Solutions · Business Insights

Ask your data anything,
in plain English.

Build an insights agent on SentientOne, connect a read-only view of your database or analytics API over MCP, and let the whole team ask questions. The agent queries live data and explains what it finds — no SQL, no dashboard sprint.

How did Q2 signups compare to Q1?
Q2 signups were 4,812 — up 23% on Q1's 3,905. Most of the growth came from the Pro plan.
Answered from your live systems · traced

Everyone has questions.
Dashboards answer few.

Every metrics question becomes a ticket for the data team or another dashboard nobody maintains.

01

Analyst bottleneck

Simple questions queue for days behind real analysis work — so people stop asking.

02

Dashboard sprawl

A dashboard for everything, and still never the exact cut you need for today's decision.

03

SQL gatekeeping

The data is all there. Most of the team just has no way to query it themselves.

04

Gut-feel decisions

When numbers take days, decisions get made without them — and nobody double-checks later.

An analyst on demand,
for the whole team.

An insights agent on SentientOne turns plain-English questions into live queries — and explains the answer, not just the number.

Plain-English questions

“Top 10 customers by revenue this month” becomes a live query — no SQL, no filters, no facets.

Reads your data via MCP

Connects to a read-only view of your database or analytics API and queries it at question time.

Read-only by design

The MCP server exposes only the queries you allow. Credentials and raw records stay in your environment.

Explains the answer

Not just the figure — what moved, where, and compared to when, in language anyone can act on.

Seconds, not sprints

Answers arrive while the meeting is still going — not in next week's reporting cycle.

Knows its limits

Ambiguous or out-of-scope questions get a clarifying question or a clean handoff — not a made-up number.

Launch your insights agent
in hours.

Four steps from locked-up data to answers for the whole team.

01

Create the agent

Name it, set its scope in a short system prompt, and pick a model — GPT-4o, Claude, or Gemini. No code.

02

Connect your data, read-only

Expose a read-only view of your database or analytics API as an MCP server. The agent discovers the query tools; raw records never leave your environment.

03

Teach it your metrics

Define what “signups”, “churn”, and “active” mean in the prompt, then sanity-check answers against known numbers in the Playground.

04

Roll it out to the team

Everyone asks in the AI Workspace, or you embed the agent in internal tools through one REST API. Every query is traced for tokens, latency, and cost.

Embed the widget
<script
  src="https://app.sentientone.ai/widget.js"
  data-agent-id="your-agent-id"
  data-style="bubble">
</script>
Or call it from your app
curl https://api.sentientone.ai/v1/chat/stream \
  -H "Authorization: Bearer $SENTIENTONE_KEY" \
  -d '{
    "agent": "insights-agent",
    "message": "How did Q2 signups compare to Q1?"
  }'

Plugged into
your data stack.

Bring your metric definitions, connect your data sources read-only, and put answers where decisions happen.

Knowledge sources

  • Metric definitions
  • Data dictionaries
  • Reporting guides
  • Policy documents

Connect via MCP

  • Postgres & warehouses
  • Analytics APIs
  • CRM & billing data
  • Your internal APIs

Deploy to

  • AI Workspace for your team
  • Internal tools via REST API
  • Existing admin panels
  • Any channel, one API

Questions, answered.

Can it write to my database?

No. You expose a read-only view through the MCP server, and the agent can only call the query tools you define. Writes aren't possible by construction.

How do I trust the numbers?

Answers come from live queries against your own data, and every request is traced — you can see exactly which tool ran and what it returned. Start by sanity-checking known numbers in the Playground.

Who on the team can use it?

Anyone you invite. Non-technical teammates ask in the AI Workspace; developers can embed the same agent in internal tools through the REST API.

What data sources work?

Anything you can put behind an API or MCP server — Postgres and warehouses, analytics platforms, CRM and billing systems, or your own internal services.

Does our data leave our environment?

The MCP server runs on your infrastructure and returns only query results. Credentials and raw records stay with you — and you can self-host the whole platform.

How long does it take to set up?

Most teams are asking real questions within days: create the agent, connect a read-only view over MCP, define your metrics, and invite the team.

Give everyone
an analyst on demand.

14-day free trial · No credit card required · Cancel anytime