veltacta.
Start a project
How we work

How we work.

Five phases. Each one ends with something you can hold.

01

Discover

1 to 2 weeks
What happens

A workshop with the people who own the problem, stakeholder interviews, an audit of the data and systems in play, and a risk list. You get scope options with estimates, not one number.

You receive
  • Scope document with options and estimates
  • Risk list
  • Success metrics
02

Design

2 to 4 weeks
What happens

Flows and a clickable prototype, the design system, the technical architecture written as decision records. For AI work, the evaluation plan: which historical cases, labelled how, judged against what.

You receive
  • Clickable prototype
  • Architecture decision records
  • Evaluation set definition
03

Build

Two-week iterations
What happens

CI/CD from the first commit, a preview environment per change, automated tests plus manual QA on every increment. You see working software every two weeks, with release notes.

You receive
  • Working increments
  • Test reports
  • Release notes
04

Launch

1 to 2 weeks
What happens

Shadow traffic beside the old system where there is one, a migration plan with a rollback, monitoring wired before the switch, training for the people who will run it.

You receive
  • Launch checklist
  • Runbooks
  • Dashboards
05

Run

Monthly, under SLA
What happens

SLA support, monthly performance and cost reviews, model and dependency updates. One report a month: uptime, cost, defects, improvements.

You receive
  • Monthly report: uptime, cost, defects, improvements
Team composition

The people who scope are the people who build.

No junior-only teams, no account-manager layer.

AI controls, stated plainly

Nothing high-stakes is automatic unless you decide it should be.

Evaluation sets
Before production, from your real cases. Re-run on every change.
Human approval
Where the cost of error is high. Threshold agreed with you.
Audit log
On every automated decision, with model version and prompt hash.
Model-agnostic
We name the model in use and show how it is switched.
Data
Stays in the region you choose. No training on your data.
EU AI Act
Article 50 disclosure and content marking built into every assistant.
Ownership
All code, infrastructure, data, prompts and evaluation sets are yours at delivery. Repositories live in your organisation from day one if you prefer.
Working hours
CET / CEST. Overlap with London all day and with New York until 12:00 ET.
Support SLA · run retainer
PriorityExampleFirst responseWorkaround or fix
P1 Production down, orders blocked Within working hours; outside hours on the retainer’s terms Same day
P2 Degraded, workaround exists Same business day Within two business days
P3 Defect, no business impact One business day Next release

Exact response times are set per retainer and written into the contract.

Engagement models

Four ways to engage. Scoped, priced and agreed in writing.

The bar

The engineering standard every system is held to.

Six non-negotiables, agreed in the proposal and evidenced in every release. They are how we define production-grade, and how you can verify it.

Verified on every release · evidence delivered with it
  1. 01

    Continuous delivery

    Every change is built, tested and deployed to an isolated preview environment before review. Release is a routine event, not a project milestone.

    Evidence Preview environment per change · deployment log
  2. 02

    Quality assurance

    Automated test suites complemented by structured manual QA. Coverage targets are set per system and reported against.

    Evidence Test results in every release note
  3. 03

    Accessibility

    WCAG 2.1 AA conformance on everything we ship, the standard referenced by the European Accessibility Act.

    Evidence Automated checks per change · manual audit before launch
  4. 04

    Performance

    Core Web Vitals and response-time budgets set during design, monitored in production, and treated as acceptance criteria.

    Evidence Budgets in the proposal · live performance dashboard
  5. 05

    Observability

    Structured logs, metrics, tracing and alerting designed in from the first increment and operated by your team after handover.

    Evidence Dashboards, alert rules and runbooks in your accounts
  6. 06

    AI governance

    Evaluation sets from your own cases, confidence thresholds, human approval where the cost of error is material, a full audit trail, and model portability by design.

    Evidence Evaluation report per model or prompt change · audit log
Start a project

Tell us what you are trying to achieve. We reply within one business day.

Send a few lines about the problem and the people it affects. The reply comes from the engineer who would lead the work.

  1. 01 We read the brief and reply in writing.within 1 business day
  2. 02 A 30-minute call to understand scope and constraints.same week
  3. 03 A written proposal: options, assumptions, estimate.within 5 business days
  4. 04 Kickoff with the team that scoped it.on your timeline
Rather talk? Book a 30-minute call → or hello@veltacta.com
NDA on requestEU data residencyCET · overlap with London and New York
Project brief
about 3 minutes
Optional

We use your details only to reply. Privacy policy