AI and automation
Practical AI for Indian businesses. We find the tasks worth automating, build working tools on proven models, and keep your data and decisions under control.
Why AI now
Language models can now read documents, answer questions and draft text well enough to take real work off a team, at a cost that makes sense for small and mid-sized companies. Customers in India already expect quick answers on WhatsApp and the web, in their own language.
The risk is spending money on pilots that never reach production, or sending data where it should not go. We focus on a few use cases with clear value, build them properly and govern them under the Digital Personal Data Protection Act, 2023.
Use cases by function
Examples of what we build. Every one keeps a person in charge of decisions that matter.
Customer support
- A support assistant on your website and WhatsApp that answers common questions from your own policies
- Ticket triage that tags, prioritises and routes requests to the right person
- Draft replies for agents to check and send
Sales
- Lead qualification and follow-up reminders from your CRM
- Proposal and quotation drafts from a short brief
- Call and meeting summaries with next steps
Finance and invoicing
- Reading supplier invoices and matching them to purchase orders
- Flagging invoices with missing GST details before they are paid
- Collections reminders tailored to each customer
HR and recruiting
- Screening applications against a clear job brief, with a person making the decision
- Answering employee policy questions from your handbook
- Interview scheduling and summaries
Operations
- Demand and staffing forecasts from your own history
- Daily exception reports instead of manual checks
- Automated handoffs between tools your team already uses
Compliance and KYC document checks
- Extracting details from PAN, GST certificates and business documents
- Checking documents for missing or mismatched fields before review
- Audit-ready logs of every automated step
Marketing content
- First drafts of product descriptions, emails and posts in your brand voice
- Content in Hindi and other Indian languages, reviewed by a person
- Summaries of customer reviews and feedback themes
Our approach
Assess
We map tasks, data and risks, and rank use cases by value and effort. One to two weeks.
Prototype
A working version on your own approved data, ready for your team to try in about two weeks.
Deploy
Integration with your systems, access controls, monitoring and training for your team.
Govern
Monthly review of quality, cost and risk, with improvements shipped as we learn.
Responsible AI and data privacy
The controls we put around every AI system we build.
Data minimisation
Only the data a task needs is sent to a model. Personal data is masked or removed wherever the task allows.
No training on your data
We use vendor settings that do not train models on your data, and never use client data for training without written consent.
Human review
Decisions that affect customers, employees or money are reviewed by a person. The system suggests; your team decides.
Audit logs
Inputs, outputs and actions are logged so you can see what the system did and why, and correct it.
Vendor due diligence
We review each AI vendor's data handling, storage location and security before it touches your data, and record the results.
DPDP Act alignment
Purposes, consent, retention and data principal rights are mapped for each use case, in line with the Digital Personal Data Protection Act, 2023.
Packages
Start small and scale what works. Each package has a fixed fee in a written proposal.
AI readiness assessment
A ranked list of use cases, data and risk review, and a costed roadmap. One to two weeks.
Pilot
One use case built and tested on your own approved data, with a clear go or no-go decision.
Production rollout
The pilot hardened, integrated with your systems, secured, monitored and handed to your team.
Managed AI operations
A monthly retainer: we run, monitor and improve your AI tools and report on quality and cost.
Questions about AI
Will our data be used to train AI models?
No, not without your written consent. We use enterprise settings and vendors that do not train on customer data, and we document where your data goes.
Which AI models do you use?
We choose per task from leading commercial models, such as those from OpenAI and Anthropic, and open-source models where data must stay in your own environment.
What does a pilot cost?
Each package has a fixed fee set out in a written proposal after a short scoping call. GST is added as applicable.
What if AI is not the right answer?
We say so. Some problems are better solved with a simpler process change or an ordinary automation, and we will recommend that instead.
Do you support Indian languages?
Yes. We can build assistants that work in Hindi and other Indian languages, with a person reviewing quality before launch.
Let's build in India.
Tell us what you are building and where you want to be in a year. We reply within one business day.