Skip to main content
NEW GUIDEFour ways to make a regulated firm AI-nativeGet the deck
Softblues

How we work

How does an engagement work?

We start with the work you want to change. An initial conversation establishes the context; paid discovery examines it in detail. You then decide whether to proceed with a scoped rollout or build. Ownership, acceptance criteria and support arrangements are agreed around the engagement.

  • OpenAI Select Partner
  • Google Cloud Services partner
  • Anthropic Partner Network member
  • Member of the Microsoft AI Cloud Partner Program

What happens from the first conversation onwards?

Each step should give you enough information to make the next decision. Detailed investigation and implementation are separate commitments.

  1. First conversation

    We discuss the objective, the people involved and the current systems. The purpose is to understand the situation and whether further discovery would be useful.

    You leave with. An agreed next step and the information needed to scope it.

  2. Paid discovery

    We examine representative work, sources, permissions, exceptions and review requirements. We establish a baseline and assess what needs to change, including any preparation outside the AI tool itself.

    You leave with. The findings, priorities and a defined implementation scope, with assumptions and dependencies recorded.

  3. Proposal and decision

    We turn the agreed scope into a proposal covering deliverables, responsibilities, milestones, price and acceptance criteria. The business case separates capacity released from cash savings and states the assumptions behind expected benefits.

    You decide. Proceed, revise the scope, do the preparation first or stop.

  4. Rollout or build

    We configure and connect the agreed solution, test it against real tasks and work with the people who will use it. Training, documentation and exception handling are part of delivery.

    You accept. The agreed deliverables against the criteria set before implementation.

  5. Support and improvement

    Where ongoing support is agreed, we review usage, quality, cost and exceptions with the named owner. Support responsibilities and response terms are set in the engagement.

    You decide. Expand, adjust or stop, using evidence from the work.

How does the work differ between adoption and automation?

The starting process is shared. What we implement depends on how the work needs to run.

AI adoption compared with process automation
AI adoptionProcess automation
Starting pointPeople need a governed assistant and repeatable ways to use it.A defined workflow needs to run across tasks or systems.
Typical workAccess, prepared knowledge, shared methods, rollout and training.Triggers, integrations, agent logic, exception handling and review.
Evidence to reviewUse on real tasks, output quality, review effort and cost.Completed work, exceptions, quality, review effort and operating cost.
OwnershipA business owner and people who maintain methods and sources.A process owner and people responsible for operation and changes.

What principles guide delivery?

The same discipline on every engagement, whatever the platform and whoever runs the process afterwards.

  • Start from a baseline

    Measure the current work before changing it. Include the time spent reviewing output and dealing with exceptions.

  • Keep access deliberate

    Agree which sources the system may use, which actions it may take and how permissions will be maintained.

  • Make review part of the design

    Define what a person must check, when they can intervene and what happens if the system cannot complete the task.

  • Keep decisions visible

    Record scope changes, assumptions and acceptance decisions. Agree a communication rhythm so the client team knows what is ready and what needs its input.

  • Leave an operable system

    Delivery includes the documentation, training and ownership needed to run the agreed solution after handover.

What do we need from you?

Start with a business objective, a named sponsor and someone who knows the work in detail. During discovery, we agree access to representative examples and involve the people responsible for IT, information and approvals.

You do not need to arrive with a technical specification. Detailed mapping is part of the scoped discovery work.

Where can you see the approach in practice?

Our case studies explain the problem, the work undertaken and the evidence available at each project's current stage. Our own Claude Operating System shows how we apply shared methods inside Softblues.

Key facts

How an engagement works: at a glance

Steps
First conversation, paid discovery, proposal and decision, rollout or build, then support and improvement.
First conversation
A discussion of the objective, the people involved and the current systems. It is not discovery and it produces no specification.
Discovery
Paid, scoped before it begins. It returns findings, priorities and a defined implementation scope with assumptions and dependencies recorded.
Commitments
Detailed investigation and implementation are separate commitments. You decide after the proposal whether to proceed, revise the scope, prepare first or stop.
Timetable
Depends on scope, access, data preparation, integrations and the availability of the people involved. Milestones are agreed in the proposal.
Price
The agreed scope is priced and the commercial terms are stated in the proposal. Licence costs, usage charges and ongoing support are identified separately where relevant.
Acceptance
A baseline and acceptance criteria are agreed before implementation. Released hours are recorded as capacity unless a route to financial benefit has been established.

Common questions about working together

Is the initial call the same as discovery?

No. The initial call establishes context and the next step. Detailed workflow mapping, analysis and specification are paid work, scoped before they begin.

How long does an engagement take?

The timetable depends on scope, access, data preparation, integrations and the availability of the people involved. We agree milestones in the proposal after examining those dependencies.

How is the work priced?

We price the agreed scope and state the commercial terms in the proposal. Licence costs, usage charges and ongoing support are identified separately where relevant.

Can you work with our existing tools?

Yes. We assess the platforms and systems you already use before recommending changes. This can include Microsoft 365, Claude, ChatGPT, Google Cloud and your existing business applications.

How do you decide whether the work has succeeded?

We agree a baseline and acceptance criteria before implementation. Review includes quality, use, exceptions and cost, as well as the business objective. Released hours are recorded as capacity unless a route to financial benefit has been established.

What happens if the proposed approach is unsuitable?

Discovery can lead to a smaller scope, preparation work, a process change or a decision to stop. Any acceptance remedies for implementation are set out in the contract; there is no general refund promise.

Ready to examine the first task?

Tell us what the team does today, where it gets stuck and what you would like to improve. The first conversation helps us agree whether further discovery is worthwhile.