VLIRTZ

Pricing

What this costs

Most agencies make you extract this over three calls. Here are the shapes of work we sell, what each one includes, what it excludes, and what moves a quote inside a range.

We are finalising published amounts for each engagement. In the meantime the scope, timeline and cost drivers below are accurate, and you will get a real range on the first call rather than the third.

Get a range for your workflowHow we build

How we price, and why

We sell scoped engagements with a defined outcome rather than hours. An hourly rate pays us for taking longer, which is the wrong incentive on work where the goal is a system that runs without us.

Every band on this page is a range rather than a fixed number, because a fixed price quoted before anyone has looked at your data is either padded to cover the worst case or about to be revised. What we publish instead are the drivers that decide where inside the range you land, so you can estimate yourself before speaking to us.

Build cost and running cost are separated deliberately. They get conflated constantly, and the second one is where people are surprised.

Engagements

The five shapes of work we sell

Reference amounts in EUR, excluding VAT. Market pages quote the same engagements in local currency.

AI opportunity review

On request

Teams with several candidate use cases and no agreed order of work. Ends in a written recommendation, including a recommendation not to build where that is the honest answer.

Timeline: 1 to 2 weeks

Included

  • Workshops with the people who own the workflow
  • Volume and handling-time measurement on the real process, not an estimate
  • Candidate use cases ranked by payback and delivery risk
  • A written recommendation with a scoped first build

Not included

  • Any production code
  • Access to your production systems
  • Vendor or model procurement on your behalf

Scoped agent build

On request

One well-understood workflow that a human runs today and that needs to run consistently. This is most of our work.

Timeline: 2 to 4 weeks

Included

  • One AI agent running one workflow end to end
  • Integration with the tools that workflow already touches
  • Human review gates on every action that is expensive to undo
  • An evaluation set built from your real cases, including the failures
  • Runbook and handover so your team can operate it without us

Not included

  • A second workflow, which is scoped separately
  • Data cleanup where source records are unusable
  • Ongoing operation, which is a separate retainer

AI consulting sequence

On request

Organisations that need an architecture and rollout plan their own engineers will implement, rather than a delivery partner.

Timeline: 4 weeks

Included

  • Target architecture with the build-versus-buy calls made explicit
  • Data residency and processor mapping against your regulator
  • A rollout sequence with a measurable checkpoint per stage
  • Working sessions with your engineers, not a handover deck

Not included

  • Implementation, though we can quote it separately
  • Ongoing architecture governance

Multi-workflow build

On request

Companies that have already put one agent into production with us and want the next three, sharing evaluation and observability.

Timeline: 3 months and up

Included

  • Several agents on shared orchestration and retrieval
  • Shared evaluation harness and observability
  • Role-based permissions and audit logging
  • Staged rollout with a rollback path at each stage

Not included

  • An open-ended retainer with no defined outcome
  • Work we have not scoped against a measured baseline

Sustain retainer

On request

Systems already in production that need model updates, drift monitoring, and someone on the hook when something breaks.

Timeline: Rolling, 30 days notice

Included

  • Monitoring, drift checks, and model or prompt updates
  • Defined response time on failures
  • A monthly note on what changed and what it cost

Not included

  • New workflows or new integrations
  • Unlimited feature development

What moves the number

Why two projects with the same description cost differently

These are the questions we would ask on a first call anyway. Published so you can work out roughly where you land before speaking to us.

Number of systems the agent touches
One system is straightforward. A workflow spanning a CRM, a mailbox, and a database costs more in integration and error handling than it does in model work.
How usable your data already is
Retrieval on clean, well-structured documents is cheap. Retrieval over scanned PDFs, inconsistent record structures, or three competing sources of truth is where budgets actually go.
Cost of a wrong action
An agent that drafts a reply for a human to approve needs far less guardrail engineering than one that issues a refund or files a regulatory return.
Regulatory scope
Work under FINMA, Finansinspektionen, or medical-device rules carries documentation and audit obligations that are real engineering effort, not paperwork added at the end.
Who operates it afterwards
If your team takes over, we spend more on runbooks and evaluation tooling up front. If we sustain it, that shifts into the retainer instead.

Running cost

What it costs to keep running, separately from building it

These get conflated, and the second one causes the surprises. The model bill is almost never the expensive part.

Model and API usage
Typically the smallest line. A single-workflow agent at moderate volume usually runs tens of euros per month, not hundreds.
Hosting and vector storage
Usually a low fixed cost, and often absorbed into infrastructure you already pay for.
Maintenance
The real recurring cost. Models change, your source systems change, and an agent nobody maintains degrades quietly rather than failing loudly.

By market

The same engagements in your currency

Each market page prices this catalogue locally, alongside the regulators and sectors specific to that market.

FAQ

Questions about pricing

How much does it cost to build an AI agent?
A scoped single-workflow build is the usual entry point, with a shorter feasibility review ahead of it when the use case is not yet settled. What moves the number is how many systems the agent touches, how usable your data already is before anyone cleans it, how expensive a wrong action would be, and whether your team or ours operates it afterwards.
Why publish ranges instead of a fixed price?
Because a fixed price quoted before anyone has looked at your data is either padded to cover the worst case or about to be revised. The ranges here are real, and the drivers that determine where you land inside them are published alongside so you can estimate before speaking to us.
Do you charge hourly?
Not by preference. Scoped engagements with a defined outcome align our incentives with yours better than an hourly rate, which rewards us for taking longer. For genuinely open-ended advisory work we will agree a day rate, but we will usually suggest a scoped alternative first.
What is not included in a build?
A second workflow is scoped separately. Data cleanup where the source records are unusable is a separate project, and we will say so during discovery rather than absorbing it silently. Ongoing operation is a retainer you choose rather than something bundled in.
What does it cost to run once it is built?
Model and API usage is usually the smallest line, often tens of euros a month for a single-workflow agent at moderate volume. Hosting and vector storage is typically a low fixed cost. Maintenance is the real recurring expense, because models and your source systems both change, and an unmaintained agent degrades quietly rather than failing loudly.
Will you tell us not to build something?
Yes, and it has happened. If discovery shows the source data cannot support the workflow, or that the useful deliverable is a documented process rather than software, that is the recommendation you get. It is a worse outcome for our invoice and a better one for your budget.
Do you quote in local currency?
Yes. Each market page carries the same engagements priced in that market's currency: SEK for Stockholm, DKK for Copenhagen, CHF for Zurich, EUR for Amsterdam and the wider EU, and AED for Dubai. Amounts exclude VAT.
Is there a minimum engagement?
In practice the feasibility review is the smallest thing worth buying from us. Below that there is not enough time to measure the workflow properly, and a recommendation made without measurement is the thing we are trying to avoid selling.

This page was last reviewed on .