chart rendering in Slack and Slack-to-Gitrevio user mapping.
Account-level TOTP + container scanning
Staff two-factor already ships; account-level TOTP for every user, container CVE scanning in CI, and adversarial hallucination tests on the AI layer.
Skills marketplace
v1: browse the builtin catalogue, install skills your organisation has authored, activate per team. v2: a public marketplace with builtin, community-verified and community-experimental tiers.
Codeium, Cody, Amazon Q
AI-assistant integrations beyond Copilot, Cursor and Windsurf, closing out the AI ROI matrix.
Stripe billing, official SDKs, MCP Docker distribution
Self-serve plans with seat management and EU VAT; Node.js and Python SDKs over the OpenAPI spec; a Docker channel for environments where npm and PyPI are not first-class.
Want to influence what we build next?
We prioritize based on what customers and early adopters actually need. If there's something on this list you're waiting for — or something missing entirely — tell us. It directly affects the order we build things.
Email us at info@gitrev.io