Pre-launch model testing
Early testing and feedback on frontier models from leading AI labs, before their public release.
Whatever you’re
building, we can help.
Leemer Group is an AI and technology company in Waterford, Ireland. We’ve worked with AI labs and startups from around the world, and our civic software is used by ministers and TDs in Ireland. A model, a product, a website or the admin that eats your week: the engineers behind our own software help across every layer of the stack. The Waterford company, the European AI lab and software engineering each have their own page.
WORKED WITH / AI LABS AND STARTUPS AROUND THE WORLD
Early testing for venture-backed startups
Hands-on testing and product feedback for startups backed by leading investors, ahead of launch.
Software used by ministers and TDs
DáilDex, our civic software, is used by government ministers and Teachtaí Dála in Ireland.
About DáilDexBuilders and coders ourselves
We design, code and ship our own products: Critique.sh for reviewing AI-written code, the Leemer Chat workspace, DáilDex and the Leemer Labs models.
You’re the idea. We build it.
Square Pigeons, our design-led software studio, builds software for people and companies. Bring the idea in a sentence and we take it through specification, design, engineering and launch.
WHAT WE CAN HELP WITHBring us
Bring us
the problem.
Most people arrive with a problem, not a service name. These are the ones we hear most. If yours isn’t here, ask anyway.
- We want to use AI but don’t know where to start
- We need an assistant that answers from our own documents, privately
- Our team loses hours to admin, copy-and-paste and email
- We have to know our AI feature works before we launch it
- We’re releasing a model and want honest, expert testing
- We need a custom or fine-tuned model, served fast and cheaply
- We have an idea but no product yet
- Our software project is stuck, late or hard to change
- Our website is slow, dated or hard to update
- Nobody finds us on Google or in ChatGPT
- Our numbers live in twenty different spreadsheets
- Our tools don’t talk to each other
- We’re starting a business and need it all set up properly
- We need a brand, a launch film or a pitch deck
- Our engineers want coding agents without losing quality
- We need senior technical judgement, not a full-time CTO
- We serve Irish speakers and need AI that works in Irish
- We need to get ready for the EU AI Act
TECHNOLOGIESThe tools we
The tools we
work in.
We pick the stack for the job, not the other way round, and we hand it over in a form your team can run.
AI models
- OpenAI
- Anthropic Claude
- Google Gemini
- Llama
- Mistral
- Qwen
- DeepSeek
- GLM
- Kimi
- Whisper
- OpenRouter
- Hugging Face
Training & inference
- PyTorch
- Transformers
- LoRA & PEFT
- vLLM
- SGLang
- llama.cpp
- CUDA
- Triton
- Quantisation
AI applications
- Retrieval (RAG)
- Vector search
- Agents & tool use
- MCP
- AI SDK
- Evaluation harnesses
- Voice & speech
Web & product
- TypeScript
- React
- Next.js
- Node.js
- Python
- Tailwind CSS
- Three.js
- Remotion
Data & back end
- PostgreSQL
- pgvector
- Supabase
- Redis
- REST & GraphQL APIs
- Stripe
- Analytics & dashboards
Cloud & delivery
- Vercel
- AWS
- Google Cloud
- Cloudflare
- Docker
- GitHub Actions
- Playwright
- Self-hosted GPUs
Business tools
- Google Workspace
- Microsoft 365
- Notion
- Slack
- HubSpot
- Shopify
- Zapier
- n8n
SERVICES / WHERE MOST WORK STARTS
01
AI systems
Assistants, agents and document workflows that fit how your organisation actually works, with private deployment when your data should stay with you.
We design, build and ship AI systems for a defined job: answering from your own documents, drafting and checking routine work, or running a background process end to end. Where the data is sensitive, the system can run on infrastructure you control.
- Assistants and agents for a specific team or process
- Search and question answering over your own documents
- Private and self-hosted deployment
- Model adaptation and fine-tuning for a narrow task
02
AI evaluation & assurance
Independent tests for an AI feature before you rely on it: what it gets right, where it fails and how you would know.
We build the test suite around your use case, run it against the model or product, and write up the results in plain language. The aim is a claim you can defend, not a score to quote.
- Task-specific test sets and held-out checks
- Failure analysis and red-teaming
- Model and vendor comparison on your own data
- A written report with evidence, limits and next steps
03
Product sprint
From an idea to a working, launched product. A written specification first, then a focused build with independent review.
We start by writing down what the product has to do and how we will know it does it. Then we design, build, review and launch a first version, so you leave with working software and the reasoning behind it.
- Specification, requirements and acceptance tests
- Design, build and launch of a first version
- Independent code review before release
- Optional brand, identity and launch pack
Leemer Labs · Born2 API
Our own models, by API.
Born2-9B, our code-review specialist, is in research preview on the Born2 API: €0.08 input and €0.20 output per million tokens, with €5 of free credit on every approved account. Inference is managed by Critique.sh, in Europe.
EVERYTHING ELSEEvery side of
Every side of
the business.
From the first idea to the systems that run it every day, each with its own page.
Irish-language & multilingual AI
Chat, translation and content for organisations that serve Irish speakers, evaluated with fluent speakers rather than assumed to work.
AI readiness audit
A short, written assessment of which processes are worth automating, which are not, and what to do first.
Web & software development
Websites, dashboards, internal tools and apps, built and reviewed to the same standard as our own products.
Automation & integrations
Connect the tools you already use, so routine work moves on its own, with AI only where it earns its place.
AI-assisted engineering
Help for engineering teams adopting coding agents, with review habits that keep the evidence in front of the person deciding.
Launch films, brand & identity
Launch and product videos, a modular identity system and a brand kit, the same work we did for ourselves.
Model training & ML engineering
Custom and fine-tuned models, dataset work, hosting and fast inference, down to GPU kernels, from the team behind Leemer Labs and Born.
Idea shaping & validation
Turn a rough idea into a written concept, test it with the people it is for, and decide what to build first, or whether to build at all.
Business setup & operations
The working foundations of a new or growing business: domain, email, accounts, payments, documents and the everyday tools, set up properly once.
Web design & UX
Sites and product interfaces designed around what visitors need to do, accessible, fast and easy for your team to keep up to date.
SEO, AI search & growth
Be found in search engines and AI assistants: technical SEO, structured data, content that answers real questions, and measurement you can trust.
Data, dashboards & analytics
Bring scattered spreadsheets and systems into one reliable picture, with dashboards and reports that answer the questions you actually ask.
Fractional technical leadership
Senior technical judgement without a full-time hire: architecture, vendor choices, hiring help and a plan your team can follow.
Civic & public-interest data
Source-linked search and explanation tools over public records, for newsrooms, charities and public bodies.
WATCH / WHAT WE CAN DO FOR YOU
82 seconds, with captions. Music and voice start only when you press play. The documents, findings and pages shown are illustrative. Read the transcript
How an engagement works.
- 01
Brief
A conversation about the problem, the people affected and what a good result looks like. You can draft the idea first in Build Brief.
- 02
Scope
A written scope: what we will build or test, what we will not, and how we will agree it is done.
- 03
Build
Focused work in short cycles. You see working software or evidence early, not a reveal at the end.
- 04
Handover
Code, documents and the reasoning behind decisions, in a form your team can run, change and own.
WATERFORD, IRELANDTell us what
Tell us what
you’re building.
Describe the problem and where you are now. We’ll reply with questions, not a sales script. Enquiries open as an email draft that you review and send yourself.
Questions before you write.
What AI and software services does Leemer Group offer in Ireland?
- From Waterford, Leemer Group designs AI assistants and agents, evaluates models before you rely on them, and builds websites, dashboards, internal tools and apps. Irish-language AI is tested with fluent speakers. A written specification comes before a build, and every release gets independent code review. Start at the services index or by email.
Do you work with organisations outside Waterford?
- Yes. The company is based in Waterford, Ireland, and works with organisations elsewhere in Ireland and with AI labs and startups further afield. Model access for the group’s own research can run on European infrastructure. There is no claim of offices in other cities. The first step is an email, which you review and send yourself.
How do I hire a technology company in Waterford?
- Write to hello@leemerlabs.ie or use the contact page, which opens an email draft with the service you choose. Describe the problem and where you are now. The group replies with questions. Nothing is submitted to a server on your behalf, and no price list is published on the site. A Build Brief draft can come before the first conversation.
LG / ADVISORThink it through
Think it through
with us first.
Building a company, with or without AI? Talk it over with our advisor. It knows what we build, how we work and where we can help, from the first idea to hosting your data in Europe or on your own premises.
Claude Haiku 5.5 by default, with a Born2-9B preview from Leemer LabsAn AI can be wrong. It cannot send anything for you, and it does not give legal, tax or investment advice. Please avoid sensitive details. Your messages go to our model provider to produce a reply. When it searches the web, a short search query goes to DeepSeek’s search service. This website does not store them unless you choose to save a copy in your own browser.
What are you trying to build?