Connect databases, SaaS tools, files and warehouses. Sluice builds a governed data layer, creates your KPIs and lets your team ask questions in plain English — with every answer traceable back to its source.
Plain English in, a governed dashboard out — on a real Czech retail-banking dataset (5,369 clients across 77 districts). The AI plans it, a deterministic engine writes the SQL, and every number traces to source.
Ingestion, transformation, orchestration, governance and BI all live in separate systems.
Every new source or metric becomes another ticket in someone's backlog.
Even when a number reaches a dashboard, it's hard to prove where it came from.
Sluice is the fourth option: AI plans the pipeline, deterministic code executes it, you approve every step — and it all runs inside your own cloud.
The pipeline stops after profiling and won't build until you approve the AI's plan. AI plans, deterministic code executes, you hold the gate.
The modern stack is four separate contracts — before the engineers to wire and run it.
Ranges are realistic market prices for the equivalent tools — not quotes. Your cloud data warehouse is excluded (you pay it either way).
Paste a connection string, drop a file, or pick a SaaS source. Sluice profiles every table, fixes types and encodings, runs quality checks, and masks PII before anything moves.
The AI reads your real schema — and your requirements doc, if you upload one — and proposes entities, value-verified joins, data marts, KPIs and a dashboard. You edit and approve. Nothing builds without you.
KPIs build deterministically from the approved plan, refresh on a schedule, feed Power BI & Excel, and answer plain-English questions — every number reconcilable to the raw source.
No new hires, no six-month setup, no five vendor contracts. Start free, pay only as your data grows — and get the output of a data team for a fraction of the cost of one.
Sensitive data is detected and protected automatically, every run leaves an audit trail, and your data never leaves your cloud. Governance isn't an add-on — it's the default.
Ask a question in plain English, get a board-ready answer in seconds — from numbers you can stand behind. And it feeds the tools you already use, like Power BI and Excel.
The platform is the only thing that's ours. The medallion builds and lands inside your cloud, on your compute — we store encrypted credentials and metadata, never your rows. Don't have the infrastructure yet? Run on ours, and move it in-house whenever you're ready.
Bronze, Silver and Gold build and land inside your infrastructure. Nothing leaves your perimeter.
Encrypted credentials + schema statistics — never values. PII stays masked, always.
Start on Sluice-managed compute & storage, then bring it into your own cloud on your timeline.
Dashboards were built for humans. The next thing that reads your analytics is your company's AI agents — and they need governed, deterministic, PII-safe access, not a database password. Sluice exposes your metrics over MCP: hand any MCP-capable agent (Claude, GPT, your own) a scoped key, and it can answer questions from your governed numbers — without ever touching raw data or writing SQL.
Agents read the same masked Silver & Gold the UI does — PII is masked before this layer, raw data is never exposed.
The agent proposes; our compiler verifies. Same question → same number, no agent-authored SQL, an honest "no" when it can't be resolved.
A per-project key exposes only that project. Revoke instantly; every call is logged for audit.
Connect an agent in one line
# add your governed metrics to Claude — or any MCP agent
claude mcp add sluice --transport http https://sluicedata.com/mcp \
--header "Authorization: Bearer ade_..."
Every run walks the medallion — land raw, govern & shape, aggregate to KPIs — so the numbers on your dashboard are typed, deduped, PII-masked and reconcilable to the source.
Sluice profiles every source, auto-detects headers and encodings, cleans each table, runs quality checks, and detects & masks PII — email, phone, IDs — before anything moves downstream.
Sluice conforms types, applies masking, and detects relationships from the data itself — composite keys, dirty-key normalization, multi-hop chains — so joins are verified against your rows, never assumed.
One Gold table per KPI, generated deterministically from the approved plan — deep joins, ratio-of-aggregates, running totals, month-over-month, moving averages, rank and share-of-total — all fan-out-safe, so a category total sums back to the grand total.
Not a chart. Not a dashboard. A full, approvable plan over your actual schema — entities, relationships, data marts, KPIs and the target dashboard. Upload a requirements doc and it reads it. You edit, approve, and the whole pipeline builds from it.
Entities, joins, marts, KPIs and a dashboard — approvable in one place.
Each KPI carries its SQL, its grain, and reconciles to the raw source.
Approved joins & meanings compound — each run starts smarter.
Databases, warehouses, files and APIs. If it has a connection string or an endpoint, Sluice already speaks to it — and lands it as governed tables in your cloud.
Our stake: the intelligence is ours, the data is yours. The AI plans the pipeline, generated code executes it, and every row stays inside your cloud — the output of a data team, without the data team, and without your data ever leaving.
A simple base plan plus pay-as-you-go usage — no per-seat, ever; your whole team is included. Pushdown to your own warehouse is always unmetered, and governance — PII masking, contracts, quality, lineage, audit — is in every plan. Storage is billed only if your data rests on us. Annual billing = one month free.
For evaluation — the full pipeline, free.
For a first production pipeline.
For growing data teams.
For regulated & scaling orgs. Your cloud, your compute, our platform — or ours, if you'd rather.
Pay-as-you-go, same rate on every plan: processed-GB past your plan billed per GB · ask-anything past quota billed per 1,000 questions ·
storage only if your data rests on us — ₹6 / GB-month (free when you connect your own cloud) · pushdown always unmetered.
Compare the alternative: one data engineer is ₹1–2.5L/month, fully loaded — before any tooling. Every Sluice tier returns more than it costs from the first pipeline.
Start free with your own data — or book a 30-minute call and we'll connect a source and build your first KPIs live, in your cloud.