Five ways,
one model.
We sell installations and engineer-backed operation. Five engagement shapes, fixed fees, the system we build belongs to you. No per-seat, no per-token, no metered surprises.
We do not charge per seat, per token, or per query. We charge for an installation: a fixed engagement, calibrated against your data, supervised by engineers who stay on the line. The brain we build belongs to you. We charge you to build it well.
Cleverwave is an AI installation practice. The pricing model below reflects that — fixed-fee deployments, transparent operating retainers, and no metered surprises. Compute is passed through at cost in your environment, whether that is your cloud, your VPC, or approved on-prem hardware.
Diagnostic
For mapping the highest-leverage AI insertions before you commit.
Roadmapfixed · 1 weekA workflow audit and AI opportunity map. We sit with your team, identify the first useful agent workflows, and deliver a 90-day deployment roadmap.
What's included
- Workflow audit with your team
- AI opportunity map
- 90-day deployment roadmap
- Data and integration readiness review
- Credited toward an install if signed within 60 days
Pilot install
For proving agent value on real company data.
Pilot swarmfixed · 3 weeksA scaled-down swarm so you can see agents running on your data before committing to a full install. Cloud-hosted by us during the pilot, with real integrations and operator training included.
What's included
- Orchestrator plus two specialist agents
- Three integrations
- Training for the operating team
- 30 days of support
- 50% credited toward full install if upgraded within 90 days
Best for
Teams that need proof on a contained workflow before installing a full swarm.
Book a deploymentFull install
For installing a working agent team inside one operating function.
Full swarmfixed · 8–12 weeksThe complete deployment: environment setup, full agent swarm, your stack connected, and your team trained. Scope is calibrated to your integrations, agent count, and operating complexity.
What's included
- Full agent swarm
- Deployment in your environment
- Core business systems connected
- Team training
- 60 days of engineer-backed support
- Additional specialists scoped per agent
Operate
For keeping the swarm healthy and improving after launch.
RetainermonthlyOngoing engineer-backed operation. Essential keeps the swarm healthy. Core adds new agents, integrations, and a monthly strategy call. Embedded dedicates a fractional engineer one day per week.
What's included
- Essential, Core, and Embedded tiers
- Health monitoring and fixes
- Model and tool updates
- Optional monthly strategy cadence
AI-built software
For custom tools built by the swarm and supervised by engineers.
Custom buildper projectCustom software built by the swarm, supervised by senior engineers. Scope ranges from small internal tools to full production builds.
What's included
- Internal tools, CRMs, web and mobile apps
- Engineer-supervised production delivery
- Maintenance plan available
The matrix, for the people who want
to read the matrix.
| Capability | Diagnostic | Pilot install | Full install | Operate | AI-built software |
|---|---|---|---|---|---|
| Timeline | 1 week | 3 weeks | 8–12 weeks | Ongoing | Scoped |
| Agent scope | Roadmap only | 2 agents | Full swarm | Existing swarm | Project-specific |
| Deployment environment | Assessment | Hosted pilot | Your environment | Your environment | Project-specific |
| Integrations | Mapped | 3 | Core stack | Added by plan | As scoped |
| Team training | — | Included | Included | As needed | As needed |
| Engineer support | Roadmap review | 30 days | 60 days | Monthly retainer | Project delivery |
| Credit toward install | 100% within 60 days | 50% within 90 days | — | — | — |
| Best use | Find the leverage | Prove the workflow | Install production | Maintain and improve | Build software |
What we get asked
before every engagement.
- 01
Why no per-seat or per-query pricing?
A pricing model that grows with usage misaligns the vendor and the customer. We want our agents to do more work, not fewer. Fixed engagement fees keep that incentive clean. Compute is metered, but it is your environment and your bill — passed through at cost, not marked up.
- 02
What does "walk-away clause" mean?
Every engagement defines a calibration target before we begin — usually 'match the median operator' on the in-scope decisions. If we cannot get there in the contracted window, we tell you, return the unspent portion of the fee, and leave. We have walked away from four engagements so far this year.
- 03
Who owns the calibrated brain?
You do. Decision history, supervisor logs, configuration, and the operating context we build live in your environment and are exportable on the day you decide to leave. We retain only the right to use anonymized signal in our own calibration research, and you can opt out of that in writing.
- 04
How does this compare to a chatbot subscription?
It does not. We are an installation, not a tool subscription. The closer comparison is to bringing in an engineering firm to build and operate a piece of internal infrastructure — except we stay attached to it for the life of the system, and the artifact compounds in value over time rather than depreciating.
- 05
Can you do this faster?
For a full install, eight to twelve weeks is the realistic range. First useful agents can be online in week one, but production calibration takes longer. Operators who need a smaller proof point usually start with a three-week pilot.
- 06
What happens after launch?
Most full installs convert into an Operate plan. Essential keeps the system healthy, Core adds new agents and integrations, and Embedded gives you a fractional engineer one day per week. The system is yours either way.
Fees are quoted in USD and exclude applicable taxes. International engagements are quoted in local currency at the prevailing rate on the contract date.
Compute, storage, and third-party model API costs are passed through at cost in your environment. We do not mark up infrastructure.
All engagements are bound by a mutual NDA. Customer logos and case studies are anonymized by default and only published with explicit, current written consent.