Services
Fixed scope. Real deliverables. You own the result.
Every engagement ends with something you keep: working private AI, staff who use it daily, and a roadmap a non-technical owner can read. Pick one; most clients start with the audit.
Service 01
Private AI Setup
The problem
Per-seat fees, token meters and data shipped to strangers have become the default way to adopt AI — and that model is structurally leaky. The bill scales with usage, the price rises at renewal, and your contracts and client records live on someone else's server under a privacy promise you can't audit.
Our approach
Benchmark first, buy second. We test candidate open-weight models (Llama, Qwen, DeepSeek, Mistral families) on your actual tasks, size hardware to the benchmark — often a single GPU workstation, sometimes a small server — then run a fixed-scope install: provisioning, secure network setup, model deployment, integrations with the tools you already run (email, files, CRM, helpdesk, documents), and security hardening by engineers who do real incident response. Tested under load with your team's workloads before sign-off.
Deliverables
- Sized hardware bill of materials + complete on-prem installation
- On-prem assistants: chat, drafting, document Q&A with citations
- Integrations into your existing tools, scoped and documented
- Security hardening: firewall rules, access control, update policy, proven backups
- Admin runbook and a recorded handover session for your IT contact
Service 02
Employee Enablement
The problem
The best AI in the world is a costline item if the team treats it as homework. Adoption fails when tools arrive without context, without playbooks, and without guardrails — so people ignore it, or start pasting payroll data into public chatbots. Both outcomes cost you money.
Our approach
We build an assistant around the job each person already does — the office manager drafting quotes and SOPs, the sales coordinator prepping proposals, the bookkeeper summarizing statements — and train non-technical staff hands-on, in your office, on your own documents. Written playbooks with copy-paste prompts, plus guardrails: rules for sensitive data, escalation paths, and a no-surprise policy for what the assistant may and may not touch.
Deliverables
- Role-specific assistant kits (prompts, templates, workflows) per team
- Plain-language playbook library, kept in your document system
- Hands-on training sessions for all staff, in your office, on your data
- Guardrail pack: sensitive-data rules and escalation paths
- 30-day office-hours window plus an adoption scorecard review
Service 03
AI Strategy & Vetting
The problem
Every vendor now has an AI story, and a per-seat price list attached to it. Separating capability from slide-deck — and spotting the vendor quietly training on your data — is a full-time job you don't have, on a budget you can't guess-and-check.
Our approach
A readiness audit of your data, systems and workflows; an ROI-first roadmap ranked by payback period, quick wins first; and vendor vetting with a genuine privacy and security review — data residency, training clauses, breach notification terms, subcontractors, and the exit plan for when you leave. Every recommendation comes with a build-vs-rent call and the arithmetic behind it.
Deliverables
- Written readiness report: systems, data flows, and risk map
- Costed roadmap: 90-day quick wins + 12-month plan, ranked by payback
- Vendor scorecards with privacy/security review — including the tools you already pay for
- Data privacy & security review: what may go to the cloud, what must not
- Optional quarterly check-ins to keep the roadmap honest
