Sunny Ray talks with Jeremy Rob, founder of RobClan and creator of Wayfinder, an AI-first learning management platform built for the agentic era. Jeremy explains how Wayfinder spans a full spectrum, from manually built courses to a built-in five-agent system that generates learning paths and certifications, to full MCP integration that lets teams connect their own AI assistants like OpenClaw or Claude Code to create, evaluate and manage training content. He describes Wayfinder Context, a three-tier system of corporate, role-based and private personal context files that let AI agents work consistently and privately with each employee. Jeremy shares his path from a history degree through eBay, the University of Utah, ServiceNow and Abnormal AI, shaped by raising two autistic sons and his own late autism diagnosis, which fueled his focus on enablement and resilience. He positions Wayfinder as more than an LMS, calling it a learning operating system designed to shrink technical debt and prepare organizations for a future where AI agents become the primary work interface across sales, HR, customer success and partner teams.
RobClan's Jeremy Rob unveils Wayfinder, an AI-first learning OS that turns MCP-connected agents into your team's primary work interface.
What's the brand behind what you're building, and what is Wayfinder?
RobClan is the company and Wayfinder is our primary product, an AI-first learning management platform. It spans from manually building courses to using a built-in AI agent to generate them, to connecting your own AI assistant via MCP to create, manage and grow your platform, moving toward a future where an AI agent becomes your primary work interface.
Who is the ideal customer for Wayfinder, HR, leadership, or someone else?
It's definitely the leadership team. Wayfinder serves HR, sales enablement, customer success and partner enablement teams alike, replacing the fragmented mess of five or six different LMS platforms each with separate licenses and limited seats. The idea is to flatten technical debt onto a single AI platform that serves everybody well through the interfaces they're already using.
Can you give a one-liner for Wayfinder, like OpenClaw for companies?
I'd say learning for OpenClaw is probably the best way to put it. OpenClaw is its own thing, and with what I'm doing, Wayfinder is very much a supplement to everything else you're doing. It becomes the central repository, the central workspace, for learning.
Take me back, what got young Jeremy into this world before training and AI?
I was a history major studying ancient Roman and Greek history, then worked in tech as an NEC technician, and later at eBay, which shaped my sense of what a great company looks like. I got my master's in adult education and distance learning, taught at the University of Utah's School of Computing, then moved into training roles at ServiceNow and Abnormal AI before starting Wayfinder.
Is what you're building similar to the soul.md file concept from OpenClaw, but for work?
In agents like OpenClaw, the soul.md file gives the agent long-term memory and defines how it works with the user. I generate a similar AI knowledge.md file tied to a person's learning path and role, so the agent knows how that job affects its support, like a customer success manager needing reports ready and tier-based recommendations for every call.
Who creates this context, the employee or company leadership, and how is ownership handled?
Corporate-level context is created by leadership, usually a tenant admin, covering company ethics, values and general guidelines available to everyone. Role-based context comes from the training and task analysis built for that role. Personal context is set by the individual, locked down so only they see it, and it's downloadable so it travels with them to a new job or role.
Most AI tools are content generators. Why call Wayfinder a learning operating system instead?
You can build manually like any LMS, or use our built-in five-agent system to generate learning paths, course outlines and certification exams as a stepping stone. The real magic is connecting your own AI assistant via MCP so teams like engineering, marketing and support can each evaluate and edit content with their own context in about two minutes, then you approve and publish.
Are you using RAG, and how do you prevent hallucination since companies are wary of AI at work?
The agent's content generation relies one hundred percent on the resources you provide, like the role name, description and any resource files or links you upload. In my demos I pull real job descriptions straight from LinkedIn and drop them in, and if you have a task analysis you provide that, otherwise you supply what you have so the output stays grounded.
Building something daring? Sunny talks to founders like this every day. Fifteen minutes to see if your story belongs on the stage.
Claim your pre-interview