Agent Engineering for Enterprise Data

A hands-on, two-day workshop in which you build an AI agent that works with private enterprise data: it finds what it needs in company documents and business systems, weighs what it finds, and makes a recommendation you can check against its sources.

Bring it to your team (in-house) — we run the workshop on-site, tailored to your technology stack and business needs. The exercises use sample enterprise data we provide, so none of your company’s data is involved. Email us or call (925) 918-1769.
Want to attend a public workshop? We don’t have one on the calendar at the moment—email us and we’ll let you know as soon as we schedule the next one.
Not sure yet? Start with a free lunch & learn for your team. Email us to set one up.

Foundation models don’t know your business. Your company’s documents, policies, records, and internal systems hold the knowledge that makes AI useful to your organization. A chatbot that searches a few PDFs makes an impressive demo, but an AI system you can trust with real business decisions takes considerably more engineering, and goes well beyond retrieval-augmented generation (RAG).

What makes an agent on private data trustworthy is evidence: every claim traces back to a source, and when evidence is missing, the agent says so rather than guessing. In this workshop you build an agent around that idea. The running example is a vendor onboarding agent. It examines supplier documents, checks them against company policies, queries internal records, and prepares an assessment for a procurement manager that cites its sources and flags what it couldn’t confirm. The same techniques apply to customer support, internal knowledge assistants, compliance, and other document-heavy work.

What you’ll build, and learn

The workshop runs as a series of modules, each adding a capability to an agent that grows throughout the two days:

Foundation models and adaptation. What models know and don’t know about your business, and when prompting, context engineering, retrieval, fine-tuning, or tools is the right way to close the gap.
Working with private enterprise data. Real corporate data is PDFs, tables, and records whose metadata matters. How to extract and prepare it without losing what the agent will need to cite.
Retrieval engineering. Why finding the right passage is harder than it looks, and how chunking, embeddings, hybrid search, and reranking each change what the agent sees.
Grounded answers. Assembling context so that every answer is supported by source documents, and handling information that’s missing or conflicting.
From RAG to agents. Giving the model tools and business APIs, so it can investigate a question across several systems instead of answering from a single search.
Security and privacy. Enforcing who may see which data, defending against prompt injection, and the deployment choices that sensitive data calls for.
Evaluation. Measuring retrieval accuracy, answer quality, and agent behavior, so you know whether a change made the system better instead of judging it by a good demo.

Each module starts with a short talk; then you implement the new capability in Python, and we look together at the results. Choices about document processing, retrieval, prompts, models, and tools all change how the complete system behaves, and comparing them, and digging into the failures, is much of the point. The instructors circulate throughout the hands-on sessions to help you past the bumps.

You don’t need to be an AI or machine learning specialist. If you’re comfortable programming in Python and working with APIs, you have the background you need. We provide the development environment, sample enterprise data, and starter code, so you can concentrate on the engineering rather than setup. Your agent works on that sample data, not on your company’s data, so nothing proprietary goes to a model provider during the workshop. You’ll use existing foundation models through APIs, and we’ll discuss alternative deployment approaches.

You’ll leave with a working AI agent, reusable source code, and an evaluation toolkit in a repo of your own, ready to point at your company’s data when you choose, along with the engineering principles needed to build AI systems your organization can depend on.

Who it’s for: Software developers and technical architects, comfortable in Python, who want to build AI applications and agents on their company’s own data. No prior AI or machine learning expertise required.
Length: Two days
Format: In-house at your office, or as a public workshop

Bring this workshop to your team!

To bring this workshop to your team, email us at info@artima.com or call (925) 918-1769.

You may also be interested in one of our other workshops: