AI
A structured evaluation of whether specific tasks in your business are suited to AI, and whether the data behind them supports it. It frequently concludes that some are not, which is the point.
Stated as we hear it, before any mention of what we would do about it.
Qualitative, because we do not publish numbers we cannot evidence.
Approach
Three to four weeks.
The tasks actually being done, with frequency and time observed rather than estimated. Recall of one's own repetitive work is consistently unreliable.
Week 1Each task against suitability and data availability. Most organisations find the constraint is documentation quality, not model capability.
Weeks 2–3A ranked shortlist with the reasoning, including what should not be attempted and why. A report with no rejections has not assessed anything.
Week 4Scope
Stated plainly, so there is no ambiguity about what you are buying.
How it works
How a grounded AI system answers a question
A request is matched against your own documented content, an answer is composed from what was retrieved, and anything below the confidence boundary is escalated to a person with full context.
The boundary
Naming the boundary early removes the most common source of disappointment in an engagement.
Then it has done its job and saved a build. It will also say what would make you ready (usually documentation, data definitions or a process simplification) which is generally cheaper than the project it replaced.
Not usually. Most useful first projects work against a single system or a defined document set. Data platform work becomes necessary when a project genuinely needs data from several sources at once.
A structured evaluation of whether specific tasks in your business are suited to AI, and whether the data behind them supports it. It frequently concludes that some are not, which is the point.
Organisations being asked for an AI strategy without a defined problem; Teams whose pilots keep stalling before production; and Businesses wanting a defensible reason to say no to some proposals
Task inventory: volume, variability, tolerance for error and cost of a mistake; Data assessment: whether the material a system would need actually exists and is accurate; Feasibility judgement per task, including tasks that should stay manual; Failure-mode analysis: what happens when the system is confidently wrong; Governance starting points: permitted data, decisions requiring a person, ownership; and A first project recommendation, sized to produce evidence within a quarter
There is pressure to adopt AI without a specific problem attached; Pilots have been run and none has reached production; and Nobody can say whether the data would support what is being proposed
A shortlist of tasks where AI is genuinely suited, with reasoning; An honest statement of what your data does and does not currently support; and A first project scoped small enough to prove or disprove the assumptions
A build: this exists to decide whether one is warranted; and Model or vendor selection, which follows a defined problem rather than preceding it
Connected
Describe where the work currently is and what it has to achieve commercially. We will tell you what we would scope, what we would leave alone, and whether this is the right service to be buying first.
