// Case studies

Proof, not promises.

The 5% of AI projects that reach production have one thing in common: they learn. Here is the work - across finance, construction, manufacturing and more - that crossed the GenAI Divide and shipped.

500+
data scientists in the network
2017
proving AI in production since
9+
industries delivered into
95/5
on the right side of the divide
Industry
18 results
// Featured

Selected work, end to end.

Each built to reach production and to be owned by the client - not left as a prototype.

World's first AI-powered timber design software
World-first

World's first AI-powered timber design software

A generative design engine that sizes and specifies timber structures automatically - collapsing days of manual engineering into minutes.

Read case study
Transaction-level carbon footprint for a green FinTech
Carbon per transaction

Transaction-level carbon footprint for a green FinTech

A model that attributes a carbon footprint to every transaction, giving a green FinTech disruptor real-time environmental signal at the point of spend.

Read case study
Predictive ML

Predictive machine learning to forecast FX fluctuations

Time-series models that forecast currency movement, giving the trading desk an evidence-led edge on FX exposure.

Read case study
AI-driven design pipeline in e-commerce
Design pipeline

AI-driven design pipeline in e-commerce

An automated design pipeline that turns customer photos into finished, print-ready artwork - removing the manual bottleneck between order and fulfilment.

Read case study
Carbon calculator for the earliest design phases
Early-phase carbon

Carbon calculator for the earliest design phases

A calculator that brings embodied-carbon data into the earliest design decisions - where it has the most leverage to cut a building’s footprint.

Read case study
Waste reduction in hard-drive manufacturing
Less waste

Waste reduction in hard-drive manufacturing

Machine learning applied to a high-precision manufacturing line to find the patterns behind scrap - and cut waste at the source.

Read case study
// All case studies
LLM-powered quotations and reporting in construction
Instant quotes

LLM-powered quotations and reporting in construction

An LLM trained on ten years of historical quotations that produces accurate quotes for new clients automatically.

View
LLM in freight investment
Dynamic pricing

LLM in freight investment

“Freight GPT” integrated into a trading platform to price each shipment from live signals like congestion, seasonality and cargo type.

View
LLM for parking-ticket appeal automation
60 UK regions

LLM for parking-ticket appeal automation

An automated bot that advises ticket-holders on local regulations and decides appeals against a per-region decision tree.

View
LLM for automated press-release generation
−50% time

LLM for automated press-release generation

A domain-tuned LLM that drafts life-sciences press releases automatically - estimated to halve time spent on manual content production.

View
LLM for customer-services automation
50% → 80% accuracy

LLM for customer-services automation

An LLM that reads call-centre reports to identify the faulty appliance part - so the right part arrives on the first visit.

View
LLM-powered semantic search for marketing
Instant retrieval

LLM-powered semantic search for marketing

A semantic-search system that lets a market-research platform find relevant information instantly across vast, varied document archives.

View
Information extraction for a Big-4 accountancy
NLP extraction

Information extraction for a Big-4 accountancy

An NLP strategy for a forensics department: extracting information across Word, Excel, PowerPoint, email and PDF to speed document processing.

View
HSBC - Project Pulse
First SME cash-flow app

HSBC - Project Pulse

Forecasting models and mobile components for the first cash-flow prediction app for UK SMEs, built with PhD-level experts.

View
Enhancing Kx Systems with machine learning
ML in kdb+

Enhancing Kx Systems with machine learning

Open-source ML libraries built into kdb+ so users can call models directly from one of the world’s most widely-used financial databases.

View
AIaaS platform

Sentiment-analysis AI test-drive platform

Fujitsu’s AI-as-a-Service platform, launched with a sentiment-analysis tool that lets brands measure sentiment towards their offering.

View
Demand forecasting with temporal fusion transformers
<10% MAPE

Demand forecasting with temporal fusion transformers

TFT models predicting item-level demand to under 10% error for a US restaurant chain - so stock and resources match what’s actually coming.

View
Machine vision to detect electric vehicles
Beyond-human accuracy

Machine vision to detect electric vehicles

A machine-vision system that identifies electric vehicles in large CCTV datasets from European motorways - at higher-than-human accuracy.

View
// Who we work with

A network of 500 data scientists working across finance, healthcare, retail, marketing, manufacturing, construction and more.

Sainsbury’sFujitsuSuezSeagateHappiest BabyNRCanKXStaircraftBEAROculaCrown & PawNVIDIAHSBCSymetriUnited UtilitiesIBMQWEBFirst DerivativesZebraSteelEyeSainsbury’sFujitsuSuezSeagateHappiest BabyNRCanKXStaircraftBEAROculaCrown & PawNVIDIAHSBCSymetriUnited UtilitiesIBMQWEBFirst DerivativesZebraSteelEye
“We invited Brainpool to participate in the initial definition workshop for Fujitsu's AI ecosystem in Europe. They really brought a level of knowledge and enthusiasm to the session. All stakeholders were thoroughly impressed with their experience in machine learning projects and their business model agility - I would not hesitate to recommend them!”
Fujitsu

Leigh Schvartz

Head of Product Innovation, Fujitsu UK

// Your move

Cross the divide. Own your AI.

From pilot to production in weeks - tuned to your business, deployed in your cloud, owned by you. Let’s talk about the project on your desk.