Question 1

How do engagements start?

Accepted Answer

With a short discovery engagement. We map the use cases worth pursuing, test feasibility against your actual data and produce an architecture and a costed plan — so you can decide whether to build with full information and without committing to a large project first.

Question 2

How long does a project take?

Accepted Answer

We scope pilots to six to nine weeks and full builds to three to five months, depending on integrations and internal approvals. The timeline is agreed and written down before work starts.

Question 3

Who owns what we build?

Accepted Answer

You do. Deliverables built for your engagement, along with the source code and documentation, are handed over to you. We retain our own general tooling and know-how. The specifics are set out in the agreement before work begins.

Question 4

Can our data stay inside our own environment?

Accepted Answer

Yes. We can deploy inside your cloud account, VPC or on-premise infrastructure, using open-weight models where a third-party API is not acceptable. The right architecture depends on your data and constraints, and we work that out during discovery rather than promising an outcome up front.

Question 5

Are you certified to SOC 2 or ISO 27001?

Accepted Answer

No — we are not currently certified to either, and we would rather say so than imply otherwise. We design and build to those control expectations: encryption in transit and at rest, least-privilege access, audit logging and network isolation. We will complete your security questionnaire and sign a data processing agreement before any engagement begins.

Question 6

Which models do you use?

Accepted Answer

We are model-agnostic. Depending on accuracy, latency, cost and data-residency needs we work with Anthropic Claude, OpenAI, Google Gemini and open-weight models such as Llama and Mistral — and we will tell you why we picked one for your use case.

Question 7

How do you reduce hallucinations?

Accepted Answer

Grounding responses in retrieved source material, structured outputs with schema validation, checks that reject answers the sources do not support, and human review where the cost of being wrong is high. We also build an evaluation suite so accuracy is measured rather than assumed.

Question 8

Do you handle ongoing operations?

Accepted Answer

Either way works. We can continue to operate and improve the system, or document and hand it over so your team owns it. Many engagements start with the first and move to the second.

Question 9

How is pricing structured?

Accepted Answer

Discovery and pilots are fixed-scope so you know the cost up front. Larger builds are quoted per engagement once the scope is clear. We do not publish rate cards, because a number quoted without understanding the problem is not useful to either of us.

Question 10

Where are you based, and who do you work with?

Accepted Answer

We are based in India and work with organisations across the country, delivering remotely with on-site time where an engagement genuinely needs it.

Frequently asked questions

Engagement

Security

Technical

Commercial

Still have questions?

Ready to build AI that actually gets used?