Frequently Asked Questions About Nick Orgo Solo AI Agent Business Playbook
20 answers covering everything from basics to advanced usage.
// Basics
What exactly am I selling in a solo AI agent business?
You're selling a digital employee — a vertically specific AI agent that knows the client's business, handles work autonomously, and improves weekly. You are not selling Claude Code, ChatGPT, or 'access to AI.' You package it as an unlimited offer at $5,000/month covering unlimited agents, usage, monitoring, support, security, and changes. The deliverable is a business outcome, not software or hours saved.
Why should I never mention tokens or credits to clients?
The moment clients start counting usage, the magic disappears and time-to-yes slows. Tokens, credits, and infrastructure are technical framings that turn a digital employee into a metered utility. Clients then optimize usage instead of value. Position everything as an unlimited digital employee; keep the token math to yourself as private margin protection, since most clients only ever run one to three agents.
What is the Executive Abstraction Layer?
The Executive Abstraction Layer is the universal set of pain points every decision-maker shares regardless of industry: too many emails, too many meetings, too many follow-ups, too many open loops, and context scattered across too many projects. You lead every discovery call with these universal executive pains, then layer in vertical-specific problems like case management for law firms or lead follow-up for real estate.
// How To
How do I onboard a new client in under 48 hours?
Create an Orgo workspace, spin up a cloud VM, and use an existing Hermes agent or Claude Code to deploy the new agent. Install the core stack — Composio, Agent Mail, and a template Obsidian vault pre-loaded with your vertical's people/project structure. Focus the first 48 hours on the one or two highest-pain tasks and ship a working agent. Templating the vault is what makes sub-48-hour onboarding repeatable.
How do I build a content presence for my AI agent business?
Post content demonstrating agent capabilities to your target vertical before you have paying clients, so prospects already know you when they book a call. Use AI to assist research and editing. Drive traffic to a simple website plus Calendly link. This warm-audience-first approach is the highest-leverage acquisition activity — content is 'overpowered in 2026' and compounds, unlike cold outreach that starts every call from zero.
How do I manage multiple clients as a single operator?
Run a master Hermes agent connected to all client Orgo workspaces via Orgo MCP as your operations manager. From Telegram, send long-horizon tasks like 'fix the gateway on Client X's VM' or 'upgrade Client Y's Obsidian vault' while walking, sleeping, or in meetings. Query all VMs at once ('how many Orgo VMs do I have across workspaces?'). This is the fulfillment engine of the one-person business.
How do I set up watchdogs and observability?
Instruct each agent to run a gateway watchdog that auto-restores crashed connections (Hermes handles this more reliably than OpenClaw). Configure agents to email you from their Agent Mail address whenever a cron job, skill, or gateway fails. You resolve problems before the client notices, delivering on the unlimited support promise. Set this up before anything breaks — observability before crisis, not after.
// Troubleshooting
My agent feels generic and forgets everything — what's wrong?
You're likely missing an Obsidian vault. Agents without structured context feel generic and forget the client's business between tasks. Build a markdown vault storing the client's people, projects, workflows, and decisions, and auto-sync meeting notes from Granola into it. The more organized the vault, the closer the agent gets to personal AGI for that client — the vault is what makes it truly know the business.
I'm drowning in client requests — how do I fix this?
You skipped scope control. Give each client a customer-facing Trello Kanban board (Backlog → To-Do → Doing → Done) and enforce a strict limit of one to two requests per 48-hour cycle. Meeting notes auto-populate the Backlog; the client drags into To-Do; you pull only what fits the cycle. Send Loom videos of improvements as delivery receipts to demonstrate value without live calls.
I keep pivoting between niches and can't build systems — what should I do?
You're stuck in the Diverge phase too long. Diverge Then Converge means testing two or three verticals for a bounded 30–60 day window, then committing fully to the one with the strongest market pull. Constant pivoting prevents the focus needed to build repeatable delivery systems and template your Obsidian vault. Set a hard deadline to converge, then niche down by geography or specialization.
Why shouldn't I use local Mac Minis for hosting agents?
Local Mac Minis make remote management impossible at scale and carry too much blast-radius risk. Use cloud VMs instead — they're remotely manageable, sandboxed, deletable, and scalable. Each client gets an Orgo workspace, and each agent gets its own isolated cloud VM as its home. This is what lets your master Hermes agent manage everything from Telegram anywhere.
// Comparisons
What is the difference between Hermes and OpenClaw?
Hermes is Nick's preferred agent harness — self-evolving, model-agnostic (swap models without changing infrastructure), and more gateway-stable than OpenClaw. It can take long-horizon tasks via Telegram. OpenClaw is an earlier, now-commoditized harness with less stable gateways; it's adequate for $5K/month positioning but not preferred. Gateway stability matters because crashed gateways cause client-visible downtime.
How does this playbook compare to a traditional agency model?
Unlike a traditional agency that scales by hiring people, this playbook scales by using agents to build and manage other agents, keeping it a one-person operation. Traditional agencies sell hours and deliverables; this sells a digital employee with an unlimited offer. Margin comes from most clients needing only one to three agents, and fulfillment leverage comes from a master Hermes agent, not headcount.
How does selling business outcomes differ from selling time saved?
Time saved is overused, and clients are immune to it. This playbook translates agent capabilities into revenue generated and concrete business outcomes instead — for example, 'the agent follows up on every stalled deal' rather than 'saves 10 hours a week.' Outcome framing commands higher prices and resonates because decision-makers buy results, not efficiency abstractions.
// Advanced
Should I start with a broad industry or a narrow sub-niche?
Start broad enough to test (marketing agencies, law firms, insurance), then go super niche once the market pulls you. Diverge Then Converge: after finding your winning vertical, niche down by geography ('real estate agencies in Florida') or specialization ('matrimonial law firms'). Narrow sub-niches let you pre-template an Obsidian vault, speed onboarding, and use one client as a wedge to infiltrate the broader market.
What is Composio and why does every agent need it?
Composio is a single MCP connector that links an agent to thousands of apps — Gmail, Slack, Notion, GitHub, and more — while handling authentication and tool calling. Every agent gets it because it removes the biggest security and setup friction in deployment. Instead of wiring up each integration manually, one connector gives the agent broad, authenticated reach across the client's tools.
How do I use sub-agents to configure a new client agent faster?
During deployment, give your setup agent access to Perplexity MCP (current Hermes docs), Context 7 (GitHub docs), Exa AI (real-time web search), and XMCP (Twitter/X setups). Spawn sub-agents per source to pull best practices in parallel, then have the setup agent apply them inside the Orgo VM. This parallel research-and-configure loop is the core of the Agents Build Agents principle.
How does Granola fit into the fulfillment loop?
Granola is a meeting notes tool with an MCP that closes the loop between conversations and agent context. Its notes auto-sync into Trello as task cards (populating the client's Backlog) and into the client's Obsidian vault (keeping the agent's context current). This automation means client conversations become actionable work and structured memory without manual data entry on your part.
What are vertical-specific skills and how often should I ship them?
Skills are vertical- or client-specific capabilities layered on top of the universal executive stack — for example, demand letter drafting for law firms or lead follow-up sequences for real estate. Ship them one to two per 48-hour window via Trello to control scope. Keep the first 48 hours focused on the highest-pain tasks, then add skills incrementally as delivery receipts via Loom.
Can I really run eight clients as one person with this model?
Yes — the example of an operator managing eight client VMs from a single Telegram chat shows how. The key enablers are a master Hermes agent connected via Orgo MCP, cloud VMs per agent, watchdogs preventing client-visible downtime, Trello scope control, and a pre-templated Obsidian vault for sub-48-hour onboarding. Fulfillment leverage comes from agents building and managing agents, not from hiring.