Problems we’ve solved

We have an urgent deadline. How would you approach it?

Urgent delivery · 2025

We delivered a working integration in 7 days. The partner then signed.

The challenge

An air cargo company needed a new partner contract. Before signing, the partner required a working, documented application programming interface (API) to test the connection between their systems. The work was normally expected to take about a month.

What we did

We focused the first delivery on what the partner needed to make a decision. We built 18 API endpoints for parcel and shipment data, with security controls and documentation, and put them into production on day seven.

The result

The partner signed once the API was ready. We then built the rest of the platform on that foundation, without starting again.

• 18 endpoints available through the API

Can our partners sell the platform under their own brands?

Partner sales · 2023

Partner and reseller accounts added in 3 weeks.

The challenge

A cybersecurity platform owner wanted to grow sales through partners and resellers. The platform served direct customers only, and the owner did not want a separate system for every partner.

What we did

We added a partner, reseller and customer structure to the live platform, with separate data-access permissions. Partner accounts carry their own logo and support email. Resellers can use that identity or set their own, and partners receive billing data ready to import.

The result

The client can sell directly and through partners and their resellers from one platform, with each partner presented under its own brand.

• 3 tiers: partner, reseller and customer
• Added in 3 weeks while the platform stayed live

How can we control the cost of AI?

AI · 2026

AI explanations capped by a daily allowance for each company.

The challenge

Users of a cybersecurity platform needed security findings explained in plain English. Paying for every AI request meant usage needed limits, with a way to respond if the provider changed its prices or terms.

What we did

We set a daily request allowance for each company, capped answer length and reused saved answers to repeated questions. Once the daily allowance is reached, the platform blocks new requests before sending them to the provider.

The result

Usage stays within defined limits. When we changed provider in 2026, it took one configuration change, with the previous provider retained as a fallback.

• 10 requests per company per day, by default
• 500 words per answer, at most
• 1 setting to switch provider

Can you modernise our system without disrupting customers?

System modernisation · 2021 – present

We’re rebuilding the platform in stages while customers keep using it.

The challenge

As a cybersecurity platform grew over five years, its technical foundations needed to change to support new services without disrupting existing customers.

What we did

We rebuilt in stages, running new components alongside the existing system on the same database. Feature development and sales continued throughout. The work to bring the application together is now in its final stage.

The result

Customers kept access to the service throughout the upgrades. We continue to develop the platform after five years of working together.

What if a partner changes how our systems connect?

Partner integrations · 2024 – 2025

A new partner integration, with no pause in trading.

The challenge

A package forwarder’s main partner changed how parcel data was exchanged through its application programming interface (API) and set a deadline to retire the old version. The business had to update its system and records before the cutoff.

What we did

We ran both integrations together and migrated eight years of data in batches of 1,000 records. We updated the system in stages and sent any record with an uncertain owner for staff review.

The result

The transition was completed after about five months of parallel operation, with no maintenance shutdown and no increase in support requests.

• 8 years of records
• 1,000 records per batch
• ~5 months side by side
• 0 minutes of downtime during migration

What if an external service our system relies on shuts down?

Continuity · 2024

A provider shut down. The platform kept running.

The challenge

A cybersecurity platform received one month’s notice that its breach-data provider was closing. Part of the service customers paid for depended on that data.

What we did

We connected a second data source within about two weeks, using the platform’s existing data to check for breaches. A permanent replacement was live around three months after the notice.

The result

Customers retained access throughout the transition. When the provider changed again in 2026, we used the same approach to keep the service running.

• 0 days of lost access

How do you preserve accuracy when moving our data?

Data integrity · 2026

We moved 453 tasks, 716 comments and 356 files, and invented no history.

The challenge

While moving four years of our own work from a third-party tool, we could not extract the history showing who changed each task’s status and when, despite testing every route available to us.

What we did

We moved the data we could verify and kept inferred observations separate from recorded actions. We attributed no change to a member of staff without evidence in the original data.

The result

We migrated 453 tasks, 716 comments and 356 files. Confirmed records stay separate from observations, and we invented no activity history.

• 453 tasks
• 716 comments
• 356 files
• 0 actions attributed to staff without evidence. Notes stay separate from verified change history.

Can you reduce repetitive manual work?

Automation · 2018 – present

An invoice now takes under a minute, down from a working day.

The challenge

A company forwarding customers’ overseas purchases prepared invoices by hand from spreadsheets. Combining parcel costs, fees and discounts took a working day per invoice.

What we did

We checked the pricing rules against real invoices, then built a tool that combines 23 types of charge, cost and discount in one click. It calculates shipping from parcel weight and dimensions, and customs duty for each item.

The result

The business owner reports that preparing an invoice now takes under a minute instead of a working day.

• Invoice preparation time, reported by the owner

Can you turn complex procedures into a clear workflow?

Workflow design · 2025

Staff see only the permitted options across 40 statuses.

The challenge

An air cargo handler worked with 40 parcel statuses, each with its own conditions and sequence. These procedures were undocumented, even though the partner required updates in the correct order.

What we did

We documented the procedures in plain language and agreed them with the client before development. We then made the system show staff only the permitted options, with rules kept in settings that can change as the business does.

The result

We delivered four partner-requested changes through settings, without rewriting the parcel-processing logic.

• 40 status codes
• 17 steps
• 120+ rules
• 4 rule changes made in settings, with no new release

Can our customers run phishing drills without a technician?

Security testing · three months

A button in the platform replaced manual commands.

The challenge

An employee security-testing platform needed a tool that creates simulated login pages. That tool required someone to type commands by hand and had no application programming interface (API) for the platform to control it.

What we did

We built a service that takes requests from the platform and operates the tool automatically, saving each customer’s settings and restoring them after a restart. Reaching stable operation took about three months.

The result

Customers can launch a test directly from the platform, without typing commands or handling the tool’s technical setup.

Can we keep documents consistent across branches?

Consistent branch documents · three system generations

One contract update reaches every store.

The challenge

A retail group kept a separate copy of its sale contract in each store. With nine documents in the contract, an update meant changing multiple copies of the same content.

What we did

In the third generation of the system, we replaced those separate copies with one template used by every store. Contract terms can now be updated in one place.

The result

Every store uses one shared template, with no separate local copy of the contract template to maintain.

• 1 shared template for every store

How do you handle errors after launch?

Resolving system errors · June 2021

Unmatched parcels now go to staff for review.

The challenge

Weeks after a warehouse system launched, some parcels appeared not to exist when staff scanned their tracking numbers, despite being right in front of them.

What we did

We traced four errors to one cause: tracking numbers arrived in different formats, sometimes with extra data. We rebuilt parcel identification and customer matching, and added a review queue when ownership could not be confirmed.

The result

Unmatched parcels go to staff for review instead of being assigned to a customer on a guess.

How do you catch failures that could go unnoticed?

Data reliability · automated checks

Price data stays protected when an update fails.

The challenge

A pricing platform collects data from 188 competitor websites. If a site stops returning prices, an empty result could be mistaken for a change in the market.

What we did

We made the platform keep its last valid data and report ten types of failure. At 100 unresolved faults for one site, it pauses collection from that site and resumes automatically once the faults are resolved.

The result

An empty result cannot erase saved prices. The platform flags the failure for review and keeps the last valid data available.

• 10 failure types
• Collection paused after 100 unresolved faults

Can the app work without an internet connection?

Offline access · two live stores

The product catalogue stays available offline.

The challenge

Customers needed to browse products inside stores and in areas with weak mobile reception, without losing access when the connection dropped.

What we did

We made the catalogue an app customers can install on their phones. After the first visit, it saves the pages and files needed for browsing and refreshes that saved copy with each new release.

The result

Customers can keep browsing offline across every brand using the system. Sending an enquiry requires the connection to return.

• 671 files saved on the phone for offline browsing

Can staff move between systems without signing in again?

Point of sale · until 2021

Breeder records open directly from the store account.

The challenge

Animal retail staff use a tablet to show customers breeder details and licences. Opening a separate platform meant signing in during the conversation.

What we did

We connected the breeder information platform to each employee’s store account. Once signed in to the store system, staff can open a breeder profile without entering another password or sharing an account.

The result

Staff can continue the conversation without stopping to sign in, and every breeder profile visit is recorded under the employee’s name.