Skip to content
woodfrog
AI
Companies with their data already on DatabricksUse case on Databricks, no client result claimed

AI agents that work on the data you already trust.

Your data is on Databricks, governed and clean. The next step is putting it to work: agents that watch the numbers, draft the follow-up and ask a person before anything goes out. We build them on Claude, on top of Unity Catalog.

Sample deliverable, illustrative data
Collections agent drafted 14 reminders from the Gold layer. 3 need your approval.
Reminder to Sharma Traders, ₹4.2 lakh, 38 days overdue
Reminder to Kiran Foods, ₹1.1 lakh, 45 days overdue
Credit hold for Apex Retail, above limit
Governed tables in Unity CatalogClaude agents with tools and guardrailsDrafts and alerts, approved by a person
RegisteredDatabricks partner, in Pune
CertifiedDatabricks practitioners on the team
50+Data projects delivered by the team
Week 1Fixed-fee audit
TypeUse case: what we build on Databricks, not a client story
ForCompanies with data on Databricks that want AI to act on it, safely
EngagementWeek 1 audit, then build, fixed scope, fixed fee
Runs onDatabricks on your cloud with Unity Catalog, Claude from Anthropic, MLflow for tracking and evaluation
Before and after

From the old way to the new.

Before
  • Reports that show a problem, then wait for someone to act
  • AI pilots that run on copied data outside the lakehouse
  • No way to check what an AI answer was based on
  • Security teams that block AI because access is unclear
After
  • Agents that act on the numbers and draft the next step
  • Agents that read governed tables, with the same access rules as people
  • Every answer traced to the tables and rows it used
  • A person approves anything that matters before it goes out
The problem

The data is ready. The AI pilots are not.

Many teams on Databricks have clean, governed data and a backlog of AI ideas. Pilots get built on exported data, outside the access rules, and stall when security asks what the AI can see and how its answers are checked.

A hand pointing at a chart on a tablet
The data is ready. Nobody has time to act on all of it
What was broken
  • AI pilots copy data out of the lakehouse to work.
  • Nobody can show which rows an answer came from.
  • Access for the AI is broader than for any person.
  • There is no test set, so quality is a matter of opinion.
  • Pilots never reach production.
You might be here if
  • You have a list of AI ideas and no pilot in production.
  • Security or audit asks hard questions about AI access.
  • Reports flag problems that nobody has time to chase.
  • You want AI that uses your own numbers, not general knowledge.

Why it mattered. Each month an agent is not in production, the repeat work it could take stays with your team.

The approach

Govern first, then build the agent, then test it before launch.

We give each agent its own identity in Unity Catalog, with access no wider than the job needs. The agent is built on Claude with tools that query governed tables. Before launch it is tested against a set of real questions, and after launch every run is logged and reviewed.

Where it is today
Governed tables
A backlog of AI ideas
Repeat work on people
woodfrog method
Scope the accessIts own identity, least access
Build on ClaudeTools on governed tables
Evaluate before launchThen monitor every run
What changes
Agents in production
Answers traced to rows
People approve
AI on your lakehouse, approved
Deliberately not doneAn agent that acts alone on anything that matters. A person approves every action with an effect outside the system.
We are an Anthropic Certified Services Partner and build Claude agents the way Anthropic recommends.schematic
The solution

An agent your auditors will sign off.

What we build
  • One agent for a real repeat task, chosen with you in the audit.
  • Its own identity in Unity Catalog, with access only to the tables it needs.
  • Tools that query governed tables, so every answer cites its rows.
  • An evaluation set of real questions, run before every release with MLflow.
  • Guardrails: rule checks on inputs, a confidence score and a stop button.
  • Run logs and a weekly review, so quality is measured, not assumed.
Claude runs through your cloud's approved route, and your data stays in your Databricks account. Every agent action is logged in the lakehouse, next to the data it used.
How it runs
  1. Week 1AuditData profiled, scope and fee fixed
  2. ThenBuildOne area at a time, tested with your users
  3. OngoingOperateQuarterly reviews of quality, access and spend
The impact

What we would measure.

What we would measureNo client values yet
Tasks handled by the agentPer week
Answers passing evaluationBefore each release
Actions approved first timeShare of drafts
Hours of repeat work movedMeasured against your baseline
Incidents and stopsLogged and reviewed
Put your Databricks data to work with AI.A short call, or record a message. A person replies within one business day.
Contact usCertified Services Partner, Claude Partner Network