How we work
Four engagement models
2–3 weeks
AI Readiness Assessment
Leadership teams deciding where AI should go first.
Outcome: A prioritised use-case list, a hosting and model recommendation, and a business case you can take to your board.
- Workshops with business and IT owners
- Use-case scoring: value, effort, data readiness, risk
- Infrastructure review and GPU sizing estimate
- Model choice per use case (open-source in-house, optional foundation-model backup)
- 12-month roadmap and budget range
8 weeks
Agent in 8 Weeks
Teams with one clear use case who want it in production, not a slide.
Outcome: One agent live in production on your systems, measured against the baseline agreed in week 1.
- Week 1: baseline metrics and acceptance tests
- Weeks 2–5: build on a domain accelerator, integrate with your systems
- Weeks 6–7: evaluation on your data, security review, user testing
- Week 8: go-live, runbook and handover
6–12 weeks build, then ongoing
Private AI build + support
Organisations whose data cannot leave their premises.
Outcome: Open-source LLMs running on your own servers behind a model gateway, with your first use cases built on top and supported by us.
- GPU server sizing and setup (yours, on-prem or in your data centre)
- Model serving, gateway, access control and monitoring
- First 1–3 use cases built and integrated
- Support, model updates and monthly usage and cost report
Monthly, minimum 3 months
AI Pod
Companies with a backlog of AI work and no in-house AI team yet.
Outcome: A dedicated cross-functional pod from the Surge AI team working through your backlog in two-week sprints.
- AI architect, engineers and an evaluation specialist
- Sprint planning with your product owner
- Shared backlog, demos every two weeks
- Knowledge transfer to your team built into every sprint
Principles
What is the same in every engagement
- A written baseline and acceptance tests before any build starts
- Demos on your data every two weeks
- Your data stays in your environment unless you decide otherwise
- Your data, configuration, prompts and tuned models are yours; accelerator licensing is set out in the contract
- Documentation and knowledge transfer, so you are never locked in
Book a working session
A working session with our architects on one of your use cases. You bring a process and sample documents; we leave you with a scoped plan: model, hosting, integrations, timeline and cost range.