Skip to content

Process

How We Work

Every engagement follows the same path: understand the risk, assess properly, prove value on one use case, then deploy with governance built in. Scope, acceptance criteria and prices are agreed in writing before work begins.
  1. 01

    AI Data Risk Check

    Free · 30 minutes

    What happens

    A conversation with your leadership or IT lead about how AI is used today, what data is involved and what you want to achieve.

    Who's involved

    One or two people from your side; one Novatra specialist.

    What you get

    A one-page summary of your biggest AI data risks and a suggested first step.

  2. 02

    Audit

    Fixed scope

    What happens

    A structured assessment of AI use, data flows, vendor terms, policies and controls, using Novatra AI Audit and interviews with key teams.

    Who's involved

    Leaders from IT, risk or compliance, and the teams that use AI most.

    What you get

    A readiness and risk picture, a prioritised action list and a recommended pilot.

  3. 03

    Pilot

    One use case

    What happens

    We build one high-value use case, such as document AI or a knowledge assistant, in the environment you choose, and test it on your real data.

    Who's involved

    A business owner, a reviewer and your IT contact.

    What you get

    A working system measured against acceptance criteria agreed in writing before work starts.

  4. 04

    Deploy & govern

    Ongoing

    What happens

    Production rollout with access controls, monitoring, human review and an incident process, plus training for users.

    Who's involved

    Your system owners, with Novatra support as agreed.

    What you get

    A governed production system and, if you choose, managed operation and improvement.

Engagement models

Clear scope, clear pricing

Engagement models
EngagementFormatPricing
AI Data Risk Check30-minute call and one-page summaryFree
AI AuditFixed scope, agreed in writingQuoted after the Risk Check
PilotOne use case, with acceptance criteriaQuoted after the Risk Check
Managed sovereign AIMonitoring, support and improvementsMonthly, agreed in writing

Your part

What we need from you

  • A business owner who can say what success looks like.
  • A representative sample of real documents or questions for testing.
  • A named reviewer who can check outputs during the pilot.
  • An IT contact for access and the deployment environment.
  • Timely decisions on scope, risks and acceptance.

Our principles

What never changes

  • 01

    Data under your control

    Deployed on your own servers, a private cloud, or an India-region cloud of your choice.

  • 02

    No training on your data

    Your documents never train our models or any third-party model.

  • 03

    Governed by design

    Human review, audit trails and access controls are built in, not added later.

  • 04

    Built for Indian languages and regulation

    Designed around the DPDP Act and sector rules, with support for Indian languages.

FAQ

Questions about working with us

Is the AI Data Risk Check really free?

Yes. It is a 30-minute conversation followed by a one-page summary, with no obligation and no sales deck.

Do we have to do an audit before a pilot?

Not always. If your use case and data are already clear, we can scope a pilot directly. The audit is recommended when AI use is widespread or unclear.

How are prices set?

Audits and pilots are fixed-scope engagements, quoted in writing after the Risk Check once we understand your data, systems and goals. Managed services are priced monthly.

Who owns what we build?

Ownership and licensing are set out in the written scope. Your data always remains yours.

Can you work with our IT team?

Yes. Most deployments are a joint effort: your team knows your systems and policies, and we bring the AI engineering and governance.

Find out where your data is going, in 30 minutes

Book a free AI Data Risk Check. We'll map where AI is used in your organisation today, flag the biggest risks, and send you a one-page summary. No obligation.