Bob Pulver sits down with Jacob Bank, Co-founder and CEO of Relay.app, whose career arc — from Stanford's Multi-Agent Systems Lab to founding Timeful (acquired by Google in 2015) to leading Gmail and Google Calendar product teams — represents one of the most continuous threads in AI agent development. Jacob frames AI agents not as software to configure, but as employees to hire, coach, and manage, arguing that great people managers are naturally suited to the AI era. He maps out a three-tier AI stack everyone should adopt and explores how knowledge work will be restructured, why AI literacy is non-negotiable, and how small businesses can now compete at scales once unimaginable.
Keywords
Jacob Bank, Relay.app, AI agents, agentic workflows, autonomous workers, workflow automation, small business, AI literacy, people management, Timeful, Google Calendar, Gmail, knowledge work, G&A, go-to-market, responsible AI, human-in-the-loop, SaaS evolution
Takeaways
-
The right mental model for AI agents is employee management: give them a job description, set expectations, provide feedback, and apply the same code of conduct as any team member
-
Everyone needs three AI tools: a chatbot for conversation, a copilot for real-time task delegation, and an autonomous agent platform for proactive, repeatable work
-
Relay runs on 9 humans and ~60 AI agents — and Jacob sees a path to serving 100x more customers with roughly the same team size
-
AI levels the playing field for small businesses, enabling work at a scale previously only achievable by much larger organizations
-
Jacob's three-level delegation progression: tasks you already do, tasks you're capable of but never have time for, and tasks you'd otherwise hire an expert for
-
AI literacy is not optional — it's becoming a baseline requirement for effective work, equivalent to basic computer literacy
Quotes
-
"We're all managers now — that is the skill set we need."
-
"If you have a job that is just to write the blog post about X, that job is not going to exist anymore."
-
"It's not optional. This is going to be a requirement of being an effective worker in the future."
-
"Whenever I have an AI agent doing a classification task, I always ask the AI to explain its rationale — because then you can correct it for next time."
-
"At some point you'll cross this tipping point where you don't have to tell yourself to go use AI — it'll suck you in."
Chapters
00:02 Welcome and introductions
00:56 Jacob's origin story and agent-oriented programming
02:54 From Timeful to Google
04:41 Pre-LLM AI features in Gmail and Calendar
06:15 AI coworkers vs. productivity tool nudges
07:39 Early agent research and org disruption
09:24 Restructuring knowledge work
11:45 Evolving human roles and AI literacy
13:32 The social complexity of scheduling
15:16 Credentialed jobs at risk
17:24 AI leveling the playing field for small business
18:17 Inside Relay — 9 humans and 60 agents
19:41 The three-tier AI stack
22:38 Relay as intelligent workflow automation
23:42 SaaS selection in the agent era
26:47 Platform consolidation and SaaS business models
28:13 Deploying agents across G&A, GTM, and R&D
33:16 Agent collaboration and human oversight
34:21 When to build vs. buy
37:56 Three levels of AI delegation
39:50 Scaling AI readiness across organizations
42:22 Responsible AI and the employee management lens
44:14 Evaluating agents vs. testing software
45:51 The blast radius problem
48:09 Bias, coachability, and correcting agents
49:29 Closing advice — go one step further
50:45 What's next for Relay
Jacob Bank: https://www.linkedin.com/in/jacobbank
For advisory work and marketing inquiries:
Bob Pulver: https://linkedin.com/in/bobpulver
Elevate Your AIQ: https://elevateyouraiq.com
Substack: https://elevateyouraiq.substack.com
Powered by the WRKdefined Podcast Network.
Sponsors
