Most AI engagements open with an open-ended discovery phase: a vague retainer, a discovery deck, and a meter running while everyone agrees that AI is important. We do not work that way. The front door to Segment360 is a fixed five-day Diagnostic. It costs $3,500, and that fee is refunded against the build if you decide to move forward. You know the price, the timebox, and the deliverables before we start.
What it actually is
The Diagnostic is a focused, paid week of work, not a sales call wearing a project hat. The goal is narrow and honest: decide whether there is a single, high-value workflow in your business worth automating with an agent, and produce the plan to build and operate it. If the answer is no, we tell you, and you still walk away with a real map of your stack and your data. If the answer is yes, the $3,500 comes off the build.
We charge for it on purpose. A free assessment is structured to flatter you into a contract. A paid Diagnostic is structured to give you a usable artifact even if you never sign anything else. The fee keeps both sides serious. You commit a week of access to your people and systems. We commit a solution engineer who has actually shipped and operated agents, not a generalist with a template.
The shape of the week
The week is sequenced so each day feeds the next. We are not running five disconnected meetings. We are narrowing from your whole operation down to one agent worth building.
- Strategic kickoff. You sit with a solution engineer, not an account manager. We get the real picture of how the business runs: where the time goes, where the errors happen, what is on fire, and what people quietly route around because the software does not handle it.
- Map your data and your stack. We inventory the systems that hold the data an agent would need to read or act on: your practice management system, scheduling, billing, the inbox, the phone system, whatever it is. Where the data lives, who owns it, how it moves, and what it would take to give an agent safe, scoped access inside your own cloud.
- Size the highest-value workflow. We rank candidate workflows by volume, by cost of getting it wrong, and by how cleanly an agent could own it. Frequency times pain, minus the risk of automating it badly. Most teams have a long list of things they wish were automated. The point of this step is to find the one that pays for itself first.
- Compliance and security gap analysis. Before anyone talks about building, we look at what an agent touching this data would mean for your obligations: where sensitive records live, what your boundary looks like today, and what controls need to exist before an agent runs in production. You come out of the week knowing exactly where the boundary would sit and what it asks of you, before anyone touches a record.
- Identify the single first agent. We converge on one agent: what it does, what it is allowed to do, what stays human, and how we would know it is working. One agent, scoped tightly enough that it can actually ship and earn trust.
Why one agent, not ten
The instinct after a good kickoff is to build a roadmap of a dozen agents that will transform everything. We resist that, hard. A first agent has to do something narrow, prove it is reliable under real volume, and earn the right to expand. That is the Cachalot Compass in practice: Signal Detection to find where the real value is, Deep Exploration to understand the workflow underneath the surface description, and Strategic Resurfacing to come back with one clear recommendation instead of a list. See beyond the surface, think beyond the obvious, then pick one thing and do it well.
What you walk away with
At the end of the week you get two artifacts, and neither of them is a pitch deck.
- An action-oriented roadmap. The specific first agent we recommend, why it beat the alternatives, what it will and will not do, and a realistic build sequence. It is written so your team can act on it, with us or without us.
- A capability deployment framework. How the agent would actually run inside your cloud: data access boundaries, the human checkpoints, what we monitor once it is live, and how we measure whether it is doing its job. This is the operating plan, not just the build plan, because building the agent is the easy part and running it is where the value is.
You also get the data and stack map and the compliance gap analysis as standalone documents. Even a team that walks away has a clearer picture of their own systems than they started with.
Why fixed-fee and short beats a vague retainer
A long discovery retainer rewards the vendor for taking longer. The incentives are backwards: the more uncertain everything stays, the more hours get billed. A fixed five-day Diagnostic flips that. We are paid to reach a clear answer fast, and the timebox forces us to.
Short also protects you. Five days and $3,500 is a decision a single stakeholder can make without a procurement cycle. If we are not the right fit, you found out for the price of one week, not one quarter. And because the fee is refunded against the build, choosing to continue costs you nothing extra. The Diagnostic is not a tollbooth in front of the real work. It is the first slice of it.
Where this leaves you
We design, build, and operate agents inside your own cloud, on GCP, AWS, or Azure, so your data never leaves your boundary. That is a serious commitment to make on both sides, which is exactly why we do not start with a contract. We start with a week that produces something real whether or not you continue. The Diagnostic is the honest front door: fixed price, fixed time, a usable plan at the end, and no pretending the answer is yes before we have looked. If you want to know whether there is a first agent worth running in your business, that week is how you find out.