AI consulting & implementation

Find where AI belongs in your business—and put it to work.

Find useful applications for AI. Then test them in the work your team actually does.

From opportunity assessment to a working pilot, we connect the business question, the available information and a practical implementation.

Discuss an AI project ↗

Company papers gathered into a sculptural brain. — Company knowledge, connected to a useful next step.

Ways to work together

Choose a project you can put to work.

Start with the result you need. These are starting points for a scoped engagement, not fixed packages. The quote defines the deliverables, boundaries and success test for your project.

AI opportunity assessment

For a team deciding where AI is useful—and whether to invest at all.

Deliverables, scope & handover

Deliverables: A prioritised opportunity map, comparison with existing products and a recommended first test.
Scope & cost drivers: The number of workflows and stakeholders, sample quality and the depth of tool evaluation.
Success test: The decision-maker can compare options against agreed business needs, with unknowns and a go / no-go recommendation visible.
Handover: A decision brief, assumptions and a proposed test scope; implementation is a separate decision.

Company knowledge assistant

For teams that need useful answers from procedures, project history or support records.

Deliverables, scope & handover

Deliverables: An assistant over an agreed source collection, source-linked answers and a representative evaluation set.
Scope & cost drivers: Source preparation, access rules, document variety, hosting and where the assistant needs to appear.
Success test: Agreed questions return supported answers; restricted, conflicting or missing information is handled correctly.
Handover: The agreed configuration or code, source-update instructions, evaluation cases and operating notes.

Connected AI workflow

For teams that need AI to interpret information and prepare the next step across tools.

Deliverables, scope & handover

Deliverables: One scoped workflow linking interpretation, evidence and a useful action, with review points and visible failures.
Scope & cost drivers: System access, document complexity, permitted actions and the range of exceptions to test.
Success test: Representative cases reach the right next step; uncertain cases stop for review and unapproved commitments are not released.
Handover: Workflow configuration or code, test cases, connection notes and a guide to pausing and recovering the workflow.

How scope becomes a quote ↗

Illustrative example / Company knowledge

From “Have we solved this before?” to a useful next step.

A project lead is planning a new customer installation. Relevant experience lives in completed project notes, procedures and support records. An internal assistant could bring those sources into the planning conversation without giving everyone access to every customer file.

01 / Ask

Find the relevant experience.

Search approved project history for comparable equipment and site constraints. Link each recommendation to its source and revision so the lead can inspect the original.

02 / Prepare

Turn knowledge into a working brief.

Draft a preparation checklist from the current procedure and past lessons. Separate established requirements from suggestions; flag a missing site measurement instead of supplying a plausible value.

03 / Decide

Keep the project lead in charge.

The lead checks applicability, assigns the measurement task and approves the plan. A sourced answer is not permission to change the procedure or commit to an installation date.

Illustrative workflow, not a delivered client system. Answers respect the reader’s access; a missing or unsuitable source stays visible. No live company data is connected.

Explore the company-knowledge example ↗

Implementation, testing & ongoing operation

From pilot to everyday use

Agree what “works” means.

Acceptance criteria

Test representative questions, source accuracy, permission boundaries and cases where the assistant should say it cannot answer. Include stale documents, conflicting versions and requests outside scope—not only a successful demonstration.

Deployment choices

Decide where the interface lives, how users sign in and which systems it may read or update. Hosting, model providers, retention, logging and costs are choices to review, not a blanket security promise.

Support and ownership

Agree who maintains the source collection, reviews failures and approves new actions. Handover can include configuration, test cases and operating notes; monitoring and ongoing support are scoped explicitly.

Could an existing AI product be enough?

When not to build

An existing product may be enough.

If your team already works in a supported document suite, its search or assistant features may solve the problem with configuration and better source organisation. Compare permission behaviour, source visibility, export options and ongoing licence costs before commissioning a custom assistant.

A tailored implementation becomes useful when the work crosses systems, needs specific evidence checks or must prepare a next step that packaged features cannot reliably support. Start with the smallest test that distinguishes those options.

Compare configuration, integration and custom software ↗ Explore workflow automation and system integration ↗

Start with one project.

Tell us what you want to improve. We’ll compare approaches and agree on the scope, price and what success looks like before building. You don’t need a technical brief.

Pricing & working together ↗

Discuss an AI project ↗