SmartifierSmartifier
Menu

Services

We help you decide whether a digital solution, such as AI or IoT, is right for your business and what it would bring. If it is, we build it.

How an engagement runs

  1. Assess

    One week

    We review your data, systems and security, then write a short brief on what is worth building. It is yours to keep.

  2. Build

    Four to twelve weeks

    A small team builds the smallest system that works properly, in your own code repository from day one.

  3. Hand over

    Your choice

    We train your team to run the system, or we run it for you on a fixed monthly retainer. There is no lock-in.

Four areas of work

All four are backed by the same funded research, and each ends in a working system.

AI Agents & Chatbots

Assistants that connect to your own systems, so they can book, search, answer and pass difficult cases to a person.

Ask me what TAIMWAS did, or how our engagements work.
Learn more →

AI Automation

We find the repetitive manual work an LLM can handle safely, then build the smallest system that does it reliably.

Learn more →

Agentic Engineering

For problems too big for a single agent: several agents working together over your private data, with testing and monitoring.

Learn more →

Cybersecurity

Threat modelling, secure system design, penetration testing and security training that end in a hardened, working system.

$ smartifier-scan --target api.example.com
→ Checking TLS configuration… ✓
→ Scanning exposed endpoints… ✓
→ Checking auth headers…
⚠ Missing Content-Security-Policy
⚠ X-Frame-Options not set
→ Dependency audit…
2 high, 4 moderate CVEs found
→ Report saved to ./report.json
Learn more →

Our approach

Open models, on your terms

We build on open-weight models by default, because they keep costs down and can run where your data already lives. We look at your requirements first and recommend where the model should run. If a frontier model suits the job better, we can build on that instead.

Cost-effective

Most open-weight models are free to use, including commercially. You pay for the hardware or hosting they run on, with no licence or per-seat fees.

Secure

We support self-hosted, on-premises and cloud-based deployment, behind your existing access controls and with audit logging from the start.

Privacy-aware

When the model runs inside your environment, prompts and documents are not sent to an outside AI provider.

Open-weight model in your environment

Your environment

Your dataYour applicationOpen-weight model

Nothing leaves your environment. No licence fee for the model.

Hosted frontier model

Your environment

Your dataYour application

Provider

Model

Prompts and documents cross to the provider. Billed per use.

Open-weight models in your environment compared with hosted frontier models
Open-weight, your environmentHosted frontier model
CostUsually no licence fee. You pay for hardware or hosting.Billed per use, and generally more expensive.
Where your data goesIt stays on your servers or in your cloud account.It is sent to the provider and handled under their terms.
Lock-inLow. The model can be swapped or moved.Higher. You depend on one provider’s pricing and changes.
ControlYou decide which version runs and what is logged.The provider decides when the model changes.
Who runs itYou do, or we do for you. It needs maintaining.The provider does.
Best suited toSensitive data and cost-conscious projects.Tasks that need the most capable models available.