A shortlist worth pursuing
Prioritized workflows, the people affected and the practical reasons to start—or leave them alone.
A business decision before a technology decisionFind the right
Find the right
work for AI.
Turn a promising idea into a decision you can act on. We map the workflow, test the assumptions and define what is worth building—with your team and your real constraints.
Strategy · Evaluation · Hands-on delivery
What would we actually test?
Fictional examples · Sample sourcesA real-world question
Who approves a new software subscription?
Purchasing guide · §2 · Fictional excerpt
Send new software requests to your team lead for approval before purchasing.
What we would check
- Answer matches the source
- Missing information is flagged
- Review takes less effort than searching
A real-world question
Draft a reply about delivery coverage.
Example AI draft
Thanks for your enquiry. We currently deliver within Texas. Could you share the delivery address?
Service guide · Delivery area ↓Service guide · Delivery area · Fictional excerpt
Delivery is available within Texas. Confirm the delivery address before quoting.
What we would check
- No unsupported promises
- Tone fits the business
- A person approves before sending
A real-world question
Extract the renewal date and notice period.
Example AI draft
Renewal: October 1. Notice period: not stated in this excerpt—review required.
Sample agreement · Renewal ↓Sample agreement · Renewal · Fictional excerpt
The service renews on October 1. Refer to the termination schedule for notice requirements.
What we would check
- Date matches the source
- Missing notice period stays unresolved
- Reviewer can inspect the original text
What you can take forwardA clear decision.
A clear decision.
A practical plan.
A baseline and a test plan
What happens today, what a useful result looks like and how we will check accuracy, effort and cost.
A build-or-buy recommendation
Options based on your data, existing systems, access requirements and the team that will maintain the result.
A scoped next step
A proposed pilot with deliverables, responsibilities, estimated costs and a clear decision point before a larger build.
Evidence before expansionA pilot should
A pilot should
answer a question.
Can the system do useful work with your data, at an acceptable cost, with the right level of human review? Agree the test before judging the demo.
At the decision point
- Proceed
- The agreed criteria are met. Scope the production work.
- Refine
- Value is plausible, but the workflow or data needs another test.
- Stop
- The evidence does not justify the next investment.
These are possible outcomes, not claims about completed client projects.
Work through it togetherFrom a recurring task
From a recurring task
to a tested direction.
Start small enough to learn something useful. Expand when the evidence supports it.
- Understand the current workTalk to the people doing it. Map inputs, decisions, exceptions and the effort involved.
- Check the foundationsReview data quality, permissions, system access and what errors would mean for the business.
- Test a focused exampleWhen a pilot is in scope, use representative cases—including difficult ones—and compare the results with the baseline.
- Make the next decisionDocument findings, limitations, costs and ownership. Agree whether to build, refine or stop.
Data and responsibilityDecide what AI
Decide what AI
can see and do.
Data access, human approval and operating responsibility belong in the plan from the start.
Explore private AI deployments ↗Information boundaries
Which sources are approved, where they can be processed and who can retrieve them.
Human decisions
Which outputs need review, which actions require approval and when the system should defer.
Ongoing ownership
Who monitors quality, maintains the sources and responds when the workflow changes.
When the direction is clearThe same team
The same team
can help you build.
Before we start
Practical questions.
What is included in a consulting engagement?
We agree the scope around your business question. Discovery typically covers workflow mapping, data readiness, build-or-buy options and a prioritized recommendation. A working prototype or production build is included only when it is part of the agreed scope.
Can you build the solution too?
Yes. The same studio can design, build and support the system. We scope each stage so you can decide whether to continue with us, use your internal team or choose another implementation partner.
Do we need a clear AI use case before we start?
No. Bring a recurring task, a bottleneck or an idea you want to assess. We help define the problem and consider whether generative AI, a simpler automation or a process change is the better fit.
How do you assess whether it is worth doing?
We establish a baseline and agree success criteria before building. Depending on the workflow, that can include review time, answer quality, correction rates, operating cost and adoption. Pilot results inform the recommendation; savings are not assumed in advance.
What happens to our data?
We review what data is needed, who can access it, where it can be processed and how it should be retained. Hosted, private-cloud or on-premises options are evaluated against those requirements. Data flows and vendor terms are agreed before connecting sensitive systems.
How long does it take, and what does it cost?
The schedule and price depend on the workflows, data access and depth of testing. We agree deliverables, milestones and fees before work begins. Discovery, a prototype and production implementation are separate decisions, not an automatic commitment to a full build.
Bring one workflow.
Tell us what your team repeats and what you would like to improve.