How we work · Forward deployed engineering

Installed by forward deployed engineers. Paid on results.

A forward deployed engineer sits inside your perimeter, installs MindMap OS on your hardware and stays until the outcome is measured. The first accelerator is live in 6–9 weeks. No seats, no time-and-materials, no headcount billing.

Book a scoping call →See MindMap OS →Customer stories →

Inside your perimeterYour hardware6–9 weeksPer outcomeNo T&M

Definition

Forward deployed engineering, defined.

AI built outside the perimeter and handed over at the door rarely survives contact with the core system. So the engineer moves in. A forward deployed engineer is a MindMap engineer who works from inside your network, installs MindMap OS on hardware you control, configures the accelerators to your data, and stays until the outcome the subscription is billed on — claims processed, customers onboarded, exams evaluated — is measured on your numbers. Your own teams work alongside them from the first week and are trained to run the estate; the audit trail is configured with your compliance function, so the evidence a regulator asks for is produced by the accelerator itself.

Weeks one to nine

From the first badge to the measured outcome

Weeks 1–2 · Land inside the perimeter

The forward deployed engineer works from your network, on your access policy, next to your operations team. Together you pick the first accelerator from the Library — the one nearest your highest-volume, highest-value work — and agree the outcome that will be measured.

Outcome defined. Accelerator chosen. Data sources mapped.

Weeks 3–6 · Install and configure on your data

MindMap OS installs on your hardware. The accelerator is configured to your documents, systems and rules; connectors, guardrails and the human-in-the-loop queue are wired to your processes; every decision is logged inside your perimeter.

Running on your data. Exceptions routed to your people.

Weeks 7–9 · Hypercare and the measured outcome

The engineer stays until the number is in: claims processed, customers onboarded, documents read — whatever the outcome you subscribed to. Your team is trained to run it; the bill follows the result.

Live in production. Measured. Billed on the result.

The reuse loop

How one accelerator travels

Built at customer one

Forward deployed engineers build the first version inside a customer's perimeter, on the customer's data, to the customer's rules — an engineering project, measured on its outcome.

Returned to the Library

What is reusable — the model binding, the connectors, the guardrails, the audit trail — returns to MindMap OS as a versioned tile. The customer's data, configuration and bespoke work stay with the customer.

Configured at the next customer

The same tile is installed again — configured, not rebuilt — in weeks rather than months. The first deployment of an accelerator is an engineering project; the fifth is a configuration exercise. The customer pays the same for the outcome.

How it is priced

If the outcome cannot be counted, it cannot be billed.

Every subscription to MindMap OS is paid on results. The engineer is how the work gets installed; the outcome is what you pay for.

Per transaction

A document processed, a cheque cleared, a case closed — metered on volume, inside your perimeter.

Example: Claims Router, per claim routed.

Per outcome

A share of the measured saving or recovery — we get paid when it works.

Example: revenue-leakage recovery.

Fixed subscription

An annual subscription to MindMap OS and the forward deployed engineers who operate it inside your perimeter — the Centre of Excellence model.

Engineers are priced into the subscription, never billed by the hour.

FAQ

Forward deployed engineering — the questions buyers ask

What does a forward deployed engineer actually do?

They install MindMap OS on your hardware, configure the accelerator to your data and processes, wire the connectors and guardrails, run the review queue with your team, and stay until the outcome you subscribed to is measured. They work inside your network on your access policy — the same screens your operations team sees.

Do we have to give access to our production environment?

The engineer works under your identity provider and access policy, like any contractor you already trust with production. Nothing leaves the building: MindMap OS installs on your hardware, and inference, retrieval and fine-tuning run inside your perimeter with egress blocked and logged.

What hardware do we need?

The Model Estate is fine-tuned and quantised to run on hardware you control — a GPU box in your data centre, your private cloud tenant, or, for Bhasha, an edge device offline. The scoping call sizes it against the accelerator you pick first.

Who owns what at the end?

You keep your data, your infrastructure and keys, your configuration and any bespoke work. MindMap keeps the accelerator, the model family and the deploy kit, licensed to you for the life of the deployment. What we learn returns to the Library as a versioned tile, never as your data.

How is the work billed?

On results — per transaction, per outcome or a fixed subscription for MindMap OS and the engineers who run it. No seats, no time-and-materials, no headcount billing.

Pick the first accelerator. We install it. You count the outcome.

20-minute scoping call with the engineers who will walk your perimeter, your systems of record and your regulator's expectations with you.

Book a scoping call →See MindMap OS →