Nobody fails at the demo.
They fail at the first mile.
An ARPIA engineer works inside your operation, learns the process the way your people actually run it, and comes back with a recommendation you can act on. It starts with a AI Route: a fixed number of weeks, a fixed price, and at the end you know what to build, what it touches, what it returns and what it costs. Then the same engineer builds it with your people, if that is what you decide.
Self serve, or shoulder to shoulder.
Build it yourself
Your teams work on the platform directly with Oria: explore the data, reason, author the ontology and ship governed apps and pipelines. Full control, at your pace. Seats and nothing else.
Forward deployed
An engineer embeds with your operation and does it with you, use case after use case, until your people do it alone. It is how most enterprises work with us, and it starts with a Route.
The blocker is never the model.
Enterprises do not stall because the technology is not ready. They stall because the first project has to answer questions nobody in the room can answer yet, and answering them costs a quarter of meetings before anything is built.
- The process lives in five systems and nobody can say which one is authoritative.
- The knowledge that decides the hard cases is in three veterans, not in a document.
- Nobody owns the ontology, so every team models the same entity differently.
- Procurement wants a business case that needs a result you only get after building.
A Forward Deployed Engineer answers those questions by working, not by workshopping. Two weeks inside one process produce the map of the systems, the sketch of the ontology, the use cases the process makes possible and a staged plan with a price against each stage.
Then the same engineer builds it with your people, and keeps going until your team builds without them. That last part is the point, not a service tier.
AI Route. Weeks, not a business case.
A Route is a diagnosis you can buy in one decision. You choose how many processes to put on the table, an engineer works through them with your team, and you get back what to build and what it costs. It is the smallest honest way into a platform.
For the team that already knows which process hurts.
- One process mapped end to end, as it really runs
- The systems it touches and what integrating each one takes
- The use cases it makes possible, sized by return and by effort
- A staged plan with a price against each stage
For a whole function, where the processes share the same entities.
- Everything in AI Route, across three processes
- One ontology sketch that covers what they share, instead of three
- The order to build them in, and which phase of the roadmap each one belongs to
- Governance read: what the AI would touch, and under whose approval
For the operation, when the decision is a program and not a project.
- Everything in AI Route Pro, across six processes and more than one area
- The adoption plan for the company: who starts, with which seats, in what order
- The data and governance gaps that have to close first, named and sized
- The five phases costed end to end, ready to take to a committee
If you go ahead within 60 days, the Route is credited in full against the first phase, so the diagnosis costs you nothing when it turns into work. These are preliminary service prices, from, and they adjust with the complexity of each phase and each client. Figures are in US dollars and apply to the United States. Europe and Latin America are priced for their region: tell us where you operate and the proposal comes back with the right one.
A Route ends in a recommendation, not in a build.
Nothing goes to production during a Route. What you buy is the engineer's judgement, written down: what to do first, what it takes, what it returns and what it costs. It also places you on the roadmap below, says which gate you are standing at and what closing the next one takes.
- What the process really is. Written down as it runs, with the systems it crosses, the data each one holds and who decides what.
- What it would take to connect it. An ontology sketch over those systems, and the work each integration needs, named system by system.
- What is worth building. The use cases the process makes possible, each one sized by the return it carries and the effort it costs.
- What it costs to get there. The phases you have to cross, what each one closes and the price against each, so the next decision is a purchase order and not another study.
The decision is small on purpose: take the recommendation forward with us, or keep it and take it anywhere. Either way it is yours, and it came from an engineer who spent those weeks inside your operation instead of reading a questionnaire.
Land, Expand, Automate, Scale, Enterprise.
Start with AI. Build institutional intelligence. Connect the enterprise. Automate work. Scale into an AI-native organization.
Every phase answers one question, and it does not open until the phase before it closes. The engineer runs each one with the method underneath, the same four dimensions we have used since the first deployment: what the AI knows, what it touches, what it builds and what it does. The first governed use case is normally in production between 30 and 90 days, and production means production, not a proof of concept.
Land Can your people use AI?
Your tenant, AI governance, security, audit, the AI Wallet, your administrators and the first Oria users working. Closes when the platform is governed and those people are active.
Method: Stage 0, Setup. Paid as seats, the AI Wallet and the Setup item.Expand What does your AI know?
Oria across the organization and Cerebro loaded with the institutional memory: domains, rules, procedures, product and customer knowledge, with permissions by area. The engineer trains your key users, who carry it to the rest in the language of the business. Closes when that knowledge is useful, the users are active and the cascade is agreed.
Method: Stage 1, what the AI knows. The hunt for high return use cases starts here, not at the end.Automate What can your AI access and do?
Your essential systems integrated, the ontology built over them, nodes, relations and governed actions where they belong, data quality validated and access governed by role. Closes when the data is governed, the ontology is operational and at least one real automation is running.
Method: Stage 2, what the AI touches. This is the phase that varies most, because it scales with each system integrated.Scale How much can your organization build and automate?
Pro users and Builders trained on Oria Builder, building personal and institutional applications, workflows, agents and workers over the ontology that now exists. Closes when your people build on their own and can scale it internally, with at least one application built by your team.
Method: Stage 3, what the AI builds. The point of this phase is capability transfer, not dependency.Enterprise How do you run AI as critical infrastructure?
Sovereign or dedicated execution where you need it, advanced governance, enterprise support, and a backlog of use cases prioritised by return that keeps producing. This phase does not close, and it is not about a bigger contract: it is AI running as permanent operational capability.
Method: the maturity of Stage 4, what the AI does, across everything installed.A price against every phase, not a rate per hour.
One line per phase of the roadmap. These are preliminary service prices, from, and they adjust with the complexity of each phase and each client. Automate is the one that moves, because it is priced by the systems you actually need connected.
| Item | What it delivers | Price from |
|---|---|---|
| LandArpia Setup Standard, Stage 0 | Tenant, AI governance, security, audit, AI Wallet, administrators enabled. | $3,500 |
| ExpandStage 1, what the AI knows | Oria and institutional memory across the organization, key users trained. | $3,500 per key user |
| AutomateStage 2, what the AI touches | Each essential system integrated, the ontology built and the data validated. | $2,500 per system |
| ScaleStage 3, what the AI builds | Pro users and Builders trained on live cases until they ship on their own. | $3,500 |
| EnterpriseStage 4, what the AI does | High return use cases, found, qualified and prioritised as the work goes. | Quoted per use case |
Seats are separate and they are not a project: they start at $10 per user per month, billed per day. See the seats and the AI Wallet →
The measure is what you build without us.
Capability, not dependency
Every stage ends with your people able to do something they could not do before. Stage 3 does not close until one of them has shipped an application on their own.
The platform stays yours
The tenant, the ontology, the memory and everything built on them live in your account. The engineer leaves, the core does not.
Governed while it happens
Every model call is budgeted before it runs and logged with the context and the policy it ran under, from the first session of the first stage.
ISO 42001 and SOC 2 Type 2. AI maturity Level 6 of 7, in production. Seven years of platform behind the engineer who shows up.
Request a AI Route.
Tell us the process and where you operate. We come back within 24 to 48 hours with the scope, the weeks and the price for your region.
Not sure which one you need?
Tell us the process and we will tell you honestly whether it takes two weeks or eight, and whether a Route is worth buying at all. Thirty minutes is usually enough.