🆕 Buy Box Builder + 10 Flippa Deals

Hey there 👋,

This newsletter, the Redundant thesis and the product range all just got a lot more coherent.

The way I see it, AI is going to make most of us redundant. People in employment, and business owners who get disrupted by it. So we own the disruption instead: automate your own tasks, then own income-producing assets. The guys at Unemployable* view it this way too.

The free Redundancy Audit is my top of funnel now. It scores your job task by task, hands one task back to you as recovered hours, and ends with one verdict naming one of two routes.

Option A: build one. The Agency Playbook, which I launched at the start of this month. Ten AI workers you install for local businesses and charge $999 to $1,999 for, plus the Playbook document: what to charge, who to sell to, and how the install session actually runs.

Option B: buy one. The Acquire Playbook. The fifteen buy-side workers in the previous Acquisition, a new Playbook doc (what your money actually reaches, what diligence is really asking, how to close etc), plus a toolkit. Also $299 including a year of Pro.

If you already owned the pack, you already have it. Same workers, plus the document, no extra charge, sitting in your folder.

For those who have been around a while, this second one goes back a way: I built and sold the Acquire.gg newsletter network to Scott Oldford which included the Acquire The Web newsletter (now the LetterTrader newsletter).

And this newsletter and community now runs in four sections. If you are only here for the Flippa deals, jump straight to ACQUIRE.

The trackers through 27 July: AI is explicitly cited in around 50% of 2026’s cuts so far, some 170k+ roles, with tech well past 150k for the year and March the peak month for AI-reasoned cuts. Support, engineering and middle management are taking the worst of it at the scale players. Companies aren’t trimming. They’re reallocating around agentic systems.

🏭 monday.com: ~620 (20% of workforce), restructuring to fund its AI Work Platform where people and agents operate in one workspace. Link

📱 Patreon: 93 (20%), flattening the organisation and adjusting cost structure as AI transforms how the company builds and operates, while denying direct replacement. Link

🚗 Uber: 10% of community operations/customer service team, simplifying a “too complex and siloed” structure to scale AI tools across support. Link

The free Redundant audit used to end with a score and some advice. Advice is where most of these things stop, and it is why most of them do nothing.

It now ends with a task. One task off your own list, picked for the hours it gives back and for data you can safely touch, with the arithmetic shown and rounded down. Then a second worker builds it with you, tests it against your own definition of a good result, and fixes it in front of you. An hour, and you have the hours back plus the ability to do it again.

The order matters and it took me a while to see it. The old version asked for five to ten hours a week from someone who does not have them. Now it hands over the hours first and then says what to spend them on.

Two rules run through the whole thing. Nothing leaves the building: your own machine, your own or dummy data, or whatever tool your employer already sanctions. Never a company system wired to a personal account, because that is the one move that can actually cost you the salary you are using as runway. And keep your output flat afterwards. You are not obliged to tell your employer you got faster. Telling them is the only genuinely bad move here, because it hands over the redundancy case for free and you get nothing back for it.

I ran it twice on real job ads to check it. A warehouse data analyst at a global logistics firm scored 64 and got routed to BUILD. A pharma software consultant scored 66 and got routed to ACQUIRE. Nearly identical exposure, opposite verdicts, because their money is different. Both reports come with the download.

Run it free →

This week my Claude workers read every live Flippa listing between $20,000 and $200,000 across ecommerce, SaaS, content, newsletters and YouTube. A bit over a hundred of them. Ten made the board.

Having done this kind of newsletter since 2020 with the Website Investing one I sold to Travis I’m assuming this is the range that most subscribers are interested in - if you’re looking for larger deals (or smaller) just hit reply and I’ll change this buy box. Talking of which the Buy Box Builder is a new free tool you can grab, details at the end of this section.

These are first glances, not teardowns. My workers valued them off the listing data and picked the ten worth your time. Almost everything they threw out failed the same way: a confidential dropship store claiming five figures a month at a sub-1x multiple, which is a price that only makes sense when the traffic is rented and the seller knows the lease is up.

Before you dive into any of these in detail: write the box first. The new free Buy Box Builder turns your capital, your reserve, your hours and what you already know from working life into written criteria, so you can say no to eighty percent of a board like this in under a minute. Browsing before the box exists is how people fall for whichever listing flatters them.

In 2016 I wrote a post called Productizing Services To Flip The Script (check it out on archive.org). I got it ranking top of Google for "productized services", and the comments filled up with founders of the businesses I'd listed. Twenty-seven of them: WP Curve doing WordPress support, Undullify doing unlimited graphic design, Bench doing bookkeeping, Boombait doing app icons, Authority Engine doing podcast editing.

The argument was that a service with a fixed scope, a fixed price and a fixed delivery date beats freelancing, because you stop trading hours, and beats an agency, because you stop writing proposals you lose. I'd built one myself in the UK ski trade and it was the first time I understood how to scale anything.

Here's the line from that post I keep coming back to:

Delivered by others. In 2014 "others" meant a VA or a writer off a marketplace, and that was the ceiling on the whole model. You productized the service, hired people to run the SOP, and kept whatever was left after payroll.

The others changed. An SOP with a fixed scope, a fixed input and a fixed output is the exact shape of a Claude skill. Every one of those 27 businesses was writing a spec for a worker that couldn't be built for another ten years.

So that list reads differently now. It isn't 27 businesses. It's 27 specs.

I don't think the move is to build 27 skills. I've just spent a week doing the opposite to my own catalog. The move is one worker that takes any productized service, yours or one you're looking at, and turns the SOP into an installable skill you can sell or deliver with.

I haven't built it. I'm telling you because I'd rather be wrong in public than launch something nobody asked for.

If you run a productized service, hit reply and tell me what it is and what the SOP looks like. If enough of you do, it's the next thing I build and you get it first.

I let real domainers loose on my domain operator pack. They broke it in six places in a week.

I gave Domain Operator to a couple of people who actually sell names for a living, one of them at six and seven figures. Every day for a week I got a list of what was wrong. The pack went from v2.2 to v2.7. Here is what changed, because most of it is worth knowing even if you never install anything.

It was pricing names at wholesale. With no asking price supplied, it leaned on NameBio and similar sources to suggest what to ask. Those are domainer-to-domainer prices. What a company pays to brand itself is a different market, often by an order of magnitude. Anchor your ask to the wrong one and you quote a fraction of what the name is worth. Every comp is now labelled WHOLESALE or RETAIL and an end-user ask never gets built from a wholesale number without the adjustment shown.

It would have emailed trademark holders. Offering to sell a domain to the company holding a live mark on that term is the sort of email that shows up later as an exhibit, and they can take the name instead of buying it. It now screens every lead and refuses to draft. Dead marks are ignored, pending ones are not.

It pitched dead companies. One suggested lead's own domain was itself for sale, meaning they had rebranded or folded. Every lead's site is now fetched before it makes the list.

It flattered bad leads. A company sitting on the .io of a name you own already knows you exist and has passed. That is not a hot prospect, it is a no with extra steps, unless something changed or you will flex on terms. Rebrand candidates lead now, because they have never considered your name.

Also new: whole-inventory runs with a price and terms per name, lease-to-own drafted into the outreach, B2B/B2C tagging, and funding signals screened by how weak the buyer's current domain is. A fresh raise is not an opportunity. A fresh raise plus a hyphenated domain is.

Related, and not mine: a domainer posted this week that he sold velocity.io for $100,000 last November. The buyer just raised $27M. They bought the name on the way up, not after the round.

Whether you are in employment or in a business you want out of, the community now has a channel for each of you.

And in the Pro channel I shared the Redundant watch prototype and initial samples are somewhere between China and Slovakia. The Pro channel got the first photo. Here are pics of the first couple of watches I built with my old RP logo, off my richardpatey.com site back in the day.

Cheers!

Richard

The obvious bit: nothing in this email is financial, investment, or legal advice. The deal reads are produced by AI workers from public listing data, unverified, and I hold no position in anything listed. Valuations are opinions, multiples move, and no verdict here replaces your own due diligence. If you're buying anything, verify the numbers yourself before money moves.