Work outside the four areas
Pure infrastructure resale, ERP implementation and performance-marketing retainers sit outside our practice. We will name someone better placed rather than staff up to take it.
A/01
Artificial Intelligence
& Machine Learning
Generative AI and LLM application design, retrieval-augmented systems over your own documents, computer vision, natural-language processing, production chatbots, and AI-as-a-Service delivered behind an API surface you control.
- LLM & RAG
- Computer vision
- NLP
- Chatbots
- AIaaS
- MLOps
- Retrieval over your own documents
- A private index across the contracts, tickets and manuals you already hold, so answers cite a source rather than inventing one. Evaluation harness included — an assistant nobody can score is an assistant nobody should trust.
- Computer vision
- Inspection, counting, defect detection and document capture, deployed to the edge where bandwidth or latency rules out a round trip to the cloud.
- Natural language
- Classification, extraction and summarisation over messy real inputs — the email thread, the scanned PDF, the voice note — not the clean corpus a demo uses.
- AI as a service
- Model hosting, prompt and version management, cost controls and audit logging, behind an API you own and can move.
A/02
Blockchain
& Web3
Smart contract architecture and review, decentralised applications, tokenisation and NFT infrastructure across the Ethereum and Algorand ecosystems, and Web3 integration into products that already have real users.
- Smart contracts
- dApps
- Ethereum
- Algorand
- Tokenisation
- Smart contract design and review
- Specification before code, then tests that model the adversary rather than the happy path. We will recommend an external audit before mainnet and say so in writing.
- Decentralised applications
- Wallet flows, indexers and the off-chain services that make an on-chain product usable by people who do not care that it is on-chain.
- Tokenisation
- Asset representation, custody boundaries and transfer rules, designed alongside whatever compliance regime actually applies to you.
- Integration into live products
- Adding a Web3 surface to a product that already has customers, without a migration that puts existing balances at risk.
A/03
Metaverse
& AR / VR
Virtual environment construction, real-time 3D and gaming, immersive training and simulation, and enterprise XR applied to onboarding, field service and product marketing.
- Real-time 3D
- Training sims
- Enterprise XR
- 3D gaming
- Immersive training
- Procedures that are expensive, dangerous or rare to rehearse in the real world, built so completion and error data comes back to your LMS.
- Enterprise XR
- Field-service overlays, guided assembly and remote assistance on the headsets and phones your people already carry.
- Real-time 3D
- Configurators and virtual environments that hold frame rate on mid-range hardware, because that is what your audience owns.
- 3D and gaming
- Interactive experiences for marketing and product launch, with an asset pipeline your team can keep feeding after we leave.
A/04
Data &
Predictive Analytics
IoT and event ingestion pipelines, warehouse modelling, predictive and forecasting models, and the dashboards operators actually keep open all day.
- IoT pipelines
- Forecasting
- Warehousing
- Dashboards
- Ingestion pipelines
- Device and event streams that survive the network dropping, with replay so a bad afternoon does not become a permanent hole in the record.
- Warehouse modelling
- A model the business can read, with definitions agreed in writing — most reporting disputes are definition disputes.
- Forecasting and prediction
- Models scoped to a decision somebody actually makes, monitored for drift, with an honest baseline to beat.
- Operational dashboards
- Built for the operator who stares at them all day: dense, fast, tabular figures, and designed empty and error states.
Outside the four
What we will not take on
A short list, because a vendor who claims everything is a vendor who has not decided anything.
Referred out
Declined
Engagements we cannot stand behind
A deadline that only works if nothing goes wrong, or a scope that depends on a claim we cannot evidence. We would rather lose the work than write that proposal.