AI App Development
AI apps that run in production, not in a demo
We build new AI tools for the work that eats your team's week: search across your documents, assistants for your teams, and automation across your CRM, ticketing and data. Built for production from the first slice, with evals, logging and clear rules for your data.
Organisations we've built for
Best for
When an AI app is the right move
If one of these sounds like your team, an AI app is likely worth building.
Manual work
A costly workflow still runs on copy, paste and checking.
We automate the repetitive steps, such as reading, sorting, extracting and drafting, and keep your people on the decisions.
Internal assistants
Your ops, sales, support or finance team searches five systems for one answer.
An assistant that knows your documents and data, respects who may see what, and shows its sources.
Across your systems
Work moves between CRM, ticketing and spreadsheets by hand.
Agents that act in your existing tools through their APIs, with clear limits and a log of every step.
Pilot to production
Your AI pilot impressed in a demo but never went live.
We take it to production: evals, monitoring, access control and a handover your team can run.
What you get
What we build, and what you're left with
Less manual handling, faster turnaround and output your team can check.
AI web and mobile apps
New tools built around one workflow, for your staff or your customers.
Internal assistants
Role-based and permission-aware, in the tools your teams already use.
Search over your documents
Answers from your contracts, manuals and tickets, with links to the source.
Workflow automation
Agents that read, decide within set limits and act across your CRM, ticketing and data warehouse.
Every build includes
Code in your repository from day one
An eval set that scores every release
Logging, monitoring and guardrails, including PII filters
Documentation and handover, so your team can run it
How we deliver
Three steps from idea to a tool your team uses
The same three steps as every project we run, with a check on quality and data at each one.
01
Plan
The workflow, success criteria and risks, written down with a list of what we won't do. We agree which data the AI may see.
02
Thin slice
One case working end to end on your real systems early, scored against an eval set, with a demo every week.
03
Launch and handover
Monitoring, guardrails and PII filters, onboarding for your team, and documentation so they can run and change it.
Example projects
What an AI app project looks like
Three illustrative examples from sectors we haven't shown yet: the problem, what we'd build and how we'd measure it. They aren't client projects; client case studies are added as clients approve them.
Example · Insurance
Claims intake, from inbox to claims system
The problem
Claims arrive by email with photos and PDFs, and handlers spend the first hour of each case retyping them.
What we'd build
An app that reads each claim, extracts policy number, date and damage details into the claims system and flags what's missing. The handler checks and decides.
How we'd measure it
Time from arrival to a complete case, fields right first time on a checked sample, and handler corrections per case.
Example · Manufacturing
An assistant for maintenance technicians
The problem
The fix for a machine fault is somewhere in manuals, service bulletins and years of maintenance logs.
What we'd build
An assistant on the technicians' tablets that answers from those documents, shows the source page and records which answers helped.
How we'd measure it
Time to find the right procedure, correct sources on an eval set built with your senior technicians, and repeat repairs.
Example · Wholesale
Customer orders from email and PDF into the ERP
The problem
Orders arrive as emails, PDFs and scans in every customer's own format, and the sales office types them in.
What we'd build
An app that turns each order into a draft order in your ERP, matches customers' item codes to yours and leaves anything unclear for a person.
How we'd measure it
Order lines entered without edits, time per order, and orders entered on the day they arrive.
Security and data
Clear rules for your data and the AI Act
What the AI may see, where it runs and how its output is marked, agreed in writing before the build.
No training on your data.
We only use AI tools on your code and data under business terms that exclude training, and only with your approval.
You decide where it runs.
Your systems, your cloud account, and model providers and regions agreed with you, with EU hosting where you need it.
Built with the AI Act in mind.
Users are told when they're dealing with an AI, AI output is marked where the Act requires it, and every step is logged.
Sectors and technologies
Where we've built, and what we build with
Sectors
We've built production systems for hospitals, banks, cooperatives and a national geospatial institute, so we know the data and rules your AI has to respect.
Healthcare
Financial services
Logistics & mobility
Public sector & geospatial
Technologies
The tools we use most on AI apps. The model, the hosting and the region are agreed with you before the build.
Large language models via business APIs
Retrieval (RAG) and vector search
Agents and tool calling
Evals and monitoring
Python
React
React Native and Flutter
AWS
FAQ
Questions about AI apps
Related services
Often combined with
Build
AI-Native Development
Add AI to the product you already run, shipped feature by feature with evals and monitoring.
Data
Data Foundations & Analytics
One source of truth your reports and AI can rely on: pipelines, data models and dashboards.
Build
Custom Software Development
One system built around how you actually work, in place of spreadsheets, paper or a tool you've outgrown.