AI and automation
AI features that take the routine work off your team
We build AI features into the systems we make, where they save time. Chat on your website, calls answered, quotes and documents drafted from your own data, and paperwork checked against the job, with a person approving whatever matters.
What it is, and where it fits
Where it tends to help first
- Enquiries arriving by phone, email and the website
- Quotes, letters and reports built from the same details each time
- Certificates, invoices and forms that need checking
- Notes, calls and photos that need turning into a record
- Anything somebody retypes from one place into another
If a task follows the same steps most days, it is usually worth a look.
AI is one tool among several we use. Where it does a routine job well, we build it into your system so the job gets done as part of the work, rather than in a separate app somebody has to remember to open.
Plenty of automation needs no AI at all. A reminder that goes out when a date comes round, or an invoice raised when a job is signed off, is ordinary software doing what it's told. We use AI for the parts that involve reading, writing or listening: an email to make sense of, a certificate to check, a quote to draft, a phone call to answer.
The aim is to automate as much of the routine as is sensible, so your team spends its time on the work and the customers rather than on the typing around them. Decisions that matter stay with a person.
What it can take on
The jobs we're asked about most. Most businesses start with one and add others once it has proved itself.
Chat on your website
An AI chatbot for your business website that answers questions from what you've told it, your services, areas, hours and prices if you want them shown, and takes the enquirer's details. The lead lands in your system with the conversation attached, so nobody types it up.
Calls answered
An AI receptionist that answers the phone when the office is busy or closed. It takes a message, answers the common questions, books where you allow it to, and puts urgent calls through to the person on call. Each call arrives as a written note on the right customer.
Quotes and proposals drafted
The proposal written from the job's own figures, your rates and the notes from the site visit, in your layout and your wording. Somebody reads it, adjusts it and sends it. More on quotes and pricing.
Documents generated
Reports, certificates, letters and job summaries drafted from the data already in the system, so the details match the record because they came from it. Filed against the job when they're done. See document and certificate stores.
Paperwork checked
A certificate, an insurance document, a supplier invoice or a form read and checked against the job or the asset. Dates, names, amounts and reference numbers compared, and anything that doesn't match flagged for a person to look at.
Details read in
Order details from an email, figures from a PDF, a reading or a serial number from a photo, pulled out and put on the right record, ready to confirm rather than type. One system we run watches a mailbox for incoming PDFs, reads a fixed set of fields off each one, checks every value against what the document should contain, and passes the result on to the next system automatically. Anything it cannot read with confidence is set aside for a person, rather than guessed.
Calls and notes summarised
A long call, a site visit or a thread of emails turned into a short summary on the record, with the actions picked out, so whoever picks it up next has the gist in a minute.
Inboxes sorted
A shared inbox read as it arrives, with each message tagged by type, matched to the customer or job it's about, and put in front of the right person. Routine replies drafted for somebody to check and send.
Chasing done
Polite reminders for missing documents, unsigned forms, overdue invoices and quotes that went quiet, sent on a schedule you set, and stopped as soon as the reply comes in.
Design work drafted
Layouts, visuals and marketing material drafted from your templates and brand: a flyer for a new service, a social post, a brochure page with the details filled in. Ready for you to change and approve.
Forecasts from your own data
What next month is likely to look like, based on the work already in the system, and a plain-English summary of the week for whoever runs the business. Alongside management reporting, not instead of it.
These are built into the systems we make, so they work alongside job and works trackers, compliance records, customer portals and field apps. Where the information lives in software you already run, they can often reach it through an interface or an export.
How we go about it
One job at a time
Start with one job. We pick the task that comes round most often and takes the most typing, build the AI feature for that, and let it prove itself on real work before adding the next.
Built into your system. The feature works on the records you already keep and puts its results back on them. There is no separate tool to log into and nothing to copy across.
A person approves what matters. Anything that goes to a customer, commits you to a price or changes a record that matters is drafted for somebody to check, not sent on its own. You decide where that line sits, and it can move as you get used to it.
Tested on your own work. Before it goes live, it runs on real examples from your business beside the way you do it now, so you can see what it gets right and where it needs a hand.
Every step recorded. What the feature read, what it drafted or flagged, and who approved it, kept on the record like any other change.
How a build runs from the first call to go-live is set out under how we work.
How it gets built, and how it gets run
| Fixed price | Scope, price and dates in writing before anything starts, with what the feature will and won't do written down as plainly as everything else. |
|---|---|
| Timescale | A first feature added to an existing system is usually weeks rather than months. As part of a new system, it's planned into the same dates. |
| Your data | Your system stays hosted in the UK. Where a feature sends text to an AI provider, we use providers that don't train their models on your data, under written terms, and we tell you which one before it's switched on. More on hosting and support. |
| Security | Cyber Essentials certified. A feature only sees the records it needs for its job, with the same permissions as the rest of the system. |
| Running costs | AI features carry a usage cost from the provider. We tell you what it is likely to be before you decide, and how it's covered is agreed at the start. |
| Changes | New wording, a new document type, another kind of check. What the monthly covers is agreed at the start, so nobody works it out on the day. |
Ask about our free Business Systems Assessment
An hour with whoever runs the day to day. Clear recommendations to act on.
We look at how the work actually runs and what it runs on, then send you a list of what we would change, what each change would do for the business, and what it would take. In person or online, whichever suits you. Nothing to pay, nothing to sign.
- One hour
- In person or online
- Recommendations we talk you through
Honest answers
Will the AI make decisions on its own?
Only where you want it to. Routine steps, such as tagging an email or filing a document against a job, can run by themselves. Anything that goes to a customer, sets a price or changes something important is drafted for a person to approve, and you decide where that line sits.
Is our data used to train AI?
No. Where a feature uses an AI provider, we choose providers that don't train their models on the data sent to them, and the work is done under written terms. Your system and its records stay hosted in the UK, and the provider is named to you before anything is switched on.
Does it ever get things wrong?
Sometimes, which is why it's built the way it is. It drafts and flags, a person checks what matters, and every step is recorded. We test it on your own examples before go-live, so you've seen how it handles your work before anybody relies on it.
Can you add AI to a system we already have?
If we built the system, yes, it's an addition rather than a rebuild. If it's software you already run, it depends on whether that software lets another system read and write its records, and we find that out before the price is fixed.
Does an AI receptionist replace our office staff?
It's there to answer when they can't: when the lines are busy, at lunch, out of hours. Urgent calls go straight through, and everything else arrives as a note for the team to pick up.
Talk to us
Tell us which job your team does over and over, and where the information for it comes from. That's usually enough for us to say whether an AI feature would help, and where we'd start.
Ring, 8am to 6pm Monday to Friday and Saturday mornings, or use the form below. Pick what you'd like to keep track of, add a line if you want to, and leave a name and a way to reach you.