Custom AI agents for business
Build your AI workforce
A team of AI agents that takes on the jobs your people repeat every day. They read each request, price it, draft the reply and book the work, inside the systems you already run, and stop for your yes on anything that matters.
One agent per job, working as a team
These are the agents from the film. In your business they take on the jobs your people repeat every day, hand the work along to each other, and stop for your yes on anything that matters.
Reads every request properly, pulls the history from your CRM and works out what the customer actually needs.
You: Nothing, unless something looks unusual.
Prices each job from your own rate card and flags anything outside the limits you set.
You: Approve anything above your discount limit.
Writes proposals, quotes and replies in your style, and holds them until you approve.
You: Read it, change a line, send it.
Books site visits, calls and meetings into the right diaries and sends the confirmations.
You: Turn up.
Chases overdue invoices politely, matches payments and tells you who needs a call.
You: Make the calls that need a voice.
Screens applications against your criteria and shortlists the people worth meeting.
You: Interview the shortlist.
Your team is built around your jobs, so yours might look quite different. Most businesses start with two or three agents on the work that hurts most, then add more once they have earned their place. Talk it through on a free call
A chatbot answers. An automation repeats. An agent decides.
Most people meet AI as a chat window, so that is what they picture. Here is the same enquiry landing in your inbox, handled three different ways, so the difference is obvious.
A chatbot
Holds a conversation using a script or a knowledge base. Genuinely useful on a website, and we build them, but the conversation is where it stops.
The enquiryTells the customer your opening hours and offers to take their details.
Nothing has actually been done yet.
An automation
Follows the exact steps you drew, in the same order, every time. Fast, reliable and cheap to run, until something turns up that the path did not allow for.
The enquiryCopies it into your CRM, tags it and sends a templated acknowledgement.
It cannot handle the exception, and business is mostly exceptions.
An agent
Reads the situation, works out what is needed, then uses your data and your systems to finish the job. When it should not decide alone, it escalates to a person with everything already prepared.
The enquiryWorks out what they are asking for, checks stock and current pricing, drafts the quote in your tone, and flags the one unusual line for you to approve.
The customer gets an answer in minutes, not on Monday.
A chatbot talks to your customers. An agent does the job your customer is waiting on.
Pick a job and watch your agents run it
Pick a job below and watch the team hand it along, inside your systems and using your data, from the request to the finished work sitting in front of you for approval.
1 of 6 watched
Your request
✓ Sent
Everything is done, this is the bit that needs a person
Illustrative examples based on typical business processes. Timings are conservative estimates of what these jobs take, not results from a specific client. The real builds are further down the page.
AI agents for every part of your business
Pick a department to see the agents we build most often, how they hand work along, and what changes for your team.
If your week looks like this, AI agents pay for themselves
Industry matters less than bottleneck. Find the one that sounds like your week.
You quote or estimate a lot
Every quote takes a skilled person hours, and the slow ones lose the job to whoever answered first.
Common in: Builders, installers, fabricators, agencies, manufacturers
Digger, Penny, ScribbleYou process orders by hand
Orders arrive as emails, PDFs and calls, and somebody rekeys every one into your system.
Common in: Distributors, wholesalers, ecommerce
Digger, PennyYour inbox runs the business
Everything lands in one place, and things get missed on busy days.
Common in: Service, property, professional services
Digger, Scribble, BookerYou book appointments all day
Someone spends their day arranging, confirming and rearranging, and the phone never stops.
Common in: Clinics, salons, garages, trades
BookerYou chase money every month
Invoices go overdue, reminders go out late or not at all, and cash flow suffers for it.
Common in: Any business that invoices on account
ChaseYou hire and onboard often
Screening, scheduling interviews and induction eat weeks that nobody has.
Common in: Logistics, hospitality, care, construction
Picker, BookerAI agents doing real jobs today
Three builds where agents do the reading, drafting and deciding, and a person signs off.
A quoting agent reads the survey, prices the job from their own rates and drafts the quote. The estimator checks it and sends it the same day.
An enquiry agent reads each event request, checks availability and drafts the quote. Still being built out, already handling the first response every time.
We ran it on ourselves first. A reporting agent pulls our numbers together every week, drafts the summary and flags what moved.
Your agents work inside the software
you already run
No new platform, nothing ripped out. If it has an API we can usually reach it.
Behind them we use the leading models and orchestration platforms, and we pick per job rather than tying you to one vendor. If a better model arrives next year, your agents move to it.
From first conversation to agents doing real work
No months of discovery and no strategy deck that sits in a drawer. Every stage ends with something you can hold, and the first agent is working inside month one.
The blueprint
Two to three days of senior time inside your business. We sit with the people doing the work, record their screens, and map how a job really travels from enquiry to cash.
Your first agent goes live
PennyQuoting agentLive- Read the request from Lumen Dental
- Checked six sites against the CRM
- Priced each site from your rate card
- Draft ready, waiting for your yes
We do not build for six months and hope. We take the single best opportunity from the blueprint and put it into production, in your systems, on your data, with your approval gates.
Build out the workforce
With one agent proving itself, we work down the roadmap. More agents, deeper integrations, dashboards so you can see what they are doing, and the custom CRM or web app underneath where it is needed.
Run, improve and hand over
Discount rule tightened after two edge cases. Nothing else needed you.
Agents drift. Models change underneath them, integrations move, and your business does not stand still. We monitor, tune and extend them, train your team, and document everything.
Limited offer · AI assessment and strategy
Your AI strategy, half priceStart with an AI strategy
Find out exactly where AI will pay off in your business, before you spend a penny building it.
Offer ends in
At half price
Only 5 places left
Not sure yet? Talk it through with us first, free and with no obligation.
What you get
- Kick-off callGoals and priorities agreed in 45 minutes
- Time with your teamTwo to three days of senior time inside the business
- Every opportunity scoredWhat each one returns against what it costs
- A phased roadmapCosts, savings and the quick wins first
- A full walkthrough60 minutes with you and your team
- Yours to keepUse it with us, someone else, or on your own
- £495 credited backIn full, if we do any of the work within 90 days
After the strategy
From £1,000a month for ongoing support
From £2,500for one-off builds
Fixed priceagreed before anything starts
Why designbox
Plenty of people are selling AI who have never run a business. We have, from Essex, for over a decade.
You get people, not a portal
One account manager who knows your business and who you can actually ring. You never explain yourself twice.
We build what the agents run inside
Websites, stores, CRMs and reporting are our day job. Most AI firms can only work on top of whatever you already have.
Ten years of being accountable
ICO registered, a decade of building and running business-critical systems, and clients we started with years ago. AI is new. Answering for the outcome is not.
We will tell you when it is no
Some processes should not be automated and some businesses are not ready. The blueprint exists so we can say so early.
Your data, your rules, your systems
Two questions come up in every single conversation: what happens to our data, and what happens if we leave. Here are both answers up front.
You own everything we build
No lock-in and no hostage-taking. If we are worth keeping, it will be because the work is good.
Security and control
Model and processing arrangements are covered in writing before anything touches your live data, and we go through them in the blueprint.
AI agents, answered honestly
Can’t see yours? Ask us anything on a quick call, no obligation.
What is an AI agent, in plain English?
What can AI agents do for a small business?
How much does it cost to build an AI agent?
Is this just a chatbot with a different name?
How is an agent different from an automation like Zapier?
Will it work with the software we already use?
What actually happens to our data?
Are we replacing people with this?
How quickly will we see something actually working?
Do we need technical people on our side?
What does it cost?
Do we own what you build?
What happens when an agent gets something wrong?
What if we want to stop?
Let’s find where AI actually pays off in your business
Thirty minutes, no hard sell, and an honest answer either way.
