Built to work in production.
Five examples from earlier assignments. Each one shows the problem, the solution and what it delivered. You can also read how it was built. Clients and systems are anonymised.
Faster answers to quote requests and tenders
A team had to answer large numbers of recurring questions in every tender. The answers already existed, but were spread across hundreds of earlier documents.
The challenge
Specialists in product, legal and sales answered the same questions again and again, in every tender and questionnaire.
The solution
An internal system that searches earlier answers and documents, writes a first draft and shows the source for every answer. A specialist reviews and approves before anything is sent.
The result
- Much less repetitive work.
- Earlier knowledge can be reused.
- Specialists spend their time on questions that need their judgement.
How it was built
- Earlier answers were organised into a shared knowledge base of several thousand questions and answers.
- Search works on the full original documents, so no detail is lost along the way.
- A test suite of real questions was used to measure quality, choose the approach and decide when it was time to change direction.
- The solution was handed over with automatic updates, quality tests and documented failure scenarios.
Facing a similar challenge? Book 30 minutes →
New leads assessed instantly, without an expensive external service
A sales team received new business leads through a web form and needed to know which ones were worth prioritising.
The challenge
New leads were assessed by hand with the help of an expensive external service, and the best leads reached the sales team late.
The solution
Our own system analyses every new lead the moment the form is submitted and sends an assessment to the sales team’s channel.
The result
- The external service could be cancelled.
- The sales team gets an assessment the moment a lead comes in.
- More teams started using the solution.
How it was built
- The public form never reaches the system directly. Requests go through a protected layer with authentication, rate limiting and input validation.
- Logging was designed for GDPR: personal data is replaced with anonymised values.
- Each assessment is stored as structured data that other teams can reuse. More teams started using the solution without anyone pushing for it.
- The solution passed a formal security review with no findings and has been monitored from day one.
Facing a similar challenge? Book 30 minutes →
More customers who complete sign-up
A global service where new customers apply digitally, across several markets with different payment habits and price levels.
The challenge
Applications were dropping out of a digital sign-up flow. In several markets, the payment options and pricing rules did not fit the market.
The solution
The flow was rebuilt visually and technically, with full validation, more payment providers and pricing rules set per market.
The result
- More completed applications in the same flow.
- Fewer false rejections once a missing integration was in place.
- Payment options and pricing adapted per market.
How it was built
- The review process was followed on site, end to end, from application to decision.
- That revealed a silent fault: a missing integration was behind a large share of the incorrect rejections. No dashboard had shown it.
- Pricing rules and payment options were set up per market instead of one shared setup for all.
Facing a similar challenge? Book 30 minutes →
Partner data you can trust
An internal system received messages from a large number of partners and forwarded them to the right team.
The challenge
Messages from partners were silently disappearing in an internal system, and nobody had an overview of which connections were broken.
The solution
A view that reconciles four data sources showed exactly where the errors were. They were fixed, and the teams involved got automated requests to verify their details.
The result
- Errors show up before they become problems.
- Teams verify their details automatically.
- The system handles higher volumes.
How it was built
- Four sources were checked against each other in a single view, so missing and incorrect connections became visible.
- The requests to teams were automated, with checks that stopped incorrect details before they were saved.
- A filter removes sensitive content before messages are forwarded.
- When volume hit a technical limit, the intake was rebuilt to spread the load.
Facing a similar challenge? Book 30 minutes →
A quarter of the cloud cost, and approved at audit
A data platform in a regulated environment, where data has to be handled under strict requirements and pass regular audits.
The challenge
A data pipeline in a regulated environment was expensive to run and hard to get through audits. Elsewhere, there was legacy cloud infrastructure nobody had a full overview of.
The solution
The pipeline was rebuilt to meet the data-handling requirements, and the legacy infrastructure was mapped so it could be decommissioned safely.
The result
- Cloud costs fell to a quarter.
- The pipeline passes audit.
- Legacy infrastructure could be decommissioned safely.
How it was built
- The data flow was redesigned to meet regulatory requirements and cost less to run.
- Cloud resources were mapped and security reviewed before anything was switched off.
- A streaming data platform made data available for analysis in near real time.
Facing a similar challenge? Book 30 minutes →
Do you have a problem that should be solved better?
Tell us what you want to achieve. We help you decide what is worth building and how to get it into production.