Your team typed the same invoices again last Friday.
Machines do the work. You run the business.
I build small machines that do the weekly busywork, so your team stops doing it.
You already have the tools. Accounting software, email, a CRM, WhatsApp for everything else, and a shared drive somewhere. The problem is that none of them talk to each other.
So someone on your team ends up being the bridge. They copy invoice data every Friday, they chase overdue payments from a spreadsheet, and they file project documents into the right folder one by one. That person is expensive, and the work shouldn't exist.
The fix isn't a new tool you install or a consultant you hire, and it isn't a six-month project. It's a small machine that moves information between the tools you already have. Nothing to log into, nothing to migrate, nobody to train.
Right now I have a machine that reads delivery notes, whether scanned paper or PDF, and extracts every field in under a minute for about ten cents, with close to zero errors. A person takes between two and ten minutes to do the same thing, and the error rate climbs when they're tired. The same one reads invoices and receipts too. Three problems, one pattern.
- 4h copying invoices every Friday→Extracted and filed automatically
- Quotes followed up from memory→Nudge sent 3 days after sending, no human involved
- Partner inbox: 200 emails, 2h of daily triage→Classified, drafted, queued for review
- Job orders arriving on WhatsApp, lost twice this quarter→Single intake, planner notified instantly
- Onboarding packet assembled by hand per hire→Sent, personalised, and chased automatically
- Same 5 questions answered 20 times a day→Answered instantly, escalated when needed
Before and after
Which one is yours?
Pick the task costing you the most hours. I have written up 35 of them.
- Incoming quote requests as free-text emails
- Supplier invoices keyed into the accounting system by hand
- Incoming job orders scattered across inboxes and messaging apps
- Partner or executive inbox triage
- Incoming delivery notes reconciled against POs by hand
- Recurring FAQs answered from memory or a Word cheat sheet
What I build
Automation
Flow glue between the tools you already use, so the weekly and monthly manual tasks move on their own.
- Invoices
- Emails
- Web forms
- Spreadsheets
- POS tickets
- WhatsApp messages
- Shared inboxes
- Paper-based counts
- Updated records
- Automatic notifications
- Generated reports
- Classified and routed emails
- Data extracted from one tool and dropped into another
AI agents
Custom AI assistants that read, classify, answer, or summarise, so your team focuses on the decisions, not the triage.
- Emails
- PDF documents
- Customer queries
- Meeting recordings
- WhatsApp messages
- Shared inboxes
- Recurring customer questions
- Classified emails
- Extracted structured data
- Automatic responses
- Meeting summaries
- Routed tasks
- Agent-drafted replies for human approval
One machine, measured.
Delivery notes go in: scanned paper or PDF. The AI extracts every field, edge cases get flagged for human review, and the result is fully digitised.
Pre-production. Real numbers.
From prior work and pilot conversations
Suppliers' delivery notes arrived by email or into a shared folder, one PDF per note, no two suppliers formatting theirs the same way, and someone had to read and key each one in by hand.
Each note goes from PDF to validated, structured data in under a minute, for a few cents in extraction cost.
Pricing
No fine print.
Most projects land between $3,500 and $11,000, depending on how many systems have to talk to each other. I am building five of them free this quarter, as pilot cases.
Automation
What's included
Discovery meeting
Flow design
Integration with the customer's existing tools
Testing and launch
30-day support after go-live
What's not included
Third-party subscriptions (Zapier, Make, and equivalents)
Migration of legacy data
Unplanned integrations
Scope changes after delivery
AI agents
What's included
Discovery meeting
Agent design (role, boundaries, fallback rules, human-in-the-loop checkpoints)
Model configuration
Initial tuning
Testing and launch
30-day support after go-live
What's not included
AI API costs (OpenAI and equivalents)
Creative content generation
Decisions with legal or financial liability
Scope changes after delivery
Tell me your case. I'll tell you honestly if I can build it.
- A 30-minute call where you describe the recurring task and I tell you whether I can build it.
- No slides, no pitch, and no follow-up emails afterwards.
- If it's not a fit, I'll say so.
30 minutes, free. If it's not a fit, you'll know on the call.
I'm picking 5 businesses this quarter to build pilot cases with.