Featured Guests
Christopher Parsons, Founder and CEO, Knowledge Architecture
Webinar Summary
Knowledge agents are emerging as one of the most promising applications of AI for AEC firms—helping teams accelerate learning, scale expertise, and streamline workflows.
In this webinar, Christopher Parsons, Founder and CEO of Knowledge Architecture, will provide a live demonstration of Synthesis Knowledge Agents, showing how agents are created, refined, and deployed to solve real business problems.
Along the way, Christopher will share what we're learning from firms participating in the Synthesis Knowledge Agents private beta, highlighting the patterns and strategic insights emerging from early adoption.
We'll explore:
A live demonstration of Synthesis Knowledge Agents, including how to create and refine agents from scratch
Real-world examples spanning informational, creation-based, and process-based use cases
The progression from informational AI assistants to workflow-transforming knowledge agents
The common characteristics shared by firms making the fastest progress with knowledge agents
How firms are capturing and structuring knowledge to power more capable agents
Where knowledge agents are headed and what that means for AEC firms
This session offers a practical look at how knowledge agents are evolving and what early adopters are teaching us about the future of AI-enabled work in AEC firms.
Enjoy!
Webinar Timeline
Synthesis Knowledge Agents
00:00 Welcome and Introduction
02:29 The Agent Capability Spectrum
06:55 Product Demo: Overview of Synthesis
09:28 Level 1 Demo: Informational Agents
18:31 Level 2 Demo: Creation-Based Agents
23:34 Level 3 Demo: Process-Based Agents
29:22 Building an Agent Live
35:06 Agent Analytics & Feedback
36:27 Insights from the Private Beta
40:37 The Case for Leaders Building Agents
41:41 What's Next: Agent Analytics, Search Connector API, Synthesis MCP Server & Semi-Autonomous Agents
Q+A
50:39 Will there be any standard agents or agent templates released by KA alongside the functionality?
52:04 How does the agent access specific project details? Does this data need to be included in the project data in OpenAsset/Deltek?
52:26 Do agents employ computer vision in any capacity for visual items? For example, can I upload a PDF set and have it checked against symbology and other standards?
52:59 If an end user creates an agent, can a global admin or agent admin consume it and make it a global agent?
53:29 Are there templates for agent output like the NDA comparison table so they always come out the same?
54:11 Can we configure agents to also search Autodesk Revit help or other software?
55:59 Are there costs associated with these features?
55:17 Are the agents available under the basic AI search offering?
55:30 Can the L&D assistant create true/false or multiple-choice questions with correct answers based on a learning course to create an assessment?
56:14 Can we create a content governance assistant?
56:28 Are documents uploaded to an agent viewable to KA?
57:01 Can we disable certain users from being allowed to build agents?
57:24 Is there a suggested preferred order of operations between setting up agents and developing LMS if we're not yet part of the LMS beta?
58:36 How are knowledge agents changing what companies decide to put into their LMS? Will the LMS eventually become obsolete?
59:46 As these agents mature, how do you prevent memory staleness, and how do you handle controlled writeback so the system stays trustworthy without polluting the source of truth?
1:01:05 Where does the cost of tokens reside?
1:03:09 Is there persistent conversation knowledge for users and teams — in other words, memory for users or projects like Claude or ChatGPT?
1:04:32 Can agents access content optimization and analytics?
