Forward-deployed AI consulting in Hyderabad
Hyderabad is one of the six Indian cities that host most of the country's global capability centres, and it is growing quickly. The centres here build for parent organisations in banking, life sciences, technology and enterprise services. The question for a Hyderabad centre is the same one a Bengaluru centre asks: who owns the agent after the build, and how the parent can inspect it.
I am Mayur Patil, based in Nashik, India. I work with Hyderabad centres on forward-deployed AI consulting, and I am in Hyderabad often for engagements.
What the numbers say
Zinnov counts more than 355 GCCs in Hyderabad. Pune and Hyderabad together hold about 24 per cent of India's GCC research and development talent, and the six cities including Hyderabad host 92 per cent of the country's GCCs. Zinnov's BFSI analysis counts 51 BFSI GCC units in Hyderabad, out of 333 nationally. These figures describe a city where the first agent is usually built inside a centre, which makes the operating model the first decision.
The state's AI mission
The Telangana government's AI mission, T-AIM, is an AI-specific organisation set up in partnership with NASSCOM. The government describes it as driving the execution of six pillars of the state's AI framework: datasets and the data exchange platform, talent and education, research and innovation, high-performance computing labs, governance, ethics and privacy, and enabling AI adoption with community collaboration. For a centre, the governance pillar matters most, because it sets the expectations a state-backed partner will look for in a build.
Pilots that stall in Hyderabad
Three reasons come up most often. The first is that the business unit that will use the system is not in the room, so the centre builds something the unit did not ask for. The second is that the evaluation is a demonstration rather than an agreed test set. The third is that the agent can act, but no one has listed the actions that need a person to approve them.
A fourth is the handover. A centre that builds an agent for a parent needs the parent's risk team to see the record, and it needs a named owner in the parent who will run the system when the centre's team moves on.
What the rules mean for a centre
The Digital Personal Data Protection Rules were notified on 13 November 2025. Most of their obligations apply about eighteen months after notification, around May 2027. A centre that processes Indian personal data through an agent needs a data map, a lawful basis and a working route for people to exercise their rights before that date. The India AI Governance Guidelines, released on 5 November 2025, are non-legislative, but a parent's risk team will ask whether a centre has followed them. The India pillar sets out each with its source.
One example use case
The example below is an illustration of the work, not a client engagement.
Example: a vendor-onboarding agent for a BFSI centre. A centre that supports a bank's procurement team receives onboarding packs from new vendors, each with registration papers, bank details and certifications. The proposed agent checks each pack against the bank's checklist, flags missing items, and drafts the message to the vendor. It does not change the vendor master record. The readiness review settled three points. Vendor bank details are personal data of the vendor's staff, so the agent sees only what the checklist needs. The agent never changes the vendor record: a procurement officer approves each change. And the evaluation set is built from past onboarding packs whose outcomes the procurement team has already reviewed. The first slice caught packs where certifications had expired, which became a check the agent now runs on every pack.
What the readiness review produces
The AI Delivery Readiness Review is two weeks on one workflow, with the business unit in the room. It produces four written things: a map of the decisions the agent makes, a data map that lists the personal data each step reads, an approval list with the named approver for each action that changes a record, and an evaluation plan with the test cases and the owner who signs them off. It ends with a recommendation to proceed, reframe or pause.
Who owns the agent in a centre
The question that decides most Hyderabad projects is who owns the agent once it runs. In a centre that serves one parent, the owner is usually a business leader in the parent, and the centre's team keeps the system running for that leader. In a centre that serves several business units, each unit needs its own owner, and the centre needs a rule for which changes it can make without asking. Writing that rule down before the build saves a dispute later, when a model update changes what the agent does.
The owner also decides the evaluation. The owner signs off the test cases, because they know what a correct answer looks like in the business. A centre that builds the evaluation without the owner tends to test what is easy to test, not what the business needs.
Working with a parent in another time zone
Hyderabad is on India time, so there is no offset to manage inside the centre. A parent in the US or Europe is a different matter. I run back-to-back full-day corporate trainings, mostly in IST, and some programmes run in US Pacific time for US-based teams. Overlap with a parent is agreed per project, usually as a short window for decisions and reviews, with the build done in the centre's working hours.
How an engagement runs
An engagement starts with the two-week AI Delivery Readiness Review on one workflow, with the business unit in the room. It ends with a written recommendation to proceed, reframe or pause. If the work proceeds, an architecture sprint decides the operating model for the first system, and a forward-deployed delivery engagement builds it with the centre's engineers.
Three ways to start
AI Delivery Readiness Review. Two weeks on one workflow, with a written readout and a recommendation.
Architecture Sprint. Two to four weeks on one use case, with the architecture, recorded decisions and a backlog.
Forward-deployed AI delivery. One bounded use case, built with the centre's engineers, with a runbook and a named owner.
Each is quoted on its scope. The engagements page describes what each includes. I do not publish prices.
Relevant work
The organisations I have worked with and can name are listed on the about page. Among them are Walmart, Novo Nordisk India and Tredence. Walmart's procurement teams have been through a three-day programme on Microsoft 365 Copilot and agent tooling, and Novo Nordisk India has taken generative and agentic AI enablement. I do not describe the detail of that work beyond what the clients have agreed to share.
Where to go next
The FDE consulting hub covers the three ways to engage. For the centre model, read the global capability centres page. For the rules, the India pillar is the reference.
Common questions
Can a Hyderabad centre run forward-deployed work for a business unit in the US or Europe?
Yes. The operating model decides who owns each system. A centre can work with one business unit at a time, with a named owner in the business and a runbook the centre keeps. Overlap hours are agreed for each project.
What does the DPDP timeline mean for a centre processing Indian personal data?
The Digital Personal Data Protection Rules were notified on 13 November 2025. Most obligations apply about eighteen months after notification, around May 2027. A centre should map its personal data and design its agents to meet those obligations before that date.
Does Telangana's AI mission give centres a programme to join?
The Telangana AI Mission, T-AIM, is an AI-specific organisation set up with NASSCOM to drive the state's AI framework. Its six pillars include governance, ethics and privacy. Whether a particular centre takes part in a T-AIM programme depends on the programme, which I would check with the mission directly.
Do you run on-site work in Hyderabad?
Yes. I am in Hyderabad often for engagements. Discovery and the handover can be on site, and the build can run from India with a fixed overlap for decisions.
Sources
- Telangana government, Artificial Intelligence initiatives (T-AIM and the six pillars of the state AI framework)
- Zinnov, five shifts defining India's GCC story in 2025 (355+ GCCs in Hyderabad; 880+ in Bengaluru)
- Zinnov, the role of six cities in India's GCC growth (Pune and Hyderabad hold about 24 per cent of GCC R&D talent)
- PIB, Digital Personal Data Protection Rules notified (November 2025)