Pricing
Priced per approved story.
One billing unit: a story, one piece of work small enough to specify, test and review on its own, that you approved and we delivered under governance. No seats, no tokens as a line item, and bug stories are never billable.
- Per approved story
- $60–$85
- Support, flat
- $500 / month
- Governed Pilot
- $3,500 · 4 weeks
The billing unit
One approved story. One invoice line.
You pay for exactly what you asked for, approved before work started and verified before it closed.
- The unit never changes. There is no scope to negotiate.
- Nothing is billable until you approve it: the story, or the epic charter it belongs to.
- Story size is enforced by the platform: a story too big or too vague to test on its own is rejected before you can approve it.
- The price falls as your team takes on more. There is no penalty for building your own capability.
Never on your Arkoria invoice
- Bug stories
- Rework on a story blocked at review
- Specifications rejected before approval
- Seats
- Tokens as a line item
Three tiers
The name says what Arkoria manages.
Delivery and the AI, the AI only, or neither. Each step hands your team more of the work, and costs less per story.
-
Project
Managed Delivery
Arkoria engineers build it, on Arkoria inference. A direct path to production, and to throughput numbers from your own codebase.
$85per approved story
≈ R$ 425
- For
- Companies with a roadmap and nobody free to build it
- Who builds
- Arkoria engineers
- Contract
- A Statement of Work, priced per approved story: no fixed scope to negotiate, no hours to count
- Inference
- Included
- Support
- Included during the project
-
Product
Managed Inference
Your team builds it, on Arkoria inference. A fixed price per story, however much the model costs, with no AI provider to manage.
$70per approved story
≈ R$ 350
- For
- Teams that build, and want one predictable bill
- Who builds
- Your team
- Who operates
- Your tech leads; your stack and domain rules become rules the platform enforces
- Inference
- Included
- Support
- $500 / month, or $5,000 / year prepaid
-
Product
Bring Your Own Inference
Your team builds it, on your inference. Your choice of model, your existing commitments, and full ownership of the data posture.
$60per approved story
≈ R$ 300
- For
- Teams with their own model contracts and data rules
- Who builds
- Your team
- Who operates
- Your tech leads; your stack and domain rules become rules the platform enforces
- Inference
- Your provider account; token spend is yours
- Support
- $500 / month, or $5,000 / year prepaid
Pilot, project, product
Start on your own codebase, let us run the first project, then take the controls.
The lowest price on the list is the one where your team runs it.
-
Step 01 · 4 weeks
Governed Pilot
$3,500 fixed · ≈ R$ 17,500
One workload, up to five users, one epic of up to ten stories shipped under governance. What the pilot includes.
-
Step 02 · Per engagement
First project
Managed Delivery · $85 / story
Arkoria engineers operate the platform under a Statement of Work. An Arkoria forward-deployed engineer provides live support and enablement as your team takes the controls.
-
Step 03 · Ongoing
Product
Bring Your Own Inference · $60 / story + support
Your tech leads run the pipeline on your own infrastructure and inference account. Your code, backlog and documentation are yours throughout.
Delivering software for clients? Systems integrators build on Arkoria as delivery or referral partners, on terms agreed with each partner.
Support subscription
A flat fee. Never per seat.
For Managed Inference and Bring Your Own Inference. Support is included during a project.
Unlimited tickets
No per-incident charge.
12 hours a year of guided support
Calls and enablement sessions with our engineers.
Knowledge base
How-to guides and enablement material.
Upgrades and security patches
Always on the current release.
Support pricing
Annual prepay saves 16.7%, the only discount we offer.
What counts as a story?
A unit of work small enough to specify, test and review on its own. The specification gate checks every story against INVEST (independent, negotiable, valuable, estimable, small, testable) and rejects the ones that fail.
Who approves a story before it becomes billable?
You do: the story itself, or the epic charter it belongs to. Nothing is billable before a person on your side has approved it.
What is never billable?
Bug stories, whether the platform found the defect or you did: we are never paid to create a defect and then paid again to fix it. Rework on a story blocked at review, which goes back to the agents until it clears the gate and stays one story on the invoice.
What stops stories from being cut smaller to raise the bill?
The specification gate holds story size in place, and that call is made by an independent reviewer, never by the agent doing the work.
Why not price per token, or per seat?
Tokens and seats measure what AI costs the vendor. An approved story measures what you asked for and received. Pricing on our cost would reward us for consuming more. Pricing on your value rewards us for delivering more, with fewer defects. More on why.
Do we pay for inference separately?
Not under Managed Delivery or Managed Inference, where inference is included in the story price. Under Bring Your Own Inference you use your own provider account, so token spend and your tech lead's time are yours, along with the model choice and the data posture.
Which coding agents does it work with, and where does it run?
Natively on Claude Code today, with OpenAI Codex coming soon. The hub runs in Arkoria's cloud and reads your repository through GitHub's API, opening and merging pull requests on projects that use pull-request review; the agents that touch your source run on your infrastructure or ours, confirmed in writing for each account; and onboarding never modifies your source, CI, database or dependencies. Where it runs, and what it touches.
What exactly does the Governed Pilot include?
Four weeks from receipt of requirements, for one workload and up to five users: a planning report, a configured environment with the gates switched on, an exported INVEST backlog you keep whether or not you continue, one epic of up to ten stories shipped with traceability, a governance report, and up to two hours of handover.
Production cutover, migration and certification are out of scope. The price is $3,500, fixed: half on signature, half on delivery.
How do the prices in reais work?
Prices are set in US dollars. Amounts in reais are shown at an assumed policy rate of R$ 5.00 per US$, for illustration only; the applicable exchange policy is stated in each proposal, never a spot rate.
Prices in USD; BRL equivalents at an assumed policy rate of R$ 5.00 / US$, for illustration only. The applicable FX policy is stated in the proposal, never a spot rate. Under Bring Your Own Inference, token spend and your tech lead's time are additional to the per-story price. Commercial terms are indicative and subject to a definitive agreement.
Start small
Start with the project you keep putting off.
The terms are above; tell us the project.