01 — STRATEGY
The question is almost never whether AI could do something here. It usually could. The question is where it would pay for itself, what it would cost to keep running, and which parts of your business are better left alone.
AI strategy is deciding where AI belongs in a business before anybody builds anything. It is the Diagnose week every Perspicality engagement opens with, sold on its own: we watch how the job is actually done rather than how it is described, then name where AI would pay for itself and where it would cost more than it saves. What comes back is a plan someone can act on: what to build first, what running it takes, and where the honest recommendation is to build nothing.
02 — BUILD
An AI strategy engagement starts inside the work rather than in a workshop, and ends in something somebody can act on. Each piece below is here because AI decisions go wrong at the same points.
Diagnose opens every engagement; sold on its own, it ends in a plan rather than a build. We watch the work happen before anything is recommended — the clicks, the handoffs, the workaround somebody invented years ago and never mentioned. The process a company describes and the one it runs are rarely the same, and only one is worth planning against.
Every candidate goes through the same test. Does the task repeat. Could a competent person write down the logic they use. Does it happen often enough that a small gain adds up. And can a wrong answer be caught and undone before it costs you. Work failing that last one stays with a person.
What comes back is short and specific: what to build first and why it goes first, what running it takes, who owns it, and how you will know it worked. A plan that stops at "use AI in operations" is one nobody can start on Monday.
Part of what you pay for is the recommendation not to build. Some work is better simplified than automated, some should be deleted outright, and some belongs with a person. The federal AI risk framework builds the same stop into its own process: its mapping stage ends in a go/no-go decision on whether to develop or deploy at all. A no is a result rather than a failure to find one, and ours comes with the reasoning.
03 — PROOF
ChaChing shows what naming the constraint correctly looks like, which is the judgement an AI strategy engagement sells. Reps ready to close with no machine feeding them is not a closing problem; it is a function nobody has stood up.
ChaChing had reps ready to close and no machine feeding them. Read as a closing problem, it invites a fix aimed at people who were never the constraint. Read correctly, it names a missing function: building and staffing outbound — sourcing, enriching, sequencing, following up — is one most teams never get around to standing up. What ChaChing hired us for was the building, and we operate that function end to end today.
04 — PROCESS
Four steps, the same four on every system we run: we watch the work, agree what gets built, ship it, then operate it.
We spend a week inside the work, watching how it is really done, and name where the time and money leak.
We write down what the system will do — owners, hand-offs, what happens when it is wrong — and you approve it before anything ships.
We build on your stack and ship a working tool in weeks, not quarters. Your team uses it the day it lands.
One-time setup, then flat monthly to monitor and tune it. Each system frees up time and data the next one builds on. If it does not earn its keep, fire us.
05 — FIT
AI strategy is the right call when the pressure to use AI arrives before any clear picture of where it would help. If you can name the outcome but not the work that gets you there, that gap is what to close first.
06 — LIMITS
A strategy engagement is worth only what somebody can act on afterwards. Each limit below names a way this kind of work turns into an expensive document nobody opens.
This is not a deck. What lands is a short written plan with an owner and a running cost against every call in it. A strategy nobody can act on is an expensive way to feel prepared.
Sometimes the honest answer is don’t. If the work does not suit AI, or being confidently wrong would cost more than being right would save, that goes in the plan — we would rather lose the build than sell you one.
Watching how the work is done is not a performance review. We read the process, not the people, and what gets written down describes the work rather than who is slow at it.
We do not assess a process nobody intends to change. If the decision is already made and the assessment exists to ratify it, that is theatre, and we would rather not be in it.
07 — FAQ
Paying someone to tell you what to do about AI invites scepticism, and these answers are written for a sceptic.
A short written plan: the work we watched, where AI pays here, what to build first, and what to leave alone. The reasoning sits next to each call, so your team can disagree with a specific line rather than a conclusion.
Mostly in where it starts and what it leaves you holding. A readiness assessment scores a company against a maturity model; this starts inside your actual work and ends with a first build, an owner and a running cost. The score is not the useful part. The sequence is.
No. The plan is yours, written so another team could execute it. If we do build it, this week was that engagement’s Diagnose step, not an extra one. We build and run most of what we recommend, which is also why the recommendation is worth something: we would be the ones who have to make it work.
Then that is what we tell you, with the reason and with what would have to change first. The blocker is rarely the technology — usually it is a process two people describe differently, or data no system can read without somebody cleaning it up.
Weeks, not quarters. The observation runs a week; writing it up and pressure-testing it with your team takes a little longer, because a plan your own people have not argued with is not finished.
08 — RELATED
Most engagements touch more than one of the systems we build, because each one feeds the others. The links below are the closest matches to this service.
Tell us what isn't working and what done looks like. We come back with what we would build first and what it takes to run it.