Email triage
Incoming mail sorted, routed to the right person and filed against the right client.
AI automation & agents
AI that clears the admin.
Agents that read the email, find the client, draft the reply, update the record and file the document, then wait for a person to say yes. Practical AI inside the systems you already run, not a chatbot bolted to your website.
Sound familiar?
The admin grows faster than the team.
Sorting email, typing invoice details in, writing the same replies and summarising the day: it’s work that has to be done and doesn’t need a person to do the first draft.
- Staff spend the morning sorting and forwarding email.
- Invoices and receipts are typed in by hand.
- The same twenty questions get answered from scratch every week.
- You’ve tried ChatGPT, but it doesn’t know your clients or touch your systems.
What gets built
The parts, shaped to you.
Pick the ones you need. Each is built around how your team already works, then joined up on one set of data.
Drafted replies
Replies drafted from your records and your tone, sent only after someone approves.
Document reading
Invoices, receipts and forms read into structured data, checked by a person before saving.
An AI teammate in your admin
Staff ask in a chat inside your own portal, and it carries out the task in your CRM, with approval and a full trail.
Summaries and briefings
A morning briefing of what came in, what’s overdue and what needs a decision.
Photo to data
Snap a photo, get a structured result: an item identified, described and priced for review.
Guardrails
Approval steps, audit logs and limits on what an agent can touch, so nothing changes without a person seeing it.
An honest first step
A short discovery to find where AI saves real hours, and a ‘not worth it’ list for where it doesn’t.
Connects to
- Gmail or Outlook
- Your CRM or ERP
- Google Drive & Sheets
- Claude
- OpenAI
- Gemini
- Google Lens
A good fit if
- Your team spends hours on email, data entry or repetitive replies.
- You have a system of record (or want one) for the AI to work inside.
- You want people to stay in control of what goes out.
Not the right call if
- You want AI for its own sake. If a plain automation or a better form fixes it, that’s what I’ll suggest.
Proof
Built before, running now.
Second-hand retail · Built & deliveredSecondhand Malta
Photo in, listing out: Google Lens identifies the item and Gemini drafts the copy and a price for staff to review.
Read the case study Software for photographers · Live · v1.5imagicA desktop product with a 30-tool MCP server, so AI assistants can run its whole pipeline.
Read the case study SEO & content agency · 2021–2025LJCAI drafting in an agency’s content workflow from August 2022, then the GPT API inside Google Sheets.
Read the case studyHow it runs
From first call to daily use.
The same five steps on every project, sized to yours.
- 01A short call or the quote builder
Tell me how the business runs today: the tools, the workarounds, the bit that drives you mad. The quote builder takes about five minutes and gives me a scoped brief.
- 02A fixed quote, in phases
I reply with a fixed price and a phase plan. The must-haves go first, so you are using the system while the rest is still being built.
- 03Previews as it’s built
You see the real system on a private link while it is being built, with your own data in it as early as possible. Changes are cheap at this stage, so this is where we make them.
- 04Your data moved across
Spreadsheets, an old system, email threads: the data comes with you, checked against the source so nothing is lost on the way.
- 05Launch, then care
Your team gets a walkthrough, and I stay on for fixes and new features, either on a monthly support plan or pay as you go.
FAQ
Questions, answered.
Will AI send things to my clients without us checking?
Not unless you want it to. The default is that agents draft and prepare, and a person approves before anything is sent or changed. Every action is logged.
Is our data used to train AI models?
Business API access from the major providers is not used to train their models by default. I build on those APIs and keep your data in your own systems; the details for your provider are set out in the proposal.
Which AI models do you use?
Whichever fits the job: Claude, OpenAI and Gemini models, and Google Lens or vision models for images. The system isn’t locked to one provider.
Can AI work with our existing CRM?
If it has an API, usually yes. If your CRM is spreadsheets and email, it often makes sense to build the system of record first and add agents on top.
Where do we start?
With one process that eats the most hours. Pick ‘AI & automations’ or ‘AI teammate’ in the quote builder and describe it, and I’ll scope a pilot.
Can AI projects get Digitalise Your SME funding?
AI is listed among the eligible digital solutions, and advisers report a separate AI allowance in the current call. Rules change between calls, so check the Malta page and confirm the current terms before relying on it.
Often built together
Related systems.
Custom CRM
Enquiries, clients, pipeline and follow-ups in one place, shaped to how you sell.
IntegrationsWix, Shopify, Gmail, WhatsApp, Stripe, calendars and accounts, synced both ways.
Custom ERPQuotes, stock, staff, jobs and invoicing on one shared set of data.
Quoting & invoicingQuotes from your rate card, invoices with VAT, deposits and automatic chasing.
How pricing works · Digitalise Your SME grants · All case studies
Start here
Get a fixed quote.
The quote builder opens with the parts on this page already switched on. Add or remove what you like, and I reply with a fixed quote.