I'm Griffin. I connect the tools you already use so information moves on its own, build simple dashboards, and add AI only where it pays its way. You only see what needs your yes.
AI sorts, drafts and filters. You only see what needs a decision.
Built for scaleSites I've built and run have served millions of visitors a year.
IndustriesE-commerce & digital productsOnline courses & membershipsPhotography & creative educationHospitality & bookingsNonprofits & givingTravel & aviationSportsEngineeringLocal trades & services
The detailed, rule-following work that has to be done right every time. Example from a typical small-business workload, not a guarantee. Where you land depends on your tools, your rules and how much you want to approve yourself.
AI isn't always the answer, and that's the point. Plenty of what I build is made with AI but runs on plain rules, automatic checks and your sign-off.
Each recurring job gets a time, a destination and a status. Once it says AUTOMATED, it runs on its own.
your workflows · gate 1
| Time | Workflow | To | Status |
|---|---|---|---|
| 07:30 | Order sync | Customer DB | Automated |
| 08:15 | Income dashboard | Owner | Automated |
| 09:00 | Social graphics | Approval | On time |
| 10:30 | Review requests | Customers | Boarding |
| 11:45 | Checkout smoke test | Store | On time |
| 13:00 | Translated captions | Courses | Scheduled |
| 14:20 | Bounce cleanup | Email list | Scheduled |
| 15:30 | Error watch | Slack | Scheduled |
Not another all-in-one platform. Small views on your own data, each answering one question. Everything below is sample data.
A short route with clear stops. Before anything gets built, you see the plan, the price and what you'll get.
Tell me where the week goes. I'll tell you honestly what can come off your plate, with AI or without it, and what can't. Book it.
We map every place work gets copied, chased or re-typed and rank it by hours saved. You get a board like the one above, filled in with your real jobs, plus a price.
I build only what the plan earns. Usually some mix of:
Plain-English notes and a walkthrough with your team. We agree up front on who hosts it and what support looks like after.
Arrival: your hours, backSlow website? I also tune sites for speed, search and AI-search visibility, so the people asking ChatGPT for “a good dentist nearby” find you.
Ask about a site tune-up →
I started as a designer, became a developer, and today I build and run AI agents, automations and web apps every day for real businesses. My clients have been all over the map: the US, Australia, India and beyond. And I've worked from more than 60 countries along the way. You work with me directly, start to finish. No hand-offs to a junior, no agency layers.
Ten things I'm good at, and the rules I follow while doing them. Open any line for the thinking and the proof.
Keep every old URL and redirect the rest. Rehearse on a staging copy, back up before every write, never email past buyers mid-move, and walk the whole upsell chain with cheap test orders.
Unlock by purchase, never by hand. Stop double purchases at the source.
“Accepted” isn't “delivered.” Suppress bounces and complaints automatically, cap how often people hear from you, and stop link scanners from confirming sign-ups.
The thank-you page should hand over the gift above the fold. Rewards should feel personal, not generic.
Plain-rule gates are cheap, predictable and explainable. Anything unmatched goes to a person, never a guess, and a person approves anything a customer will see.
Ask only people who finished or bought, and only once. Hold every review for moderation, never mix old counts with new reviews, and disclose incentives.
Quarantine before deleting. Watch for problems on a schedule instead of waiting for a customer to report them. Audit code other people wrote.
Measure before and after. Fix the underlying data so answer engines quote the right numbers.
Keep private data on your own server. If a number looks too good, check the data before celebrating.
Design is still the first thing people judge, before they read a word.
The two builds with the most moving parts, then a few more. Names left out on purpose; the details are real.
It searches ten marketplaces and social platforms for creators in your niche, learns their audience size, what they sell and how to reach them, and scores each one on what actually predicted earnings for past partners. Every lead moves through a clear pipeline from new to signed.
Outreach is drafted in your own voice, learned from hundreds of your sent emails, with nothing invented. Nothing sends until a person chooses to: campaigns start in practice mode, duplicates are merged, and every job reports if it stops.
List the products, tiers, prices, tags and course access once. The app builds the checkout options, purchase notifications, customer-list automations and student access across four systems, renders checkout images, and attaches translated captions.
It checks that email countdown timers match the real deadlines and test-drives every checkout variant in a headless browser to confirm the right lines and totals. What used to take hours of error-prone clicking becomes a guided, repeatable run.
Every customer since 2014, from five different checkout systems, merged and de-duplicated into one list (a CRM-style database) that feeds the help desk.
A spreadsheet schedule turns into on-brand graphics and drafted captions. A person approves each one before it's scheduled, and the drafts learn from their edits.
Eleven kinds of systems, built and running for real businesses, including three of my own. Pick a gate.
Unique graphics made by rule, in bulk, and synced to where they're used.
AI translates against your glossary, past translations and house rules; native speakers approve.
Recordings become readable pages and print-ready PDFs.
One settings file sets up a whole sale, checks itself, and waits for your sign-off.
Every customer and every dollar in one place, de-duplicated and kept in sync.
Private places where partners, staff or clients log in and find their stuff.
Tools that search the web for you, score what they find, and ask before acting.
Assistants trained on your own material, with spending caps on every one.
Work gets prepared automatically, then waits in a queue for a human yes.
Sell it, unlock it, and keep members moving through it.
Businesses I build and run myself, so I feel every rough edge before a client does.
Usually not. If you or your team spend a few hours a week copying, chasing or re-typing, there's likely something worth automating. The call will tell us.
I quote after the call, once we both know what's worth building. There are no published prices, and you see the price before any work starts.
Builds that run on plain rules and connections between your apps (APIs) have little or no ongoing cost beyond tools you already pay for. AI features have usage costs; I set spending caps and tell you the expected range up front. Optional maintenance plans are available for a low monthly fee: I keep everything running smoothly, make sure nothing breaks when the outside services it connects to change their APIs, and can add a few small features along the way.
We agree on what each tool can see and where your data lives. Nothing goes out to customers without your approval unless you choose that. Your data can live on my servers for a low monthly fee, or on a server you choose, control and pay for. With a maintenance plan, I'll need access to that server.
You get plain-English notes on how everything works, and we agree up front on who hosts it and what happens if we stop working together. Every app I build also has a searchable help section, so your team can look things up without calling me.
No. I build data dashboards that show what's happening with income, expenses and leads. Your accountant still does the books and taxes.
Thirty minutes on a video call, no slides. You'll leave with two or three things worth automating, whether or not we work together.