Data Fastlane runs a data team's whole operating system — a dual-lane model with service queues for ad-hoc work and projects for planned initiatives, domain-aware stakeholder intake, and AI that clarifies, scopes, and (when you want) implements. We're picking 3 design partners to help us ship it.
Control plane runs in our infrastructure. The data plane — the orchestrator and workers that touch your warehouse and repo — runs inside yours. Warehouse credentials, dbt keys, API tokens: all stay in your VPC. We never see them. We never store them.
A PM platform for data teams, with AI woven in where it helps — not forced where it doesn't. Everything below is buildable today against the current codebase. Design partners get it free.
Branded intake URL per workspace (datafastlane.io/{org}/submit). Domain-verified email auth — no stakeholder seats. Every request lands in the right Queue or Project with full context.
Five templates built for data work: New Model, Metric Definition, Data Quality Issue, Source/Integration, Analysis/Research. No generic "describe your request" forms — every template asks the right questions for that kind of data work.
Snowflake, BigQuery, and Postgres schema introspection plus hybrid search over your dbt docs and runbooks. Every AI decision grounded in your real tables and institutional knowledge.
A service Queue for ad-hoc tickets. Project lanes for planned initiatives. Requests can float wherever they belong. A PM model that fits how data teams actually split their work — not a one-size-fits-all backlog.
Multi-round conversational agent that asks the right follow-ups per request type, grounded in your warehouse schema and KB. Turns vague stakeholder asks into a scoped PRD your team can approve in one click.
For AI-eligible tasks, Temporal Cloud workflows draft the PRD, split into tasks, write dbt/SQL with quality-gate retries, and open a reviewed PR. Humans approve at every stage. The rest of your work moves through the PM normally.
Kanban plus list views across your workspace. Assignees, due dates, comments, statuses, priorities. Drill from Task → Request → Project/Queue without losing the thread. PM basics, done right for data work.
First pass on every AI-generated PR. Checks dbt best practices, schema consistency, test coverage, and flags pre-existing failures against main. Your engineers see the bot's notes before they open the diff.
Bring your own OpenAI, Anthropic, or internal LLM keys. Route every AI call through a control plane you own — with per-team budgets, token monitoring, and model switching. Your data plane runs entirely in your VPC.
LIVE INTEGRATIONS
This is the opt-in AI pipeline for AI-eligible work. Manual work moves through the same Queues and Projects with none of the AI steps forced on you.
A stakeholder hits your hosted intake portal, picks a request type (New Model, Metric, Data Quality, Source, Analysis), and fills a template built for that kind of data work. Jasper asks domain-specific follow-ups, grounded in your warehouse schema and KB.
Jasper generates a structured PRD. Your team reviews and approves in one click, or edits in place. Nothing moves without sign-off.
The approved Request is decomposed into Tasks, routed into the right Queue (ad-hoc service work) or Project (planned initiative). Each Task is classified AI-eligible vs needs-human, with complexity score and dependency ordering.
Your Workspace Lead reviews the plan on the board. Approved Tasks move to Stage 3. Flagged Tasks go to a human assignee. Nothing starts without sign-off.
For AI-eligible Tasks, the agent writes dbt models, SQL, and tests with a quality-gate retry loop, then opens a PR. Our AI Code Review Bot runs first — dbt best practices, schema consistency, test coverage, pre-existing-failure detection.
Your engineers review the bot's notes plus the diff. Your existing CI runs dbt compile and dbt test. Two approvals. Ship. Request to merged PR, measured in hours instead of weeks.
Data Fastlane's data plane runs inside your infrastructure. Your Snowflake / BigQuery / Postgres credentials, your GitHub tokens, your LLM API keys — all stored in your environment, encrypted at rest. We orchestrate. Your secrets never leave your VPC.
Current roadmap items — design partners get a direct line to what ships next and a vote on what jumps the queue.
One New Request entrypoint with a mode toggle: AI Workflow or Manual. Context-specific fields per request type. Eliminates the split between intake and manual request flows.
See Task → Request → Project/Queue everywhere an asset is rendered. Clickable breadcrumb chips on cards, rows, modals, and detail headers. Never lose your place in the hierarchy.
Full manual request detail page with inline task-creation UI. Build out a request's task list without leaving the page. For teams that want the PM without the AI pipeline.
Check fastlane.yml into your repo: LLM provider/model, quality gates, requirements template, concurrency. Version-controlled, reviewable, team-managed.
Searchable activity log for every agent action, approval, edit, and deployment. Exports and user attribution. Compliance-ready event history, filterable by Workspace.
Guided first-run setup connects your warehouse, GitHub, Jira/Linear, and KB in under 20 minutes. First hosted intake URL live before your coffee gets cold.
Three companies. Free access during the preview. A real seat at the table while we ship this.
▪ We're taking 3 design partners for this cohort. ▪ Data stays in your VPC. ▪ 30-minute intro call — no commitment.
We're looking for 3 data teams. If this describes you, let's talk.
We're looking for three data teams to work with us hands-on. You get the full product free, white-glove setup in your VPC, and a direct line to the founder. We get sharp feedback, real workload, and a partner for every roadmap decision.
30-minute intro call. No commitment. If we're a fit, you get lifetime early-adopter pricing when we GA.