Tools

What You'll Build

An overnight pipeline that finds businesses for you to buy. Not the ones already listed on BizBuySell that 500 other buyers have seen. The ones whose owners haven't decided to sell yet, but whose age, tenure, or business signals say they might be ready if someone made the right call.

The pipeline takes a one-page "buy box" you write in plain English (industry, size, geography, asking price range) and gives you, by morning, a ranked list of real companies that fit. For every company on the list, it has already pulled the owner's name, age, time in business, employee count, and a few signals about whether they might be ready to sell. And for each one, it has already drafted a one-page deal memo you can read before you pick up the phone.

This is the system Moran Pober's acquisition firm uses. It is also the system you can use to buy your first business.

Why This Works

The hardest part of buying a small business is not the negotiation, the financing, or the closing. It is finding a good one to buy.

Business brokers exist to solve this. They charge 10-12% of the deal value, often with a minimum of $25,000 to $50,000. In exchange they show you their listings. But broker listings are picked-over. The same five buyers are looking at the same listing the day it gets posted. The good deals are gone in 48 hours. And the listing you're looking at is, by definition, a business someone has already decided to sell. That decision pushes the price up.

The real opportunity is off-market. A business whose owner is 67, has been running it for 30 years, hasn't told anyone they're tired, but would say yes to the right offer. There are tens of thousands of these owners. There is no list of them. Finding them is grinding work.

Until now grinding work was the only way. You'd build a target list from D&B or ReferenceUSA, pay a research firm to enrich it, hire a college kid to skip-trace owners. One round of 200 targets would take a junior analyst two weeks.

Claude does the same work overnight. It reads your buy box, queries the database, enriches every target with public-record signals, throws out the ones that don't fit, and writes a one-page brief on every survivor. By morning you have a call list with the homework done.

This is the entire game. Buyers who build their own off-market lists are not competing with each other on price. They are competing on who calls first.

How the Pipeline Works

The system runs in five stages. Each one feeds the next.

Stage 1: The Buy Box

Before any AI touches anything, you write your buy box. This is a plain-English description of the kind of business you want to own:

Save this as a Claude Code skill. Now every member of your team is screening against the same criteria, and the AI can read it on every run.

Stage 2: The List

A second agent reads the buy box and queries a business database. The most accessible options for SMBs:

The agent pulls every company that matches the buy box, with no filtering yet. For a typical search you'll get 300-2000 raw targets.

Stage 3: Enrichment

For each company on the raw list, a third agent fills in the details that don't come from the original database:

Clay handles a lot of this in one pass. Whatever Clay can't fill in, Claude attempts via LinkedIn, the company's website, and public records.

Stage 4: ICP Qualification

Now the agent throws out the ones that don't fit. This is where a sub-agent does the work: read the enriched record, compare against the buy box, kill anything that doesn't match.

You'll typically lose 60-80% of the list at this stage. Wrong industry, wrong size, wrong geography, owner too young, owner too rich to care, business growing too fast (the owner won't sell).

What's left is your call list. Maybe 50-150 names from a starting pool of 2000.

Stage 5: The Deal Memo

Before you pick up the phone, the agent writes a one-page brief on every survivor:

You read the memo, decide if you want to call, and call.

Step-by-Step Setup

Step 1: Write Your Buy Box

Spend half a day on this. Be specific. The narrower the buy box, the better the results. "Profitable small businesses" is not a buy box. "Residential HVAC service companies in DFW with $800K-2M revenue, owner age 55+, run by the owner" is a buy box.

Save it as a Claude Code skill (/skills/buy-box.md) so every agent in your pipeline reads from the same file. When you tune the buy box, every downstream agent uses the new version automatically.

Step 2: Pick a Database

Start with one. Apollo is the most SMB-friendly. You can run a query, export a CSV of 1000-2000 companies, and have something usable on day one.

If you can get access to D&B Hoovers through a corporate or library subscription, that's better data, but Apollo is enough to start.

Step 3: Build the List Builder Agent

A Claude Code agent that:

  1. Reads your buy box.
  2. Converts it into Apollo search filters.
  3. Pulls the result set.
  4. Drops the raw list into a Google Sheet.

This is a one-prompt agent. The whole job is "read the buy box, translate to filters, run the query, drop the rows."

Step 4: Build the Enricher

A second agent that walks through every row in the raw list and fills in the missing fields. Hook Clay in via its API for the bulk enrichment. For anything Clay misses, the agent does manual LinkedIn and website lookups.

This one is slower because LinkedIn rate-limits and websites vary. Plan for it to run overnight on a list of 1000+ rows.

Step 5: Build the Qualifier

A third agent that reads each enriched row against the buy box and writes either PASS or KILL with a one-sentence reason. The killed rows stay in the sheet but get hidden. The passes move to a separate tab labeled "Call List."

This is also a one-prompt agent. The buy-box skill does the heavy lifting.

Step 6: Build the Memo Writer

A fourth agent that takes each PASS row and writes a one-page deal memo into a new Google Doc (or a new row with a long-form text field). Use a template so every memo follows the same structure.

This is the agent that saves you the most time. Reading 50 memos that all follow the same shape lets you decide who to call in 90 minutes. Doing 50 memos by hand is two weeks of work.

Step 7: Wire the Daily Cycle

Set the whole pipeline to run weekly. Most business databases don't change fast enough to merit daily refreshes. Once a week is plenty.

Schedule:

You spend Monday morning making calls instead of researching them.

Adapting This for Other Hunting Workflows

The "buy box, list, enrichment, qualifier, brief" pattern is not unique to acquisitions. The same shape works for:

Commercial real estate sourcing. Buy box becomes a property profile (asset class, geography, vintage, cap rate target). Database becomes CoStar or Reonomy. Memo becomes a property brief.

B2B sales prospecting for high-ticket services. Buy box becomes an ICP. The qualifier and memo writer save the BDR team hours of research.

Hiring for a specific senior role. Buy box becomes a candidate profile. Database becomes LinkedIn. Memo becomes a candidate brief.

Investment scouting. For angel investors and small VCs, buy box becomes a thesis. Database becomes a startup database. Memo becomes a one-page investment brief.

In every case, the same five stages apply. Start narrow, build the list, enrich, qualify, brief.

Gotchas and Tips

What the Pros Are Doing Differently

The acquisitions.com walkthrough hints at a more advanced version of this pipeline. They use Claude Code's sub-agents to run multiple buy boxes in parallel from a single buyer, so a holding company can hunt for plumbing companies and accounting firms at the same time without confusion. They also feed completed deals back into the qualifier so the agent learns which signals actually predicted closes.

You don't need either of those to start. The five-stage pipeline above is enough to find your first business.

Keep Reading