Back to site
// 1 → 10

What one person can build with AI.

24 systems, built by one person. Each one replaces what used to take a team — or wasn’t possible at all. Scroll, or use ↑ ↓, to flip through them — and open any one for the full record.

01Client · a software company

A system for next-level sales prep

Done in minutes, behind the scenes — and sharper every call.

Paste a potential customer's website and, two minutes later, the salesperson has a complete, fact-checked briefing — who they are, whether they're worth pursuing, and how to win them. Every call teaches it to do the next one better.

Open the full record
  • Prep per call30–60 minutes of hand-research across a dozen tabsabout 2 minutes — paste and go
  • What the team learnsleaves with whoever ran the callstays in the system and builds up
  • Each callstarts from zerosharpens the next one
  • To run it at scalea dedicated sales-operations teamone person and the tools they already use
  • What the salesperson walks in witha huncha fact-checked brief and a clear recommendation
02Client · a software company

Always-on research on your customers and rivals

A full report every week, gathered automatically.

It reads everything people say about the company and its competitors online — reviews, forums, social, video — and writes a clear summary every week. The kind of research a team used to do once a year.

Open the full record
  • How often it runsonce or twice a year, if at allweekly for customer pain, monthly for competitors
  • Effort per rounddays to weeks of manual reading and taggingset up a profile once, then collect the report
  • What you get backa stale snapshota live feed of real quotes and competitor moves
  • Cost to run againanother full sprintnear zero — the engine just runs
  • Serving more than one companya whole new project each timeswap one settings file
03Client · a software company

A system to get your business recommended by AI

Show up when ChatGPT answers — not just on Google.

When someone asks ChatGPT or Google's AI "what's the best tool for this?", does it name you? This finds out, shows where you're invisible, and fixes it — so the AI starts recommending you.

Open the full record
  • What gets measuredwhere you rank on a Google results pagewhether the AI names you when it answers
  • How you'd check itask the AI a few times by handa scored, repeatable read across many questions
  • Knowing where you standa guessa health score and share against named competitors
  • Knowing what to dono clear next stepgaps ranked by effort against payoff
  • Fixing the gapswrite pages and hopedrafts built the way AI assistants prefer to read
04Client · a small local business

A full operations team for a small business

Scheduling, marketing, and the numbers — run by one person.

For a small, owner-run business: AI handling the scheduling, marketing, follow-ups, staff training, ads, and the numbers — all the jobs a big company hires a whole team for, run by one person.

Open the full record
  • Internal toolsnone — no budget for a software hirecustom scheduling and lead capture, built in
  • Knowing which jobs paya gut feel, never time to checka profit analysis that answers it plainly
  • Running the adsdone badly, between everything elseplanned and run with results auto-reported
  • Training field staffad hoc, whenever someone had a minuteready-made material and a question-answering base
  • Who runs all of ita team a small business can't affordone person, with AI doing the heavy lifting
05Experiment

A live coach for your video calls

Real-time nudges, private, on your own laptop.

A Mac app that quietly helps you while you're on a video call — reminding you of your key points, nudging you when you're talking too much, reading the room — completely private, running on your own laptop.

Open the full record
  • Where the AI runsin the cloud, on a serverentirely on your own laptop
  • Speedlag that breaks 'real-time'fast enough to coach you live
  • Ongoing costa per-minute charge for every callnone — it runs on hardware you already own
  • Privacya private call streamed to a servernothing leaves the device
  • Setupper-app integrations and audio pluginsone app, using the system's own recording path
06Product

A system that answers every new lead instantly

A real reply in seconds — day, night, or weekend.

The instant someone reaches out, they get a real, helpful reply by text or email — answering their questions and booking them in — even at 2am, with nobody watching the inbox.

Open the full record
  • Time to first replyminutes to hours, if anyone is watchingseconds, automatically
  • After-hours and weekend leadssit until Monday and go coldanswered the moment they arrive
  • What the lead gets backa generic autoresponder, or silencea real answer to their actual question
  • To cover the inboxsomeone on shift watching for new inquiriesone agent, always on
  • Best leads (off-hours)the ones most often lostthe ones now captured first
07My own product

Your entire work life in one system you own

Contacts, projects, email — together, on your own machine.

Contacts, projects, email, calendar, notes — all working together as one system on your own machine, instead of a dozen apps that don't talk to each other and bill you every month.

Open the full record
  • Where your data livesscattered across a dozen companies' cloudson your own machine, owned by you
  • How the tools connectthey mostly don't — you copy between themthey share one foundation and feed each other
  • What it costs to runa stack of monthly subscriptionsclose to nothing, on a subscription you already have
  • Who controls iteach vendor, on their termsyou, locally
  • Getting more useful over timeeach app improves in isolationthe pieces compound as they share more
08Client · a software company

A system that turns scattered customer feedback into answers

Built on what the company already owned — a fifth of the SaaS price.

It reads everything customers are saying — support tickets, in-product surveys, public reviews — and turns it into a dashboard, a Monday-morning digest, and plain-English answers on demand, with the customer's actual quotes attached. Replaces a ~$10K-a-year tool with weeks of work on infrastructure the company already ran.

Open the full record
  • Where feedback livesscattered across tickets, surveys, reviews, sales callsone place, inside the warehouse already running
  • Effort per questionhours of copy-paste and hand-synthesis into a doca one-line question, answered with source-linked quotes
  • Cadencea manual reading sprint once a quarter, if at alla live dashboard, a Monday digest, and answers on demand
  • Annual cost~$10K/yr for the off-the-shelf SaaSwell under $2K/yr, on infrastructure already paid for
  • Where customer data ends upanother vendor's database, behind another loginnever leaves the company's own warehouse
  • Time to set upa months-long vendor rolloutabout four weeks to ship v1
09Client · a global consumer brand

A publishing engine for marketing — brand-safe and self-tuning

Pages a marketing team ships in minutes — then quietly tune themselves.

Landing pages a marketing team can publish without engineers — brand-locked components so pages can't go off-brand by construction, A/B tests on by default, and a loop that critiques and rewrites each live page against its own goal until it stops getting better. A weeks-per-page release cycle became pages shipped any day.

Open the full record
  • Who can ship a pageengineering, queued behind sprintsthe marketing team, any day
  • Time per releasea sprint or twominutes to push live
  • A/B testingone-off, often unreadon by default, results read by the system
  • Brand safetyreviewed by hand, slowed releasesenforced by the components themselves
  • After launchnothing — the team had moved onthe page critiques and rewrites itself until it stops improving
  • Cost of trying ten ideasthe cost of ten engineering projectsthe cost of one
10My own product

A five-agent growth agency, running my own practice

Five named agents do the work; I approve strategy, not every post.

Five AI agents — a Growth CEO, a trend researcher, a content writer, a distributor, and a performance analyst — running BetterStory's marketing on their own. Each has a role, a budget, and a reporting line. I approve strategy and audit output instead of writing every post.

Open the full record
  • Marketing for a one-person practicean agency retainer, or done in the gaps between client worka named five-agent team running it as the work
  • Cadencesporadic, when you have a Friday afternoonweekly content, daily distribution, monthly review
  • Trend awareness'I'll catch up on the field later'a Scout agent reading the field on its own
  • What gets postedwhatever you remembered to writewhat the CEO and Scout agreed was worth saying
  • Your roledo every step yourself, or hand it all overapprove strategy, set budgets, review output
  • Costan agency retainer or a part-time hiremodel calls and the orchestration runtime
11Client · a software company

A marketing department in a box

Campaigns, leads, and web pages — produced on demand.

A set of AI tools that each handle one marketing job — writing campaigns, finding new customers, building web pages that improve themselves — and hand the team finished work it can use the same day.

Open the full record
  • How the work gets madeby hand, one piece at a timean engine produces it on demand
  • How long it stays usefulstale within weeksregenerated fresh whenever needed
  • Who it takesa full-time person per jobone operator running all the engines
  • How the jobs connectseparate teams, separate silosengines that feed each other
  • Where the output goesscattered docs and dashboardsstraight into the CRM, docs, and Slack
12Client · a software company

A system that writes your monthly business review

A finished deck, built from a 15-minute chat per leader.

Each leader has a quick 15-minute chat with an AI, and the monthly review assembles itself into a finished, polished deck — accurate and current, with nobody stuck building slides.

Open the full record
  • How leaders give inputfill out a templatea 15-minute guided AI interview
  • Building the deckone person stitches it by hand over daysit assembles itself in minutes
  • How current it isstale the day it shipsbuilt from fresh answers each cycle
  • What it showsa polished restatement of the planwhat's actually moving, checked against the metrics
  • What leaders spend time onthe logistics of reportingthe decisions the review is for
13Client · a software company

A system that keeps leadership in the loop

Weekly updates, written automatically from the team's chat.

The team jots a few words in their chat app as they work, and the weekly update for leadership writes itself — no status meetings, no one keeping a board up to date.

Open the full record
  • Reporting efforta maintained board, status meetings, weekly write-upsabout five minutes a day of tagged one-liners
  • Who writes the updatesomeone, by hand, every weekthe system drafts it; a human just approves
  • Tracking goalsceremony nobody had time for, so it lapseda light weekly check-in that runs itself
  • A project that didn't workquietly disappearslogged as a lesson, not a failure
  • What leadership seesstale, or skipped that weeka fresh update every week
14Product

An AI that answers your phone and books the work

Natural conversation, every call caught, around the clock.

It picks up every call, has a natural conversation, and books the appointment — so a busy business never misses a customer, even after hours. One system can answer the phones for many businesses at once.

Open the full record
  • A missed callusually a lost job — the caller moves onanswered live and turned into a booking
  • Phone coveragea receptionist or answering service per businessone shared engine serving many at once
  • Evenings and weekendsvoicemail, and the work goes elsewhereanswered the same as any other hour
  • Cost to cover the phonea fixed salary or service fee per locationa shared system that scales across all of them
  • What the caller experiencesa recording, or no answera natural conversation that books their slot
15My own toolkit

The toolkit that lets one person build all this

Every project starts from everything already learned.

A reusable set of building blocks I've made, so every new system starts from everything I've already figured out instead of from scratch. It's the reason one person can ship this much.

Open the full record
  • Start of a new buildrebuild the plumbing from near zeroinherit a working kit on day one
  • Lessons from the last projectevaporate when it shipscarry forward, baked into the kit
  • Checking the output is goodbolted on by hand, late, every timeautomatic quality checks built in from the start
  • Keeping results trustworthyguardrails re-learned per projectproven guardrails reused everywhere
  • Pace of shippingeach project starts sloweach build starts at a higher floor
16Open-source project

An open-source runtime for autonomous AI 'companies'

Many agents, one org chart — the platform my Growth Agency runs on.

An open-source orchestration platform (Paperclip — I fork and contribute to it) that turns a team of AI agents into a 'company' with an org chart, budgets, governance, and a full audit log. If OpenClaw is an employee, Paperclip is the company — and the five-agent Growth Agency that runs my marketing is one company built on it.

Open the full record
  • Coordinating many agentsa research topic, impressive in demosa runtime you can run concrete companies on
  • Keeping agents on taskthey drift, loop, or lose the goalan org chart and a CEO hold the group to one objective
  • Cost controlrunaway token spend, no ceilingmonthly budget per agent — they stop when they hit it
  • Auditabilitygood luck reconstructing what happenedevery tool call traced, full immutable log
  • Running more than one companyanother stack from scratch each timeone deployment, many companies, full data isolation
  • Human involvementconstant supervision, step by stepyou're the board — approve hires, set budgets, override
17Product

A system that builds a business its website

A finished site from public info — no designer, no agency.

It finds a business, builds it a proper website from information that's already public, and puts it online — no designer, no developer, no meetings. A finished site before anyone's been hired.

Open the full record
  • Who builds the sitea designer and a developerthe pipeline, automatically
  • Getting starteda kickoff call and an intake processnothing — it starts from public information
  • Time to something realweeks of back-and-fortha live preview link produced on its own
  • What the business sees firsta quote and a timelinea finished version of its own site
  • Cost to get onlinean agency bill many can't justifyclose to nothing for the first draft
18Experiment

A system that turns ideas into launched products

Finds demand, builds the winners, plans the launch.

One system that scans for what people actually want, judges which ideas are worth it, builds the promising ones, and plans how to sell them — the work of a whole team, run as one pipeline.

Open the full record
  • Who runs the funnela team across several rolesone person operating a pipeline
  • Time to explore an ideaweeks to months per stagefast and cheap enough to run many at once
  • Cost of a dead endexpensive — staffed time spentlow — the pipeline absorbs it
  • How many ideas you can testa few you can afford to bet onmany, run in parallel
  • The real bottleneckcan we afford to explore thiswhich results are worth taking forward
19Experiment

A team of AI assistants that actually remembers

Picks up where it left off, and gets better over time.

Five AI helpers, each with its own job and a memory that lasts — so they pick up where they left off and get more useful over time, instead of starting from nothing every time.

Open the full record
  • Memorynone — each session starts coldpersists in a shared store across sessions
  • Number of agentsone, doing everythingfive, each a specialist
  • How work continuesevery interaction from scratchpicks up where the last run ended
  • Over timeno accumulation — nothing compoundscontext builds, and it gets more useful
  • Character of the systema tool you querya team that remembers
20Experiment

A public interactive explainer for how AI changed SEO

A consultant's deck — replaced by a free five-minute interactive.

A single-file HTML deck that walks anyone through what SEO and marketing analytics used to look like, what changed with AI, and what the work actually is now. Open the link, click through, walk away with a clean mental model. Free, public, no SaaS in the way.

Open the full record
  • Formatworkshop, NDA deck, or long essaysingle-file HTML deck, click-to-flip
  • Accesspaywall, signup, or you-had-to-be-thereopen the link, no signup, no analytics on the reader
  • How long to grasp the ideaan hour, if you sit through itfive minutes of clicking
  • Where it livesa SaaS slide tool with a SaaS billone HTML file, host anywhere
  • What it teachesbuzzwords with examplesthe old workflow vs. the new one, side by side
21My own toolkit

A personal operations platform where the agent is the operator

Most days, I don't open it — I delegate in plain language.

A bespoke personal CRM, integration toggles, and a strict secrets model — built on top of an autonomous-AI-employee runtime so an agent does most of the operating, and I delegate in plain language instead of clicking through apps.

Open the full record
  • Where the work happensacross a stack of consumer-SaaS appsone runtime, with an agent doing the operating
  • How you reach your dataopen the right app, search, scrollask the agent in plain language
  • Integrationsa fragile chain of webhooks and Zapierexplicit toggles, each priced and budgeted
  • Where secrets livein whichever apps you set up firstone model, one place, one set of rotation rules
  • What the platform 'looks like'a UI you click throughan agent you delegate to
22Client · a B2B SaaS company

Personal outreach, written one prospect at a time — at campaign scale

Hand-writing hundreds of truly personal pages is impossible for a team → an engine that writes each one individually, at scale, without making anything up.

Every prospect gets their own landing page and their own message — written by name and to their actual situation, not merged from a template. One run produced hundreds of genuinely personal pages, each one fact-checked so nothing about the prospect is invented.

Open the full record
  • How personal each message isa first name merged into a templatea page written to that prospect's real situation
  • How many you can doa handful a week, by handhundreds in a single run
  • What stops a made-up claimwhoever proofreads it, if anyoneevery fact tied back to the research brief
  • Quality controlspot-checked, or not at allvoice, links, and emails checked on every page
  • Knowing what workeda guess once replies trickle intracking on every link, read in one place
23Client · a B2B SaaS company

No inbound lead reaches sales un-researched

Half an hour of manual digging per lead, often skipped when volume spikes → every lead pre-researched before sales makes contact, in about two minutes.

Paste an email or a website. About two minutes later there's a sales-ready brief in the team's workspace — who they are, how well they fit, and how to approach them — so every inbound lead arrives pre-briefed instead of cold.

Open the full record
  • Research per lead30–60 minutes across a few tabsabout 2 minutes — paste and go
  • Which leads get researchedthe ones someone had time forevery one, automatically
  • How consistent it isno two worked up the same waythe same workup, every lead
  • Knowing if they fita quick gut read, if anya fit score with the reasons behind it
  • What sales opens witha name and an email addressa sales-ready brief, ready to act on
24Client · a small service business

Lifting a website out of a page-builder so AI can run it

A site rented from a page-builder you can't deeply control → fully owned code an AI can operate at the drop of a prompt.

A marketing site trapped in a closed page-builder, rebuilt as standalone code the owner fully controls — every image and font hosted by them, nothing calling out to someone else's servers. Once it's owned, AI can run sitewide SEO upgrades, publish content, and sweep dead links on demand.

Open the full record
  • Who owns the sitethe page-builder hosts and controls itthe owner — it's their code
  • What it depends ona third party's servers for images and fontsnothing external — every asset is local
  • What you can changeonly what the editor allowsanything — it's real code
  • Sitewide changespage by page, by hand, in the editorall 560 pages at once, on a prompt
  • Who can operate ita person clicking through the builderan AI, running upgrades on demand

And this is only a fraction of what’s now possible.

Start a conversation