People have AI subscriptions, but the team has no shared workflow or way to judge whether they help.
One useful workflow. A measured pilot.
Make AI useful in the work your team already does.
Find a practical place for AI in engineering, an internal workflow, or your SaaS product. Start with a defined problem, test one bounded pilot, and give your team a clear basis for deciding what to adopt next.
Talk to JonathanScoped individually
Discovery and pilot scope are agreed around one workflow, the systems involved, and the evidence needed to judge the result. Tool subscriptions, ongoing operation, and broader rollout are priced separately.
A PILOT WITH A DECISION AT THE ENDOne workflow.
One workflow.
Evidence to act on.
- ChooseA task, an owner, a baseline.
- TestQuality, review effort, exceptions, cost.
- DecideExpand, revise, or stop.
Define success before the demonstration.
IS THIS THE RIGHT FIT?
A useful task, with someone accountable.
- SaaS founders deciding where AI could help their product or team.
- Engineering leaders introducing AI-assisted development without losing review discipline.
- Teams with repetitive document, data, or system-to-system work they want to assess for automation.
SITUATIONS I CAN HELP WITH01 02 03
A prototype looks promising, but data access, exceptions, review, and integration with existing systems are unresolved.
You need to decide whether an AI feature is worth building, how to evaluate it, and who will operate it.
WHAT YOU LEAVE WITH
A pilot you can evaluate.
A prioritized opportunity shortlist and, when included in the agreed scope, a working pilot, an evaluation scorecard, operating notes, and a recommendation to expand, revise, or stop.
An opportunity and constraints review
Map one workflow, its current effort, the people involved, and the data it uses. Compare AI-assisted, conventional automation, and process changes against the same objective.
A bounded pilot
Where agreed, implement one narrow workflow with representative inputs, explicit access permissions, and a clear human review or escalation path. Keep the initial commitment small enough to evaluate.
Engineering practices your team can use
For AI-assisted development, define useful repository context, task boundaries, code review, tests, and approval points. Record where a person remains responsible for judgment and release.
An evaluation and handoff
Compare the pilot with a baseline for quality, time, review effort, exceptions, and operating cost. Document limitations, ownership, and a recommendation on the next investment.
THE ENGAGEMENTA clear process.
01 02 03
A clear process.
Shared expectations.
Choose the workflow
Identify the actual bottleneck and an owner. Agree on the inputs, permitted data use, boundaries, and evidence that would make the pilot useful.
Build and evaluate
Test representative cases, including failures and ambiguous inputs. Track the effort to check and correct outputs alongside the time the tool appears to save.
Decide what happens next
Review the results with the team. Expand, revise, or stop the pilot based on the evidence, and assign responsibility for anything that moves into regular use.
SCOPE & BOUNDARIESKnow what
Know what
you’re engaging.
This is a scoped adoption or implementation engagement. It does not include unlimited automation, guaranteed savings, model training, or ongoing support unless agreed. Production data use, provider access, approvals, and operating responsibilities are defined before a pilot begins.
What kinds of work could we assess?
Examples include development and review workflows, document extraction, internal knowledge retrieval, structured data handoffs, and a proposed AI feature. These are possible project types; discovery determines whether a particular use case is suitable.
Do we need an AI tool already?
No. Start with the workflow and constraints. We can evaluate tools or work with an existing approved provider. Licensing and any required accounts are agreed separately.
How will we know whether the pilot worked?
Agree a baseline and acceptance criteria before building. Review accuracy or usefulness, time spent checking and correcting, exceptions, cost, and team adoption. A result that does not justify rollout is still a useful decision.
Can this be part of Technical Advisory?
Yes. Opportunity selection and adoption decisions can be part of an agreed advisory engagement. Building a pilot or rolling it out is separately scoped so the time, deliverables, and ownership are clear.
WORKING TOGETHER
A small pilot with a decision at the end.
Keep the first engagement tied to one workflow and a named owner. The goal is enough evidence to decide whether the change deserves a wider commitment.
Engineering workflow
Assess a repeatable development, testing, review, or documentation task. Agree repository context, review responsibilities, and the checks required before work can ship.
Operational workflow
Assess a document, knowledge, or data-handoff task. Define permitted inputs, access boundaries, review queues, and what happens when the system is unsure or unavailable.
Product decision
Evaluate a proposed AI feature against a real user task. Identify representative test cases, unacceptable errors, latency and cost constraints, and the person who owns the outcome.
A useful stop or go decision
Review pilot results alongside the baseline and the effort needed to operate it. Record what is ready, what needs another iteration, and what would need to be true before expanding.
SEE THE WORKUseful before
Useful before
we speak.
Open a sample or planning tool. No signup required.
A DIRECT CONVERSATIONHave a technology decision
Have a technology decision
you need to get right?
Tell me what’s happening. I’ll let you know whether I can help
and how I’d approach the next step.