Featured Guests
Christopher Parsons, Founder and CEO of Knowledge Architecture
Sean Mahan, Director of Software Engineering, Knowledge Architecture
Webinar Summary
Ever wonder how AI solutions actually work?
In this behind-the-scenes session, Christopher Parsons (Founder and CEO) and Sean Mahan (Director of Software Engineering) share how Knowledge Architecture designed and built Synthesis AI Search, tracing its evolution through multiple generations of design and engineering. You’ll see the real problems they set out to solve, the obstacles they overcame, and the lessons they learned along the way.
This is a deep dive for the curious—ideal for anyone who wants to understand how AI retrieves, reasons, and improves over time. We explore architecture diagrams and code concepts in plain language, connecting each layer of the system back to design strategy and business outcomes.
By understanding how AI search works, you’ll also gain insight into how to capture and structure knowledge more effectively, helping your firm better connect the right person to the right knowledge at the right time.
We close with a look ahead at how Synthesis AI Search is laying the foundation for knowledge agents—AI-powered assistants that will help your firm make better use of its collective knowledge, streamline workflows, and accelerate learning.
Enjoy!
Webinar Timeline
Introduction
00:00 Overview of Synthesis + Knowledge Management
03:09 As Little AI as Possible
PART 1: General Knowledge Base Search
05:03 Goal: Improve User Experience for General Knowledge Base Searches
09:02 Design Approach (v 0.5): Retrieval Augmented Generation (RAG)
10:28 Rapid Prototyping
13:50 Design Appoarch (v 1.0): Hybrid RAG + Content Chunking
18:50 Context Engineering
21:20 Comprehension Upgrades
PART 2: Advanced Employee Search
27:10 Goal: Connecting People to People
31:28 Design Approach (v 1.1): Agentic RAG + Query Tree
37:37 Employee Filter Query Tool
40:44 Additional Context
44:44 Testing
PART 3: Advanced Project Search
45:14 Goal: Connect People to Projects, Clients, and Consultants
46:38 Design Approach (v 1.2): Agentic RAG + Disambiguation
52:00 The Lookup Tool
53:26 Relationship Filter Query Tool
56:08 Calculations + Tools for Math
Part 4: What’s Next?
59:10 Synthesis LMS Search, More AI?, Composability
1:02:16 In Design: Knowledge Agents
Q+A
1:08:10 Can using certain keywords in the prompt help steer the AI to a specific tree?
1:11:04 Are you introducing MCP for tools?
1:12:58 Do you provide an open API for custom integrations?
