Plan a pilot
Startwell AI private compute core, voice hub, microphones and earpods

The working commercial model

Private AI.Predictable by design.

Start hosted when the business is tiny. Move to your own private compute when the team, workload or data boundary demands it.

Indicative commercial model · subject to pilot scope and final hardware specification

In partnership withNVIDIA

Hardware OEMSCAN

Private AI infrastructure · configured for Startwell AI

Two deployment routes

Start at the sizeyou are today.

A five-person company should not need to buy an appliance before it can prove value. Startwell Hosted removes that barrier while keeping a clear route into dedicated on-premise compute.

01 / Startwell Hosted

For the tiny business.

We host and operate the intelligence layer in an isolated managed environment. You get the company memory, approved connectors and voice workflows without buying a Spark on day one.

  • Designed for 1–5 enabled people
  • No private compute purchase upfront
  • Isolated company workspace
  • Approved connectors and governed memory
  • Meeting hardware and earpods optional
  • Defined migration path to your own Core
Working subscriptionFrom £750 / monthIndicative · onboarding and optional hardware separate
02 / Startwell Core

Your AI. In your building.

A dedicated Spark-based Core runs the company neural net under your control. Add another capacity unit as the organisation, models and workloads grow.

  • Designed from 5–10 enabled people
  • One private AI Core per ten people
  • On-premise company memory and models
  • Air-gapped and regulated options
  • Room and personal voice hardware sized separately
  • Multi-Core expansion without changing platform
Working subscription£2,500 / monthPrivate compute hardware + onboarding separate

Deployment planner

Size the whole system.Compute and capture.

Choose the enabled people, equipped meeting rooms and personal voice users. Teams of five or fewer begin on Startwell Hosted; larger deployments move onto dedicated private compute.

Recommended deploymentStartwell CoreDedicated private compute for 20 enabled people
150100
0510
050100

Physical sizing rule1 hub per equipped room · 6 clip-on mics per hub · 1 earpod set per personal voice user

Dedicated AI cores2one per 10 enabled people
Charging hubs2one per equipped room
Clip-on microphones12six per charging hub
Earpod sets6one per named voice user
Platform subscription£5,000per month · indicative
Annual platform£60,000before hardware + onboarding
Physical deployment2 rooms2 hubs · 12 mics · 6 earpod sets

Planning illustration, not a quotation. Hardware is priced separately once room coverage, spares, charging, resilience and the final product specification are confirmed. Compute capacity also depends on models, workloads and concurrency.

What you pay for

Three clearcommercial layers.

No mystery token bill. No buried data trade. The deployment is separated into assets you own, the platform we operate and the work needed to adopt it safely.

  1. 01
    One-time

    Hardware

    Private compute, hub, microphones, earpods and any network or resilience equipment required.

  2. 02
    One-time

    Onboarding + assurance

    Deployment design, security baseline, data boundaries, connector approval, configuration and team launch.

  3. 03
    Recurring

    Platform subscription

    Startwell AI software, model and memory orchestration, updates, guardrails, support and the agreed operating capacity.

How the deal moves

Prove it small.Scale what works.

01 / Frame

Pick one painful question.

Agree the users, systems, guardrails and measurable outcome.

Commercial workshop
02 / Pilot

Run it on real work.

Deploy a bounded, paid pilot with read-only connectors first.

Fixed pilot scope
03 / Deploy

Host it—or install the Core.

Small teams begin managed. Larger or controlled teams move directly onto private compute.

Hosted from £750 · Core £2,500/month*
04 / Expand

Add teams, not complexity.

New ten-person units, department packs and approved connectors join the same company memory.

Capacity-led expansion

*Indicative working model. Final proposal follows technical sizing, security review and hardware specification.

Start with a paid pilot

One use case.One team. Real evidence.

We will choose the hosted or private-compute route, define the guardrails and give you a clear proposal for the smallest deployment that proves value.