Generative AI that stays inside your walls

agodonnell builds generative AI on your own documents and data, inside a secure, governed platform. The documents, the search index, the models, and the application your team uses all stay within your control.

Built for legal, healthcare, financial services, and other organizations where confidentiality is not optional.

How it works Background

Where Private GenAI Creates Advantage

Ask questions across thousands of documents and get answers with citations
Find clauses, precedents, and prior work in seconds instead of hours
Keep privileged and regulated data inside a governed boundary
Replace scattered AI subscriptions with one controlled platform
Give non-technical staff a simple tool they will actually use

The model is the easy part. The value is in governed, trusted data.

How It Works

Most firms already see what AI could do. The blocker is trust: where the documents go, who can see them, and whether the answers can be checked.

Step 1

Assess

Identify the document collections and questions that matter most, and define the security and access rules they require.

Step 2

Secure the Data

Bring documents into a governed lake on Databricks, with access controls and audit trails through Unity Catalog.

Step 3

Build the Assistant

Index documents with vector search and connect models served inside the platform, so every answer comes from your own sources, with citations.

Step 4

Put It to Work

Deliver a simple application for staff, measure answer quality, and improve it as usage grows.

Technology is the implementation layer. The goal is faster, more confident work on information you already own.

Who This Is For

Mid-market organizations with large document collections and strict confidentiality requirements: law firms, healthcare organizations, insurers, financial services firms, and professional services.

Thousands of documents that are hard to search and easy to lose track of
Confidentiality obligations that rule out sending data to public AI tools
Staff already experimenting with AI on their own, without controls
No in-house team to build and govern an AI platform

Every engagement is built around depth of impact, not breadth of clients.

Beyond Documents: Market Signals

The same governed platform can combine your internal data with external market signals, so leadership sees risks and opportunities before they show up in financial results. The approach is familiar to macro traders; the difference is applying it to the operating decisions of a mid-market business.

Map input cost exposure against commodity prices, freight rates, and FX
Spot margin pressure building before it reaches the financial statements
Use AI to turn signals into plain-language briefings leadership can act on

Unified data first. Intelligence on top.

About

agodonnell is the independent practice of Aaron O'Donnell.

I started in finance, supporting cross-border relationships across Latin America at Merrill Lynch and JPMorgan Chase. Since 2009 I have built analytics and data platforms for large federal programs, where security, auditability, and data quality are requirements, not preferences.

Today I focus on bringing generative AI into those same governed environments.

Common thread: trusted data, clear answers, and outcomes you can measure.

Credentials

Databricks Certified Generative AI Engineer Associate
Databricks Certified Data Engineer Professional
Databricks Certified Data Engineer Associate
MS Data Science, Northwestern University
BBA Finance, University of Texas at Austin (McCombs)
Bilingual English and Spanish

Endurance Efforts

Proud to always work for the PR.