Greg Isenberg Tiny AI Agent Business Builder

// When should I use the Tiny AI Agent Business Builder framework?

Use this skill whenever you want to start a small, boring, high-margin business using AI automation — especially when you have a vague idea, a niche you know, or just want a systematic way to generate and immediately prototype 'tiny AI agent business ideas' without writing code from scratch.

// What inputs do I need to start building a tiny AI agent business?

• Domain or niche of interest
  The market, asset class, or industry you want to hunt in (e.g. restaurant equipment, expired domains, mobile apps, SaaS businesses). Can be left open for the AI to suggest.
• Target monetization stylerequired
  How you intend to make money: flip, broker fee, retainer, relaunch, or sell-as-a-service. Determines which buyer and liquidity point to target.
• Budget ceiling
  Maximum acquisition or bid budget per asset (e.g. under $200 per domain, under $2,500). Sets the scoring filter for the agent.
• Delivery channelrequired
  Where you want the agent to surface deals — Slack channel, email, Telegram, etc.
• Exclusion criteria
  Any asset types, histories, or categories to filter out (e.g. no adult or gambling history on domains).

// What are the core principles behind the Tiny AI Agent Business Builder?

// How do you apply the Tiny AI Agent Business Builder step by step?

1. 1

   Choose your idea lens and identify the feed

   Pick one of three lenses: (a) places with constant change — job boards, auction sites, app store rankings, court filings, marketplaces; (b) neglected assets — expired domains, dead Product Hunt launches, fallen app-store rankings, distressed restaurant equipment; (c) competitive intelligence — competitor pricing pages, changelogs, founder tweets, job postings. The feed must be public, machine-readable, and update frequently.

2. 2

   Write your one-liner using the Feed → Asset → Trigger → Buyer → Monetization chain

   Format: 'Monitor [feed source] for [asset type] that meet [scoring criteria], flag when [trigger event], deliver ranked list to [channel] so I can [monetization method] to [buyer type].' Example: 'Monitor expired domain drops and GoDaddy auctions for domains with DR 20+, clean backlink profile, no adult/gambling history, under $200 — deliver 10 picks every morning to Slack — flip to newsletter operators or SEO agencies.' This one-liner is your entire brief to the AI agent.

3. 3

   Paste the one-liner into your AI agent tool and answer its clarifying questions

   A well-configured agent will ask: What are your niche keywords or categories? What is the delivery channel and format? Which data sources need credentials or API keys? What is the scoring model? Answer these concisely. If you don't have a preference, say 'you pick the best ones' — the agent will make reasonable defaults. Do not over-specify upfront; iterate later.

4. 4

   Set up your delivery channel (Slack webhook or equivalent)

   Create a dedicated channel per business idea — do not mix multiple agent outputs into one channel. In Slack: create new channel → go to api.slack.com/apps → add a new webhook → paste the URL into your agent config. Ask the agent to walk you through this if unfamiliar; it will produce step-by-step instructions.

5. 5

   Run the agent and review the first deal card batch manually

   Let the agent scrape, score, and surface its first 10–15 results. Inspect every field: spread calculation, buyer value estimate, contact info, outreach draft. Look for data bugs (broken links, HTML entities in emails, wrong price units). Do not automate outreach or purchasing until you have validated at least one full cycle of output quality.

6. 6

   Talk to the agent in plain language to fix bugs and expand scope

   Treat every correction as a conversation: 'I noticed the links in the cold emails look like this — fix it so no HTML entities bleed through.' Or: 'Increase the budget ceiling to $2,500.' The agent updates its own logic. You do not need to rewrite the original prompt. This is the core interaction loop.

7. 7

   Apply the three quick-screening questions before scaling

   Before enabling the agent to run autonomously (overnight, daily heartbeat), verify: (1) Is there urgency? — the asset or deal will be claimed by others if not acted on quickly. (2) Is there spread? — the gap between acquisition cost and resale/broker value is large enough to be worth acting on. (3) Who pays first? — you can identify a specific, reachable buyer before you commit capital or time.

8. 8

   Enable overnight / scheduled runs and configure the heartbeat

   Once output quality is validated, turn on 'prevent sleep' so the agent stays active. Enable 'heartbeat' (checks for pending events every 30 seconds) only once the business is generating revenue — it consumes tokens. Set the agent to deliver its ranked brief at a fixed daily time so you wake up to actionable deal cards.

9. 9

   Choose and lock in your liquidity point

   Pick exactly one monetization method per agent idea: (a) Flip — buy asset cheap, resell at market value; (b) Broker — connect seller and buyer, charge 15–30% fee, zero inventory risk; (c) Retainer — sell the daily intelligence brief as a subscription service (e.g. competitive intelligence for $9.99/month — 'agents are the new SaaS'); (d) Relaunch — acquire a dead asset, improve monetization, grow again. Mixing methods in one agent creates confusion; build separate agents per model.

10. 10

    If selling as a service, vibe-code a landing page and create a deal card as your sales artifact

    Ask the agent to generate a landing page based on a reference style. The deal card (used market value, acquisition price, spread, broker fee) is your core sales artifact — it is what you send to the obvious buyer. The landing page is what you send to inbound prospects who want this intelligence delivered to them regularly.

// What are real-world examples of tiny AI agent businesses?

// What mistakes should I avoid when building a tiny AI agent business?

• Skipping the one-liner step and jumping straight into building — without a compressed brief, the agent produces a generic tool rather than a targeted business asset.
• Enabling automated outreach or automated purchasing before manually reviewing the first batch of deal cards — bugs like broken links or HTML entities in emails will damage your reputation at scale.
• Mixing multiple business ideas into a single Slack channel — makes it impossible to act quickly on individual deal cards and creates noise that kills the daily habit.
• Turning on the heartbeat token loop before the business generates revenue — heartbeat consumes tokens continuously and is a cost sink until there is cashflow to justify it.
• Choosing an asset without an obvious, reachable buyer — the Feed → Asset → Trigger chain is worthless if you cannot answer 'who pays first?' before committing time or capital.
• Mistaking a one-time output for a business — the value of these agents is in the daily recurring brief, not a single scrape. If the agent doesn't run on a schedule, it is a research tool, not a business.
• Assuming the agent is broken when it goes quiet — it may be autonomously reconfiguring or rebuilding itself. Check the app status before troubleshooting.
• Trying to build a 'billion-dollar startup idea' instead of a tiny, boring, immediately cash-flowable business — the methodology is optimized for speed to first dollar, not scale.

// What do key terms like deal card, one-liner, and heartbeat mean in this framework?

Tiny AI Agent Business Idea

A small, boring, immediately actionable business concept — not a venture-scale startup — that uses an AI agent to automate the monitoring and screening work, targeting $1,000–$3,000/day in cashflow. The emphasis is on 'tiny' and 'boring' as features, not bugs.

Feed → Asset → Trigger → Buyer → Monetization

The five-node chain that every viable tiny AI agent business must complete: a live data feed surfaces a mispriced or neglected asset; a trigger event flags it as actionable now; an obvious buyer with money exists; and a clear liquidity point (flip, broker, retainer, relaunch) closes the loop.

Dead Domain Flipper

A specific tiny AI agent business pattern: the agent monitors expired domain drops and auction platforms against a scoring criteria list (DR threshold, clean backlink profile, exclusion filters), ranks the top picks daily, and the operator flips winning domains to newsletter operators, SEO agencies, or rebuilds them as content sites.

Local Liquidation

A tiny AI agent business pattern where the agent monitors restaurant closures, liquidation auctions, and bankruptcy filings in a city, calculates the arbitrage spread between acquisition price and used market value, and the operator brokers the deal between the distressed seller and a new buyer for a 15–30% fee with zero inventory risk.

Hiring Signal Hunter

A tiny AI agent business pattern where the agent monitors job boards daily for roles that indicate budget is being deployed (e.g. growth hires, SDR teams), enriches company records to find decision-makers, and auto-drafts personalized cold outreach referencing the exact job post — used to fill a consulting or agency pipeline.

Should I Even Call? Memo

A one-page AI-generated acquisition brief that pulls financials, cross-checks reviews and web mentions for a business listed for sale, and delivers a go/no-go signal in minutes. Can be used personally for deal screening or productized and sold as a service.

Deal Card

The structured output unit of a tiny AI agent business: a single asset record showing acquisition price, used/resale market value, spread percentage, suggested broker fee, contact info, and a draft outreach message. This is the primary sales artifact sent to the obvious buyer.

Agents Are the New SaaS

A monetization philosophy shift: instead of selling software seats (per-seat model), you sell an AI agent delivering a specific outcome on a recurring basis (outcome-based model). The customer pays for the daily intelligence brief or the deal card, not access to a dashboard.

Obvious Buyer

The specific, reachable person or business with money who will predictably pay for the asset or intelligence the agent surfaces — e.g. newsletter operators for premium domains, new restaurant owners for used kitchen equipment, marketing agencies for hiring-signal leads. No obvious buyer = no business.

Liquidity Point

The specific mechanism by which the operator converts an agent-sourced asset into cash: Flip (buy low, sell at market), Broker (connect parties, charge 15–30% fee, no inventory), Retainer (sell recurring intelligence as a subscription), or Relaunch (acquire dead asset, improve monetization, grow). Must be chosen before building.

Heartbeat

An agent setting that checks for pending events and runs a maintenance task every 30 seconds to keep the agent continuously active. Off by default to conserve tokens; recommended to enable only once the business generates revenue.

One-Liner

A single compressed sentence that defines the entire business brief for the agent: naming the data feed, asset type, scoring criteria, delivery channel, and end buyer. Writing the one-liner before touching any tool is the mandatory first step of the methodology.

// FREQUENTLY ASKED QUESTIONS