Inteeka
Security advisory.PHP2Shell affects the software behind most websites. Free check: is yours exposed?Check now ›

2026, in review

Built. Shipped.
Running every day.

This year we designed, built and ran the software behind two real businesses: a high-volume print and production company we operate ourselves, and a hospitality training and compliance client. Here is what we shipped, why, and what it changed.

changes shipped to production this year on one platform alone
4,800+
changes shipped to production this year on one platform alone
of them in a single week
540+
of them in a single week
automated test files guarding that work
3,300+
automated test files guarding that work
end-to-end customer journeys checked in a real browser
65
end-to-end customer journeys checked in a real browser

Most studios show you mock-ups. We show you software that took orders, booked consultations, trained staff, ran a production floor and fixed its own problems, every single day, all year.

How a feature is born

Nothing here started as a feature list.

Every item on this page began as friction someone felt. We find it, fix it in small steps, ship it the same day and keep looking after it.

From the floor

A feedback button on every internal screen. Staff report a snag from the page it happens on, and it becomes a ticket with the context attached.

From customers

The questions that keep arriving by email, chat and phone. If people keep asking, the product should answer.

From the numbers

Checkout steps, funnels, recorded visits and the journeys people abandon. We look for where intent leaks away.

From the owner

A daily review of what slowed the business down. The best brief is the thing that annoyed someone this morning.

  1. 1

    Spot

    Find the friction: a repeated question, a manual step, a lost sale.

  2. 2

    Build

    Small, reviewable changes, with the business rules written down as tests.

  3. 3

    Ship

    Previewed, checked and live the same day. Switches decide when customers see it.

  4. 4

    Look after

    Tests, alerts, audits and backups that keep it working long after launch.

01 · Customer facing

Commerce that sells while you sleep.

For our print and production business, the shop had to answer every question a customer has before they ask it: what will it cost, will my file work, what will it look like, and when will it arrive.

Instant pricing, in plain words

Customers type the job the way they would say it and get a real price straight away.

How we spotted it
Quote requests arriving by email for jobs the site could already price. People did not want to fight an options form.
Why we built it
Answer at the moment of intent, not a day later in an inbox.
How we built it
A language model reads the request and maps it onto the real catalogue. The price always comes from the same engine as checkout, never from the model.
How we look after it
A library of real customer phrasings is replayed against every change, so a request that priced correctly yesterday still does today.

What it changed for us

Fewer quote emails for standard work and a shorter path from first visit to basket.

What it could do for you

Let customers ask in their own words, answered from your own data and your own rules.

Files checked at the door

Upload a file and get an answer in seconds: size, bleed, resolution, colour and page count, in plain English.

How we spotted it
The same handful of file problems caused most of the back-and-forth and most of the reprints.
Why we built it
Catch problems before payment and production, not after.
How we built it
An automated pipeline inspects every upload, scans it for malware and explains each issue in plain words. A person steps in only when one is genuinely needed.
How we look after it
Real problem files become test fixtures. Every rule has a test, and fair-use limits keep the free check free.

What it changed for us

Far fewer reprints and much less email ping-pong before a job can start.

What it could do for you

If you receive files, forms or documents from customers, they can be checked the moment they arrive.

Proofs and one-click approval

Customers see exactly what will be produced, approve it online, and nothing moves until they do.

How we spotted it
Approvals lost in email threads, and doubt about which version was signed off.
Why we built it
One clear record of what was approved, by whom and when.
How we built it
Each order gets a secure approval page: the proof on a plain background, approve or ask for a change, and reminders that stop the moment it is done.
How we look after it
Approval pages are pinned by snapshot tests, so the page a customer approved cannot quietly drift.

What it changed for us

Faster sign-off and an audit trail for every job.

What it could do for you

Any sign-off that matters: designs, documents, quotes or contracts.

Delivery dates you can trust

Checkout shows the date first, lets customers choose a production speed for each item and prices every delivery option live.

How we spotted it
"When will it arrive?" was the question we were asked most.
Why we built it
Certainty sells. A clear date beats a vague promise.
How we built it
A working-day calendar with cut-offs, per-item speeds, live carrier options, split delivery to several addresses, and switches that put new options on sale when the business is ready.
How we look after it
Date logic is tested around midnight, weekends and bank holidays, and switches flip without a release.

What it changed for us

Fewer "where is my order?" messages, and customers choose, and pay for, the speed they need.

What it could do for you

Turn your operational limits into a promise customers can see before they pay.

02 · Behind the scenes

Operations that run themselves.

The work customers never see: the inbox, the quotes, the production floor and the machines. This is where most of the hours were, and where most of the hours came back.

An AI inbox with human sign-off

Incoming emails are read, matched to the right order and drafted into a reply. A person checks every claim before anything is sent.

How we spotted it
The inbox was the bottleneck: the same lookups for every message.
Why we built it
Reply fast without losing accuracy or our own voice.
How we built it
Agents gather the facts, draft in the sender’s own voice and stage the reply in a staff console. There is one send path, and nothing leaves unreviewed.
How we look after it
Every house rule on tone, wording and what we never promise is written down and tested.

What it changed for us

Faster, consistent replies, with people spending their time on judgement instead of lookups.

What it could do for you

Customer support at scale, with people firmly in control.

A live production floor

Job sheets with scan codes, live boards on the wall and a dispatch station that labels the parcel, books the carrier and tells the customer.

How we spotted it
Walking the floor to find out what was due, and re-typing details at dispatch.
Why we built it
Everyone should see the same truth at the same time.
How we built it
Every job sheet carries a code. A scan moves the job on, the boards and kiosks update live, and a dispatch scan prints the label, books the collection and sends tracking.
How we look after it
Screens run unattended on locked-down devices, and alerts tell us when something sticks.

What it changed for us

Less chasing, fewer missed dispatches and a calmer floor.

What it could do for you

Make the status of every job visible, from the desk to the van.

Machine-ready files, automatically

Approved artwork is prepared for the presses and cutting tables without anyone touching it.

How we spotted it
Skilled hours spent preparing files by hand, the same way, every time.
Why we built it
Hand preparation is slow, and it is where mistakes creep in.
How we built it
A toolkit turns approved files into production files to each machine’s exact conventions: layout, cut paths, registration marks, layers and names. Then it drops them where the machine collects them.
How we look after it
Output is compared with known-good files on every change.

What it changed for us

Production starts sooner and file-preparation errors are rare.

What it could do for you

Wherever people re-key or reformat data for the next system, software can do it exactly the same way every time.

Custom quotes on autopilot

Bespoke requests are analysed, priced, written up and followed up, ready for a person to approve.

How we spotted it
Bespoke quotes took longest, and were the easiest to forget to chase.
Why we built it
Speed wins custom work.
How we built it
Analysis drafts the lines from the request and its attachments, the quote document is generated, and gentle follow-ups are scheduled.
How we look after it
Every quote keeps its history, and guardrails limit what automation may change.

What it changed for us

Custom quotes go out sooner, and fewer slip through the cracks.

What it could do for you

Any business that quotes bespoke work, from trades to professional services.

And dozens of small wins.

Feedback on every screen

Report a problem from the page it happens on. It arrives as a ticket with the context.

New-order announcements

The team hears about new work the moment it lands, wherever they are.

Missed-call follow-ups

An email goes out after a missed call, and cancels itself if a later call connects.

A staff portal

Time off, training, documents to sign and policies, in one place.

Reports that read themselves

Sales, funnels and visits summarised into what to do next.

Uploads scanned for malware

Every file from the outside world is checked before anyone opens it.

03 · Client work

Training and compliance, made software.

For a hospitality training and compliance client we build and run the whole platform: consultations, courses, incident records, on-site audits, safety records and a shop. Every feature below is live and used by their customers.

Bookings and video consultations, in-house

Customers pick a slot, pay if it is a paid session, and join a video call that has already checked their camera, with a phone line if it fails.

How we spotted it
Consultations depended on a third-party scheduler, and calls lost their first minutes to camera and microphone trouble.
Why we built it
Own the journey from booking to follow-up, and make every call start on time.
How we built it
Our own slot engine, a video room per booking, a test-your-setup page, a switch-to-phone fallback, an intake form before paid sessions and calendar sync for the host.
How we look after it
Slot calculations are tested around time zones and fully booked days, and every booking email carries the setup link.

What it changed for us

Calls start on time with the customer’s context in hand, and the whole journey is ours to improve.

What it could do for you

Any business that sells time: consultants, clinics, tutors and advisers.

Self-paced courses and exams

Lesson-by-lesson courses with a fresh set of exam questions on every attempt, bought and taken at any hour.

How we spotted it
Customers wanted to train their staff on their own schedule, not when a trainer was free.
Why we built it
Training that sells, and is taken, around the clock.
How we built it
Courses are data: lessons, a rotating question bank and results. Each new course arrives as a draft for the subject expert to review before it goes on sale, and sign-in links survive corporate email scanners.
How we look after it
Because a course is data, an update is a review rather than a rebuild, and the expert signs off every question.

What it changed for us

A line of revenue that does not need a trainer in the room.

What it could do for you

Onboarding, compliance training or customer education, sold or given away.

Voice-to-text incident logs

Staff speak an incident into a phone. It is transcribed, filed against the right site and held for a manager to approve.

How we spotted it
Busy teams rarely have time to type, so records were written up long after the event, if at all.
Why we built it
Records should be made in the moment, which is exactly when an inspector expects them.
How we built it
A recorder in the browser with guided microphone permissions, transcription, a review screen with the audio beside the text, and an approval step.
How we look after it
Each failure we met on real phones, from blocked microphones to a recorder that never started, was fixed at the root.

What it changed for us

Incidents are recorded in the moment, without adding admin to a shift.

What it could do for you

Any team that has to log what happened: care, construction, retail, security or field service.

An offline-first audit tool

Tablet checklists and photos that save without signal, and a report emailed to the customer at the end of the visit.

How we spotted it
Audits were written up after the visit, and the places being audited often have no signal.
Why we built it
Finish the report before leaving the site.
How we built it
Every answer and photo is saved on the device first and synced when the connection returns; findings carry their detail inline, and the report goes to the customer with a thank-you email.
How we look after it
Built from real reports, with each improvement folded back into the checklist.

What it changed for us

Reports go out the same day, and follow-up work is offered while the visit is fresh.

What it could do for you

Inspections, surveys, site visits and maintenance rounds.

And the rest of the platform.

A site safety suite

Hazardous-substance assessments, equipment servicing, fire safety schedules with reminders, and traceability and recall.

Compliance becomes a routine with reminders, not a scramble before an inspection.

A shop that prints on demand

Apple Pay and Google Pay, approved materials sent straight to print with a signed link, and tracking emailed on dispatch.

Physical products sold with no stock and no manual hand-off.

An AI assistant across the site

Answers questions about services, the shop and courses, in the business’s own voice.

Visitors get answers at midnight; the team gets fewer repeat questions.

Follow-up that never forgets

Enquiries triaged in one view, consultations tracked from booking to outcome, and a gentle sequence of emails after every call.

Every enquiry gets followed up, without anyone keeping a spreadsheet.

Add-ons at the right moment

Extras are offered after the booking, on the confirmation page and email, so the first purchase stays simple.

More revenue per customer without making checkout harder.

A full security review, closed

Every high and medium finding fixed, row-level security on every table and secret scanning on every change.

Customer data protected to a standard we are happy to be audited on.

04 · Insights and opportunities

Every morning, a ranked list of what to do next.

We built a layer that reads what customers did yesterday, in chats, visits and checkouts, and turns it into opportunities ranked by value, effort and confidence.

Insights that turn into opportunities

Every morning, AI reviews yesterday’s chats, recorded visits and numbers, and a board ranks what to do next by value, effort and confidence.

How we spotted it
We had more data than time to read it: recordings, chats, funnels and sales, each in a different place.
Why we built it
Decide what to build next from evidence, not hunches.
How we built it
Scheduled jobs gather yesterday’s conversations, recorded visits and checkout steps, and a model summarises what stopped people buying. An opportunities board sorts every idea into a quick win, a big bet, prove first or fill-in, with a decision against each.
How we look after it
Recommendations are stored and tracked until they ship or are dropped, and a dry-run mode checks the analysis without sending anything.

What it changed for us

A short list of the next best things to build, waiting each morning.

What it could do for you

Turn the data you already collect into a ranked list of what to do next.

A daily chat review

Each conversation is linked to what happened next. The ones that did not convert become suggestions, each ready to ship with one command.

Visits watched for friction

Recorded sessions are reviewed each morning for the places people hesitate, misclick or leave.

Funnels that count real visits

Weekly reports count engaged visits, not every page fetch, so the numbers mean something.

Ideas with a verdict

Anyone can add an idea. Each gets a verdict: proceed, test first, needs data or park.

05 · Live operations

From alert to fix, while the issue is still small.

Real-time operational issues are spotted as they happen, analysed by AI, and fixed: at once when the fix is pre-authorised, or after a human review that leaves the change ready to ship instantly.

From alert to fix, in minutes

Live checks spot an operational problem, AI works out why and writes the fix, and it ships at once if it is pre-authorised, or after one human click if it is not.

How we spotted it
Problems surfaced as a customer complaint, hours after they started.
Why we built it
Fix issues while they are small, ideally before a customer notices.
How we built it
Live checks watch the moving parts, such as stuck emails, late jobs, failed hand-offs and device errors. Staff can flag a problem from any screen, even with a note on a job sheet. AI gathers the context, finds the likely cause and writes the fix as a ready change with its test.
How we look after it
Fixes in a pre-authorised class, such as retries and data corrections within set limits, run straight away and are logged. Anything else waits for a person, one click from live.

What it changed for us

Issues are fixed in minutes, with a record of what happened, why, and what changed.

What it could do for you

Operations that mend themselves, with your rules deciding what needs a human.

  1. 01

    Detect

    Live checks, alerts and a report button on every screen.

  2. 02

    Analyse

    AI pulls the records, the logs and the latest changes, and names the likely cause.

  3. 03

    Propose

    The fix arrives as a ready change with its test, or as a data correction.

  4. 04

    Decide

    Pre-authorised fixes run at once. Everything else waits for one human click.

  5. 05

    Ship

    Live in minutes, with a record of what changed and why.

06 · What is next

Ideas we are already exploring.

Not promises: the directions our own operations point to next, and the ones we would most like to build with a partner.

01

Agents on the phone

Voice agents that answer, book and take orders by phone, and pass the call to a person the moment it needs one.

02

Cameras at the packing bench

Measuring each parcel as it is packed, choosing the box and the carrier, and catching a missing item before it ships.

03

Scheduling that plans itself

Production and staff rotas planned from live orders, machine time and deadlines, re-planned when something changes.

04

Finance that reconciles itself

Supplier invoices matched to the jobs they paid for, with only the mismatches left for a person.

05

An integration hub for partners

Suppliers, couriers and resellers plugging into the same APIs and webhooks we use ourselves.

06

Modules others can use

The file checks, the AI inbox and the offline audit tool, packaged for other businesses under their own brand.

07 · Working together

Six ways we work with other companies.

Start small with a sprint, hand us a whole platform, or anything in between. Each comes with the same shipping and testing discipline.

A discovery sprint

Two weeks to find the friction that costs you most, and ship the first fix.

For example: A logistics firm maps its ten most common customer emails, and ends the sprint with drafted replies waiting for approval.

Build and run it for you

We design, build, host and look after a platform for your business, as we do for our training client.

For example: A training provider gets bookings, courses, audits and a shop on one platform, improved every week.

Embed with your team

Our engineers and AI tooling alongside yours, bringing the shipping and testing discipline with them.

For example: A retailer’s in-house team moves from monthly releases to daily ones, with tests and release records.

Connect what you already have

Email, payments, accounts, carriers and your CRM joined up, with agents doing the work in between.

For example: An events company takes a booking, raises the invoice, sends reminders and follows up, without re-typing anything.

License what we have built

Proven modules under your brand, connected to your systems.

For example: A facilities firm runs our offline audit tool for its inspectors, with its own logo on every report.

Partner on a product

Co-build with suppliers and partners over shared APIs, the way our training client’s shop sends work to print.

For example: A supplier and a reseller share orders, artwork and tracking through signed links instead of email.

What it could mean for you

Your business has its own version of every one of these.

Customers who ask in their own words

And get answers from your real data, instantly.

An inbox that drafts itself

With your team signing off every reply.

Checks at the door

Files, forms and documents validated as they arrive.

Operations that mend themselves

Issues found, explained and fixed, with your rules on what needs a human.

Field tools that work offline

Audits, inspections and reports, signal or not.

Training that runs itself

Courses, bookings and follow-ups, around the clock.

Bring us the process that eats your week.

Fifteen minutes, engineer to engineer. We will tell you honestly what we would build, and what we would leave alone.

Book a free 15-minute call