Running a full marketing operation solo with 40 AI agents

Executive overview

Most professionals use AI for roughly 1% of its potential. Jacob Bank, founder of Relay.app, replaced a $50,000/month marketing contractor team with 40 AI agents costing $500/month — while remaining the sole marketing person at a nine-person company.

The "AI as intern" model undersells what's possible. AI handles not just execution tasks but strategy, coaching, and competitive analysis. The right mental model is the Super IC: two-thirds individual contributor work, one-third coordinating AI agents.

Building AI agents is the defining professional skill of the next 30 years — as foundational as spreadsheets were for the last 40.

Why "AI as intern" is the wrong mental model

• Interns handle execution; AI also does strategy analysis, competitive research, and content creation
• An AI sales coach reviewing call transcripts costs $5/week vs. $10,000/month for a human coach
• The coaching model works for any new domain: give it raw data, set the cadence, get structured feedback

How to build an agent team

• Start with one agent doing one thing; add agents one at a time
• Do not attempt a single agent handling 25 tasks — specialist agents outperform generalists
• Use a meta-agent to invoke the right specialist agents when needed
• Agents require ongoing maintenance — they are not set-and-forget
• Fire agents when the function is no longer needed; there is no emotional baggage or coordination cost

What the Super IC future looks like

• Two-thirds of the day: hands-on work — editing, publishing, talking to customers
• One-third of the day: coordinating the AI agent team
• Junior roles lose pure-execution work; management roles shrink as companies get leaner
• Every role will require both strategic judgment and hands-on output

Career risk, reframed

• The riskiest career path is stagnation inside one large company
• Breadth of network, new skills, and varied experience are more durable than cash compensation
• Optimise for personal growth and learning — uncertainty accelerates both
• Embracing AI early is the low-risk move; ignoring it is the high-risk one

Skills that will remain durable

• Clearly articulating what needs to be done — directing AI effectively
• Building genuine social connection and personal presence
• Logic and structured thinking for giving precise instructions to AI systems