Agentic AI has gone from a frontier-lab concept to a category with real revenue inside Indian companies in roughly twenty-four months. We have been tracking the Indian agentic-AI cohort closely over the last year — 47 companies as of March 2026, across eight verticals, with combined ARR we estimate at $180–$220M and combined headcount of just under 3,400. This piece is a map.
The map, by vertical
India agentic AI — companies plotted by category traction
Each point is a vertical. X-axis: number of companies tracked. Y-axis: combined estimated ARR ($M). Sales and Support are pulling away; Procurement and Finance are early but growing fast.
Two verticals — sales/GTM and support/CX — are pulling ahead on both company count and combined revenue. They share a property: the workflow being automated has well-defined input/output, an existing system of record (Salesforce, HubSpot, Zendesk, Freshworks), and a strong-enough cost line that the buying decision is uncontested. Procurement and finance are smaller today but growing fast — both have similar structural fundamentals and are 12–18 months behind sales in maturity.
Why agentic is a venture-scale Indian opportunity now
There are three things true today about agentic AI in India that were not true 24 months ago. Together they explain why the category has gone from interesting to fundable to compounding inside the same cycle.
- Frontier model performance crossed the threshold for real workflow ownership. A 2024 model could draft a response; a 2026 model can complete the workflow end-to-end with reliable tool-use. The substitution math changed.
- Indian buyers crossed the threshold for paying for agents the way they pay for software seats. The buyer who would not budget $200/seat/month for an agent in 2024 will budget $40/transaction in 2026 — and the per-transaction model fits Indian B2B procurement.
- Talent crossed the threshold. India now has roughly 2,800 engineers who have shipped a production AI agent. Two years ago that number was an order of magnitude smaller. The team you need to start an agentic company is now hire-able in India.
Revenue per FTE — the new benchmark
The most interesting metric to watch in agentic businesses is revenue per FTE. Traditional Indian SaaS, even the best, tops out around $260K ARR / FTE. The best Indian agentic businesses are running 2–3× higher because the per-customer service cost is, by design, much lower.
Revenue per FTE — Indian agentic AI vs SaaS comparisons
Traditional Indian SaaS (median)
$180.0K · $180.0K
Traditional Indian SaaS (top decile)
$260.0K · $260.0K
Indian agentic AI (median)
$320.0K · $320.0K
Indian agentic AI (top decile)
Approaching US AI-native benchmarks
$720.0K · $720.0K
$K ARR / FTE
Composite. Agentic figures are 2025–26 reported or strongly inferred ARR / current headcount. SaaS comps are 2024 annual reports.
The revenue-per-FTE thesis is what makes agentic AI an unusually attractive category for operator-led capital. The unit-economics math at exit is closer to US comparables than traditional Indian SaaS has ever been. A $50M ARR Indian agentic business in 2027 will, on current trajectory, employ ~120 FTE — a number that supports a 2x–3x ARR exit multiple in a sale or an attractive structure in a public listing.
The structural risks
- Frontier-model commoditisation. If the difference between Claude 5 and GPT-6 collapses, vertical agents lose differentiation pressure from one direction.
- Buyer commoditisation. If Salesforce, HubSpot, and Zendesk ship native agents that are 70% as good and 'free' to existing customers, vertical agents lose differentiation pressure from the other direction.
- Talent flight. The agentic engineers in India are the most competitively recruited in the country. The company that cannot retain its founding three engineers loses 18 months of progress in a quarter.
Our read: differentiation in 2027–28 will come from depth in a specific vertical (data, integrations, workflow expertise, customer relationships) rather than from model layer. The companies investing in vertical moat in 2026 will be the ones that survive the inevitable platform-side competition.
What we are funding
Pratyaya has six investments in the agentic-vertical-AI space at the time of writing. Five are in workflows where the partner sponsoring the investment has direct operating experience in the target buyer's function. We are not funding broad horizontal agents, multi-agent platforms, or 'AI for general work' — we believe the venture-scale outcomes are in the verticals where a founder can credibly out-detail a generalist agent on a specific workflow.
“The Indian agentic-AI category will produce its first $1B-revenue private company by 2029. The companies on a credible path to that outcome are visible now. There are roughly twelve of them.”
Methodology
47 companies tracked: India-headquartered or India-founded, primary product is an AI agent with autonomy over a defined workflow (not a copilot, not chat-only), and revenue or active design-partner status as of March 2026. Combined ARR is bottom-up from disclosed and inferred customer counts × deal size; combined headcount is from LinkedIn-disclosed counts as of March 2026. Vertical assignment is to the company's primary buyer function; multi-vertical companies are assigned to the largest revenue line.