Skip to content

Ways to work with Cennix.

Plan, build and run. Every engagement is scoped to your project, from a first conversation to software running in production. Monthly support after launch is optional.

Discuss your project

Project-based work, and optional monthly support.

Builds are scoped projects delivered in milestones. Managed hosting and ongoing engineering is a separate, optional monthly engagement.

Project conversation

A short call about the problem, or a brief written recommendation on how to approach it. A lightweight introduction, not a full plan.

No charge · introduction

Solution Design & Consulting

Requirements, architecture, a build-versus-buy assessment, a practical roadmap and a scoped estimate. Useful on its own or before a build.

Scoped separately · before a build

Custom Business Software

CRMs, dashboards, internal tools, customer portals and integrations, built around how your teams work.

Scoped project, in milestones · build

Product Development

Product definition, a prototype where useful, a first usable release, the production application and later iterations.

Scoped project, in milestones · build

Custom AI Implementation

Knowledge systems, agents, AI features, integrations and the infrastructure to deploy and operate them.

Scoped project, with a prototype when uncertainty warrants it · build

Managed Hosting & Ongoing Engineering

Hosting where suitable, monitoring, maintenance, support and agreed improvements after launch. Optional: a handoff is always possible.

Optional, agreed monthly scope · after launch

Software builds can follow a project conversation or solution design. AI builds may include a prototype or an audit first, where it helps. No engagement is described as unlimited: each one defines what it covers.

Two focused AI engagements.

Neither is a required first step. A Prove Sprint tests one uncertain AI capability before you build it. An AI Reliability Audit examines an AI system you already run.

Prove Sprint

For an AI idea that hinges on one uncertain capability. One capability built on your systems and tested against 30 to 100 of your own examples, ending with a go or no-go decision.

3 to 4 weeks, scope agreed per project

  1. Define the test · week 1We agree the task, the data and 30 to 100 of your real examples. Passing that test is what done means.
  2. Build · weeks 2 to 3We build the capability on your stack and run it against the test set as it develops. You see results every week.
  3. Decide · final weekYou get the results, including the failures, a running-cost forecast and a scoped plan for production. Or a clear recommendation to stop.

What you keep

  • The working prototype code
  • The test set and the results against it
  • An architecture note
  • A forecast of running costs
  • A scoped proposal for the production build
About AI prototyping

AI Reliability Audit

For an AI feature, assistant or agent already in use. We measure how often it is wrong, why, and what it costs, using your real traffic, and rank what to fix first.

1 to 2 weeks, scope agreed per project

  • AccuracyHow often answers or actions are wrong on your real traffic, and the recurring patterns behind the errors.
  • GroundingWhether answers are supported by the right sources, and where the system invents or misquotes.
  • Tool use and safetyUnreliable tool calls, prompt-injection exposure and actions taken without the right approval.
  • Cost and speedCost and latency per request, and where they can come down without hurting quality.

What you keep

  • A scorecard with today's baseline
  • A ranked fix list with effort estimates
  • A starter test set you can run on every change
  • A scoped proposal for the fixes, if you want us to make them
About AI evaluation

We work from redacted or sampled data under an NDA, with read-only access where any access is needed.

When we are the wrong choice.

Saying no early saves both sides time. If one of these is true, we will tell you, and point you somewhere better if we can.

  • You want a general technology strategy without a specific problem or outcome in mind.
  • There is no one on your side who owns the result and can make decisions.
  • The data or systems the work depends on cannot be made available to the project.
  • You need a regulatory certification. We engineer and test systems; we do not certify them.
  • You need a marketing website on its own. Websites are in scope alongside web apps and product work.

Not sure which engagement fits?

Tell us about the project. We reply to set up a short conversation about the best way to start.

Discuss your project