// AI FOR
Enterprise AI Consultancy & Agentic AI Engineering — London
We help enterprises pick the few
AI bets that matter — and get them
into production, not stuck in pilots.
Applied AI Engineering · London · Est. 2022
OUR
WORK
[ CLICK LOGO FOR CASE_STUDY ]
// TESTIMONIAL
PARTNERSHIP
3+ years
Dr Oliver Bogler
Director, NCI's Center for Cancer Training
“I have worked with Barnacle Labs for more than 3 years on a project at the National Cancer Institute (NCI)... they treated the work like a collaboration... if you are looking for a partner in AI, I couldn't recommend them more.”
WHO
WE ARE
// BACKGROUND
Our founders have been building with modern AI for over a decade — from leading AI at IBM Watson to securing early GPT-3 access in 2020.
// THE MISSION
We build production AI systems for high-stakes environments—including the US Federal Government, financial services, global enterprises, and biotech firms. We specialise in the hard parts that quietly derail AI projects, bringing creative problem-solving rooted in experience, not guesswork.
// ENTERPRISE_AI
HOW
WE HELP ENTERPRISES PUT AI TO WORK
We help organisations move from AI ambition to production — especially where workflows are complex, context matters, and control is non-negotiable.
// OUR_PROCESS
Find the opportunity
Identify the AI bets worth making, where value will come from, and what it will take to deploy responsibly.
Build the system
Design and ship production AI systems that work inside real business processes, not just demos.
Make it stick
Equip your team to govern, own, and evolve what gets built so capability lasts beyond the initial engagement.
// OUR_SPECIALISMS
Where Barnacle goes deep
// 01
AI coworkers, not just copilots
Copilots assist. Coworkers act. We build Minerva: an entirely new interface where your team can invoke, interact with, and steer AI agents to execute complex workflows.
Best for: High-volume operations and complex internal workflows
What you get: Intuitive UIs designed specifically for humans to manage and control multi-agent workflows.
[ LEARN_MORE → ]// 02
Agents with memory that gets better over time
Useful agents need more than context windows. They need memory that can work out what matters, store it in the right way, and retrieve it when it counts. That's what Alexandria is built to do.
Best for: Agentic systems that need continuity, judgment, and useful memory over time
What you get: Memory systems that help agents prioritise, store, and retrieve the right context
[ LEARN_MORE → ]// 03
Sovereign AI for high-control environments
Public AI infrastructure isn't always enough. We design sovereign deployments with full control over data, hosting, and operational risk.
Best for: Healthcare, finance, government, and strict data compliance (HIPAA, GDPR)
What you get: The full AI stack deployed natively on your own local or UK/EU infrastructure
[ LEARN_MORE → ]// DEFINE THE PROBLEM BEFORE YOU COMMIT THE CODE
Executive AI Workshop
A half-day board session for leadership teams that need a shared view of AI, 3–5 prioritised opportunities, and clear criteria for where to start — and where not to.
[ LEARN_MORE → ]// GET_STARTED
Not sure where to start?
Talk to an engineer about your workflows, constraints, and opportunities — and we'll show you where we'd focus first.
WHY
ENTERPRISE TEAMS CHOOSE US
Production-First
Working software early, not slides.
High-Stakes Experience
Government, financial services, telecoms, healthcare, biotech.
Built for Handover
Your team can own and evolve what we build.
Technical Depth
Agentic systems, memory, orchestration, guardrails.
Beyond the Brief
Creative problem-solving, not order-taking.
READ
OUR THOUGHTS
Featured blog posts
BLOG POST
NEWBuilding a Production Multi-Agent System for Document Writing
by Thibault Boutet
Architecture of a briefing generation system using 15 specialized AI agents to transform PDFs into structured documents.
BLOG POST
Building a Reproducible Multimodal Pipeline
by Kevish Napal
Part-1 in a series of posts about proteomics, cancer biology, and AI-driven solutions for oncology.
BLOG POST
Understanding Cancer and Telomerase: From Biology to New Treatments
by Kevish Napal
Part-2 in a series of posts about proteomics, cancer biology, and AI-driven solutions for oncology.
BLOG POST
Mean Pooling Beats Attention: Predicting Telomerase Activity from Whole-Slide Images
by Kevish Napal
Part-3 in a series of posts about proteomics, cancer biology, and AI-driven solutions for oncology.
BLOG POST
Understanding and Processing CT Imaging for Stroke Detection
by William Auroux
A practical guide to turning raw brain CT into training-ready data.
BLOG POST
The Dimension Dilemma: Why 2.5D Models Outperform 3D CNNs for Stroke Classification
by William Auroux
Lessons & Experiments on training deep learning models on 3D medical data.
AI FROM THE
TRENCHES
Conversations with leaders who've actually led AI initiatives
Can a Foundation Model Read the Human Brain?
with Kris Pahuja from Piramidal
Can AI Save Scientists From Information Overload?
with Dr Oliver Bogler
From IBM Watson to Barnacle Labs: Lessons From Building AI Too Early
with Duncan Anderson and JD Wuarin
Inside Shopify’s Scout: The scrappy AI tool that went all the way to the boardroom
with Rich Brown from Shopify