Technical Delivery Scoping Guide

6 min read

This page cover the technical delivery for Customer Hosted Tenants (CHT) on Azure, AWS and OpenShift — from kickoff to Go-Live, and the ongoing operation after.

At a glance

One offering: Customer Hosted Tenant (CHT). The customer owns the cloud environment, Unique operates the platform.

CHT

Time to Go-Live

6 months from kickoff

Unique effort, setup

12 SE-weeks

Unique effort, ongoing

40% FTE per tenant

Customer effort, setup

0.5 to 1 day per week

Customer effort, ongoing

5 to 10% FTE per tenant

Total ongoing effort

45 to 50% FTE per tenant

Customer must have

Approval authority

Unique effort is Solutions Engineer time. SE-weeks and Unique's FTE percentages refer to this role.

These figures are the baseline. Any divergence from the standard adds effort, duration, or both — see Section 5.

The delivery timeline starts at kickoff, not at signature. What the customer must prepare comes first and is not counted — see Section 3. Anything still open at kickoff is carried into delivery and adds effort and duration.

Go-Live means the platform is available for productive use. Forward Deployed Engineering (FDE) deliverables and product features named in a proposal are not part of it.

What these numbers exclude

The figures above cover technical delivery only: standing the platform up, integrating it, and operating it.

  • Programme delivery — project and programme management, governance, steering, status reporting, change management

  • Forward Deployed Engineering — use-case build, assistant and agent configuration, prompt and workflow work

  • Product engineering — feature development, roadmap commitments, product changes

These activities are not included in these timelines and efforts.

What is included as standard

Everything below is covered by the effort in Section 1. Anything else is a divergence — Section 5.

The setup

Standard

Architecture

The Unique reference platform for the chosen cloud, with the customer's configuration values and no structural change

Source

The customer forks the reference platform and takes releases from upstream

Cluster

Dedicated to the tenant

Environments

Two — one non-production, one production

Data services

As deployed by the reference platform — managed cloud services where the cloud offers them, in-cluster on OpenShift

Key management

The cloud key service used by the reference platform, with customer-managed keys

Observability

The bundled observability stack deployed by the reference platform, with logs and metrics delivered to one customer sink

The reference platforms are maintained as public code. What they deploy is the standard. Anything they do not deploy is a divergence, priced in Section 5.

Platform

Reference platform

Normative scope

Azure

hello-azure-v2

Full platform

AWS

hello-aws

Full platform

OpenShift

hello-openshift

Application layer only — the customer provides the cluster

The ongoing service

What it covers

Releases

Unique's release cadence applied to the tenant

Requests

Changes to the platform

Incidents

Restoration of service after degradation or outage

Duty

Monitoring, alert response and availability

Responsibilities

Covers setup and ongoing operation. R = responsible, A = accountable, C = consulted, I = informed.

Layer

Responsibility

Cloud environment — accounts, subscriptions or cluster tenancy, transit network, guardrails, provisioning

Customer RA · Unique CI

Identity and access — cloud IAM and RBAC, identity provider tenant, users and groups

Customer RA · Unique CI

Identity and access — identity provider connection to the platform

Unique RA · Customer CI

Platform layer — deployment, upgrades, configuration, connectors (reference platform scope)

Unique RA · Customer CI

Requests — changes to the platform

Unique RA · Customer CI

Custom integrations and extensions

Customer RA · Unique CI

Model provisioning and access

Customer RA · Unique CI

Monitoring, alerting and threat detection

Unique RA · Customer CI

Backup and disaster recovery

Customer RA · Unique CI

CI/CD pipelines

Unique RA · Customer CI

1st level support

Customer RACI

2nd level support

Unique RA · Customer CI

3rd level support

Unique RA · Customer CI — product engineering, outside the effort on this page

User training and use-case documentation

Customer RACI

Unique's access is scoped operational access, federated and audited. Unique monitors using its standardized tooling. Work management runs on Unique's task management and operating model, with one intake channel through Unique's tooling. The customer reviews and approves at stage gates.

What the customer must provide

Every item outstanding at kickoff is carried into delivery. It adds effort, and moves Go-Live by at least as long as it takes to resolve.

Before kickoff

Data protection

Every approval required for the data classifications in scope for the target environment — including approval to process client-identifying data in the chosen cloud and region where that applies

Security approvals

Security review complete and the guardrail list final

Network

Connectivity and DNS approved and in place. On Azure and AWS, the hybrid transit network established

Firewall rules

Approved egress from the target environment to the identity provider, software artifact sources, and any external data sources the platform must reach. On OpenShift, also approved firewall access for end-user devices

Cloud environment

Accounts or subscriptions provisioned, sized per the reference architecture. On OpenShift, the cluster provided and running

Identity

Identity provider ready to federate

Model access

Foundation model inference and embedding model inference, both enabled in the customer's own cloud, each with approved quota — where models are consumed from the customer's own cloud

People

A named owner with decision authority, plus a deputy

Timeline to Go-Live

Five stages, in order. Each gates the next.

Stage

What completes

Elapsed

Unique effort

Discovery and configuration

Customer landing zone mapped onto the reference platform, configuration values agreed and signed off

4 weeks

2 SE-weeks

Landing zone validation

Customer environment verified against the reference

4 weeks

1 SE-week

Platform build

Cluster and platform services running and reconciling

6 weeks

4 SE-weeks

Integration

Identity provider, model access and data sources connected

6 weeks

3 SE-weeks

Verification and Go-Live

Acceptance met, tenant transferred to the ongoing service

6 weeks

2 SE-weeks

Total

26 weeks

12 SE-weeks

What changes effort and duration

A divergence is any requirement that does not match the standard in Section 2. Add its delta to the baseline in Section 1 rather than rebuilding the estimate. Deltas are additive across divergences. Every divergence needs Solutions Engineering review and Solutions Engineering Lead approval before it is quoted.

Divergences are technical or operational. A request for programme delivery, use-case build or a product change is not a divergence and has no delta on this page. It is out of scope, and it is priced by the function that owns it.

The table is not exhaustive. A requirement that is not listed is still a divergence. It has no published delta and is sized at review.

Factor

If the customer wants

Setup

Duration

Ongoing (FTE percentage points)

Source structure

A customer-specific deployment configuration and source structure rather than forking the reference platform and taking upstream releases

+3 SE-weeks

+3 weeks

+15 to +25% FTE

Shared cluster

The platform on a shared cluster rather than a dedicated one

+2 SE-weeks

+2 weeks

+10 to +20% FTE

Release tooling

A different GitOps or CD controller rather than the one the reference platform deploys

+2 SE-weeks

+2 weeks

+5 to +10% FTE

Ingress controller

A customer-specific ingress controller rather than the one the reference platform deploys

+2 SE-weeks

+2 weeks

+5% FTE

Key management

An external key store or HSM rather than the cloud key service the reference platform uses

+2 SE-weeks

+2 weeks

+5% FTE

Access

No standing access — every action raised, approved and executed per ticket

+3 SE-weeks

+4 weeks

+10 to +25% FTE

Observability stack

A customer-specific observability stack rather than the one the reference platform deploys

+2 SE-weeks

+2 weeks

+5 to +10% FTE

Log and metric destinations

Delivery to additional sinks, or SIEM integration beyond one sink

+1 SE-week

+1 week

+5% FTE

Environments

Each environment beyond the standard two — per environment

+2 SE-weeks

+2 weeks

+5% FTE

Data services

Existing customer data services consumed rather than those deployed by the reference platform

+2 SE-weeks

+2 weeks

+10 to +20% FTE

Work management

The customer's task management system rather than Unique's

+1 SE-week

none

+5% FTE

Review model

Review and approval of every change rather than at stage gates

+2 SE-weeks

+2 weeks

+10% FTE

Migration

An existing tenant moved rather than deployed new

+6 SE-weeks

+8 weeks

none

Last updated