Questions owners actually ask us.

We do not publish pricing because we do not sell a packaged product. Each engagement is bespoke and fixed-fee, committed in writing after a two-week diagnostic — never before. Most first rollouts for medium businesses sit in the mid-six to low-seven-figure range in INR. Custodianship retainers start at ₹3L / month. You will never receive a surprise invoice.

Published pricing leads to mis-scoped work — you pay for features you don't need, or end up in a change-order spiral. We would rather spend two weeks understanding your operations, write you a fixed number we can stand by, and let you decide from there. The diagnostic is a small fixed fee, credited against the rollout if you proceed.

The Brief — two weeks, ends with a written, signed scope. The Engagement — eight to fourteen weeks for a single rollout, depending on complexity. The Custodianship — minimum one year, ongoing review and maintenance. You can stop between any two phases.

We prefer it. A narrow first engagement — one agent, one chatbot, one integration — lets us prove value and lets you measure the impact before committing further. Many of our long-term clients began with a single eight-week pilot.

Six per quarter is our ceiling. We are a small, senior practice by design; we would rather turn down work than under-serve it.

Primarily Anthropic Claude (Sonnet and Opus) for reasoning-heavy workflows, OpenAI GPT-4 class for some generation tasks, and open-source models (Llama, Qwen) where on-prem deployment matters. We choose per-task, not per-brand.

Yes. Most of our integration work lives on top of what you already use. We have deployed on Tally, Zoho Books, Zoho CRM, Salesforce, SAP B1, Microsoft 365, Google Workspace, and half a dozen industry-specific systems. If you already have it, we integrate with it — we do not ask you to rip and replace.

Both, where appropriate. For orchestration we often use n8n (self-hosted) which makes the system legible to your future IT team. For agent logic we write proper, typed code. We never leave you with something you cannot maintain.

Generic AI tools fail for three reasons: (1) they are not wired into your actual data, so they guess or hallucinate; (2) they have no opinion — they help whoever opens them but change no workflow; (3) no one owns the rollout. Our engagements are designed as the opposite of all three: grounded in your data, wrapped around a specific workflow, and handed over to a named custodian on your side.

Our default is your cloud account (AWS, Azure, GCP) or your on-prem. We do not operate a multi-tenant SaaS — your data does not sit in a Svachalita database. For the AI inference itself, we use enterprise-grade endpoints with no-training commitments (Anthropic Claude for Business, Azure OpenAI); prompts and outputs are not used to train models.

Every agent we build logs every action — input, output, data touched, human who reviewed. Role-based permissions are standard. Leadership gets a weekly digest automatically. We do not ship anything your compliance officer cannot inspect.

Always. Before the diagnostic begins. We carry professional-indemnity insurance and will happily provide a certificate.

Every deployment has a human-in-the-loop boundary — the agent drafts, a human approves. For customer-facing chatbots, we use retrieval grounding (answers cite source documents) and escalation triggers; when in doubt, the bot hands to a human. Accuracy is something we measure and report on, quarterly.

No. An AI agent removes the repetitive substrate — reconciling spreadsheets, drafting templated emails, chasing status updates — that juniors resent and that burns out seniors. You keep headcount; your people spend their day on work that deserves them. Every client we have worked with has seen retention improve, not reduce, after rollout.

Carefully, and honestly. We run a 90-minute listening session with the affected team before architecture begins. People's fears are usually specific (losing the part of the job they love, looking slow, being measured differently) — and usually addressable. Rollout includes training cohorts. No one is surprised by their own tool.

Not to begin. A rollout needs a named custodian on your side — someone who meets us weekly, makes decisions, and owns the handover. They can be operations, finance, HR, or IT. By month twelve, that custodian can run most of the day-to-day without us.

Yes, for well-scoped tasks: WhatsApp triage, customer Q&A, form extraction, summarisation, basic drafting. Full open-ended reasoning in Hindi is improving monthly; we choose tasks AI is ready for today and tell you honestly when something is not. Every Indian-language deployment is reviewed by a native speaker before it touches a customer.

Major regional languages (Marathi, Gujarati, Bengali, Tamil, Telugu, Kannada) work well for the same narrow-scope tasks as Hindi. Rajasthani dialects and mixed-script WhatsApp messages are a known hard case; we handle them with explicit escalation rules rather than pretending the AI has understood.

Yes. We deploy only on the Meta WhatsApp Business Cloud API through a verified BSP. It is the only way to be compliant, scale beyond a few hundred messages a day, and access features like button-replies and templates. We handle the onboarding — green-tick verification takes around two weeks.

Yes, based in Alwar, Rajasthan. Discovery always begins on-site — we do a week at your office wherever you are. Rajasthan, NCR, Mumbai, Bengaluru, Pune, Ahmedabad, Hyderabad — all routine. Further afield, we build a per-engagement travel plan.

GST-registered in Rajasthan. Proper tax invoices, compliant with Indian accounting practices. We do not sub-contract critical work overseas; your data stays where we said it would.

Two weeks of stabilisation at no additional cost. Then, optionally, a custodianship retainer — we monitor, patch, and extend as models improve. Many clients choose this for the first year; some take over in-house earlier. There is no penalty for either.

You already own the code, the documentation, and the custodianship handbook. We run a knowledge-transfer session with your chosen in-house lead at no additional cost. We will also refer you to a local maintenance engineer if helpful.

Yes — named references after a signed diagnostic. We check with clients before making any introduction. A few clients have asked to stay private; we respect that without exception.

Not a canned demo — we find those misleading. What we can do is a 30-minute listening call, and if there's a fit, an on-site half-day where we walk your operation and point at three places we'd start. That half-day is free.