Key Takeaways

  • William Imoh, founder of Hackmamba, automated core company operations by building a multi-agent system in Grok Bot instead of hiring administrative staff.
  • To scale operations to oversee 25 people and chase a 100x revenue mandate, Imoh assigned operational personas mapped to characters from The Lord of the Rings.
  • The system runs an automated Sunday night operator review answering three questions: what to start doing, what to stop doing, and what to keep doing, plus the three highest-impact activities for the week.
  • By routing work through a central Chief of Staff persona rather than dumping all tasks into one general chat thread, Imoh keeps operational contexts separated.
  • Imoh operationalized this structure through The Fellowship Multi-Agent Org Chart.

The Fellowship Multi-Agent Org Chart

Chief of Staff ('Gandalf'): Acts as the primary orchestrator and system architect. Generates and manages all subordinate agent personas and coordinates overarching business routines.

Engineering ('Aon the Elf'): Oversees the technical codebase, handles architecture, and drives complex code implementations alongside frontier models.

Finance and Accounts ('Gimli'): Manages account receivables, balances, invoicing, and bookkeeping routines across financial tools like QuickBooks.

Sales and Recruiting ('Aragorn'): Recruits candidates, negotiates partnerships, and drives parallel inbound and outbound deal cycles.

Customer Success ('Samwise'): Monitors customer channels and support tickets, assisting clients with direct issue resolution.

Product Management ('Frodo'): Acts as the core product owner responsible for managing delivery, defining epics, and executing user journeys.

Growth and Marketing ('Legolas'): Drives content ideation, growth marketing playbooks, and multi-channel asset distribution.

When This Works (and When It Doesn't)

This setup works when you are a solo founder or small team running multi-functional operations without expanding administrative headcount, giving distinct persona scopes to automated agents. When Imoh set out to expand Hackmamba, he faced a clear operational bottleneck. “I put together a mandate internally and I called it 100x,” (play from 25:14) Imoh said. “This was a way for us to go 100x our current revenue goals and the way we're set up as a company without drastically increasing headcount.” (play from 25:20)

Context collapse kills most multi-agent experiments. When a founder gives one LLM window access to financial statements, outbound sales copy, and pull requests, the model mixes tone and drops constraints. Imoh avoids this by compartmentalizing responsibilities into fixed personas. The Chief of Staff builds and supervises the rest of the team: “I was first asked to create a chief of staff. I just called him Gandalf. Straight up. And then I asked Gandalf to create everyone else.” (play from 29:17)

Where this breaks down is edge-case governance and tool permissions. If an agent like Gimli executes live invoices or writes directly to QuickBooks without an approval gate, minor hallucinations turn into accounting disasters. Similarly, autonomous outbound recruiting from an Aragorn persona risks sending awkward or off-brand messages to candidates if it runs without human oversight.

What to Do With This

Pick the single functional role consuming the bulk of your administrative time this week. Build an agent specifically scoped to that role instead of throwing the tasks into your default chat history.

Set up a dedicated Chief of Staff bot in your workflow tool. Direct it to establish an automated Sunday evening review routine. Have it pull your open tickets, recent client messages, and calendar entries to produce an operator brief before your week begins. Specify the exact prompt Imoh uses: identify one thing to start doing, one thing to stop doing, one thing to keep doing, and the three highest-impact activities for the upcoming week.