Tools and Integrations
MouseCat agents take the same investigative steps your analysts do: they write and run code, query your warehouse, and pull evidence from your internal systems and vendors.
The Toolset.
Code execution
Agents write and run Python in an isolated sandbox with no network access, to pivot on data, compute aggregates, and test a hypothesis against real numbers.
SQL
Agents write real queries against your warehouse. The tool schema is generated from your table registry, so new tables become available to the agent automatically. Read-only by construction; a failed query comes back to the agent to correct.
Evidence lookup
Every fact an agent relies on is retrieved through a store that records what was looked up and when — which is what makes the citation trail real rather than reconstructed.
Your vendors and internal APIs
MouseCat integrates with the data providers and internal services your team already relies on, inside your network.
Browser use
Agents navigate vendor portals and internal web tools the way an analyst does — for the evidence that lives behind a login rather than in a table.
import pandas as pd
# Transfers on the account in the 24h before the flagged payment
tx = df[(df.account_id == acct) & (df.ts > flagged.ts - pd.Timedelta("24h"))]
# Minutes between the payee being added and its first payment
tx["age_at_use"] = (tx.ts - tx.payee_created_at).dt.total_seconds() / 60
baseline = history[history.account_id == acct].age_at_use.quantile(0.05)
print(f"first use of payee: {tx.age_at_use.min():.1f} min")
print(f"90d baseline p05: {baseline:.1f} min")
first use of payee: 2.8 min 90d baseline p05: 41.0 min
Get started.
We'll walk you through MouseCat live.