C E H P O I N T
The Company That Will Make
Foreign AI Irrelevant in India
A sovereign AI infrastructure company building research, domain models, autonomous agents, and controlled distribution — so India's 1.4 billion people are never dependent on Silicon Valley for intelligence again.
Round
Seed · open
Instrument
Equity, on incorporation
Valuation band
₹25–75 Cr
Confidential · Investor Presentation · August 2026
01
The Case, In One Slide
Why This Is Not
Another AI Pitch.
If you read nothing else, read this. Six facts, each independently verifiable, that together describe an unusual position: a company already earning revenue in a market where almost everyone else is still burning it.
01 · We are already trading
Founded 2017. Nine years of continuous operation, 100+ delivered projects across 12 industries, paying customers today. This is not a pre-product team asking you to fund an idea — it is an operating business asking you to fund its next form.
02 · The infrastructure exists
Gateway, two models, 13 agent tools, vision, voice, image and video generation, a security practice, and a full sub-licensing hierarchy — all in production now. Nothing on the revenue slide depends on software we have yet to write.
03 · Distribution is built in
Incubators, accelerators, universities and agencies buy capacity and resell it to their own startups, students and clients. Our customers acquire our end users. Acquisition cost per end user: effectively zero.
04 · The incumbents are failing
95% of enterprise GenAI pilots return nothing measurable (MIT). Over 40% of agentic projects will be cancelled by 2027 (Gartner). The causes are integration, cost and trust — not model quality. That is the exact gap we are built for.
05 · Regulation runs our way
India's DPDP Rules bite fully by 13 May 2027. Sending customer personal data to a foreign AI service is third-party processing the Indian company stays liable for. Every enterprise in India has a deadline to find a domestic answer.
06 · The state is spending
IndiaAI Mission: ₹10,372 crore committed and 38,000 GPUs deployed, scaling toward 1,00,000. Public money is building the compute layer. Almost no Indian company is positioned to capture the commercial value on top of it.
The asymmetry, stated once: the Indian AI market is forecast at ₹1.1 lakh crore by 2034, the demand is being created by regulation and public spending, the failure rate of the foreign alternative is documented at 95% — and the company positioned against all three is already profitable enough to offer you a share of its profit rather than only a share of its hope.
02
The Crisis
India Is Being
Digitally Colonized.
THE REALITY NO ONE WANTS TO SAY ALOUD
  • ChatGPT has 100 million+ weekly users in India. Every conversation, every query, every thought — flows to servers in America. India is giving its best thinking away for free.
  • India's ₹11.7 lakh crore AI economy runs entirely on Nvidia chips, Azure, AWS, and Google Cloud. Not a single layer of the stack is Indian-owned.
  • OpenAI, Google, and Perplexity are entering India with aggressive pricing — Bernstein called it a "wake-up call": India's entire AI application layer is about to be commoditized by foreign firms.
  • Indian enterprises face 3,247 cyberattacks per week — double the global average — with zero indigenous AI-powered defence.
  • The models powering Indian healthcare, finance, and government can be switched off, repriced, or geo-blocked overnight by a foreign CEO.
WHAT THIS MEANS FOR INDIA
  • Every Indian business using OpenAI is renting intelligence from a company that answers to American shareholders, not Indian interests.
  • India's 22 scheduled languages are being mapped by foreign companies who will own the commercial rights to understanding how Indians think and speak.
  • When the next geopolitical crisis hits, India's AI infrastructure can be turned off like a tap — just as Russia's was.
  • The Indian government is spending ₹10,000+ Cr on the IndiaAI Mission — but no Indian company is positioned to capture the commercial value of that investment.
  • The window to build a sovereign AI industry is closing fast. In 3-5 years, foreign incumbents will be irreversible.
03
What's at Stake
This Is Not a Market Opportunity.
This Is a Civilizational Moment.
₹1.1 lakh crore
India's AI market by 2034 — growing at 26.5% CAGR. Every rupee of this that goes to a foreign company is a rupee India never gets back.
1.4B
People who will use AI in the next decade. If American companies own the models, they own the relationship with every single one of them.
22
Languages that need AI built from scratch — not translated from English, but understood from the ground up by machines trained on Indian thought.
₹11.7 lakh crore
India's digital economy — 12% of GDP — currently running on foreign infrastructure. This is the equivalent of building a highway and letting someone else own the toll booth.
"India's AI market will reach ₹11.7 lakh crore by 2032. Data centre capacity will grow 5× by 2030. The question is not whether India will adopt AI — it's whether India will own its AI."
— Wood Mackenzie / Ventura Research, August 2026
Figures originally published in US dollars are shown converted to rupees at approximately ₹83 to the dollar.
04
The Evidence · Published Research
Enterprise AI Is Failing.
The Reasons Are Documented.
Every figure on this slide is from published third-party research, not our own marketing. Together they describe a market that has spent several lakh crore rupees and has very little to show for it — and they name the precise failure modes we built the company around.
95%
of enterprise generative-AI pilots delivered no measurable P&L return, against ₹2.5–3.3 lakh crore of investment. MIT attributes it to the integration gap — generic tools that demo well and break inside real workflows.
MIT Project NANDA, State of AI in Business 2025
40%+
of agentic-AI projects will be cancelled by end-2027 — from a Gartner poll of 3,400+ organisations. Causes cited: escalating cost, unclear value, inadequate risk controls. None of them are modelling problems.
Gartner, June 2025
58–88%
hallucination rate for general models on verifiable legal questions in Stanford testing; purpose-built legal tools still hallucinated 17–33% of the time.
Stanford RegLab / HAI
May 2027
final compliance deadline under India's DPDP Rules, notified November 2025. Personal data sent to a foreign AI service is third-party processing the Indian company remains liable for.
MeitY, DPDP Rules 2025
WHERE AI ACTUALLY FAILS
  • Context, not capability — a generic model knows the world but not your catalogue, your policies or your customers
  • Unforecastable cost — per-token billing in a foreign currency that finance cannot budget and cannot cap
  • Unverifiable output — confident answers with no source, in domains where being wrong is expensive
  • No accountable counterparty — a foreign vendor, foreign law, and a support queue instead of a name
  • No language depth — English translated at the edge, failing the 90% of India that does not transact in it
WHAT CEHPOINT ANSWERS WITH
  • Grounding as a toggle — the model answers from the client's own knowledge base, no re-integration project
  • Fixed rupee pricing — a published price against a published request limit, invoiced with GST
  • Citations by default — live web search and URL reading return sources, not assertions
  • An Indian counterparty — Indian law, Indian invoicing, a named team, and our own security practice behind it
  • Indian languages first — a dedicated multilingual model, proven publicly on Hindi and Bengali scripture retrieval
Why this matters to an investor: the 95% failure rate is not evidence that the market is bad — it is evidence that the incumbents are selling the wrong shape of product. Every failure mode above is an integration, pricing, trust or language problem. Those are exactly the problems a domestic, full-stack, accountable provider is structurally better placed to solve than a foreign API.
05
Our Thesis
We Don't Need to Beat OpenAI.
We Need to Make Them Irrelevant.
The mistake every Indian AI startup is making: trying to build a better ChatGPT. That's a losing game — they have ₹8 lakh crore in compute and a decade of head start. The winning move is completely different: build AI so deeply Indian, so domain-specific, so integrated into Indian workflows that a foreign model becomes useless for Indian reality.
WHAT OPENAI CANNOT DO
  • Understand a Bengali farmer's crop disease described in dialect
  • Navigate India's 700+ district-level regulatory variations
  • Integrate with India's unique digital public infrastructure (UPI, Aadhaar, ONDC)
  • Provide court-admissible cybersecurity audits under Indian law
  • Deploy on-premise in Indian government buildings with data sovereignty guarantees
  • Price for Indian purchasing power without American board approval
WHAT CEHPOINT DOES
  • Build from scratch: Models trained on Indian data for Indian contexts — not fine-tunes of foreign models
  • Own the stack: Gateway → Models → Agents → Security → Infrastructure — every layer Indian-owned
  • Control distribution: Not open-source chaos — organized, guarded, responsible deployment
  • Democratize access: Free tools for everyone, powerful models for the top 100 companies, domain models for every industry
  • Provide everything: Infrastructure + models + customization + deployment + support — one accountable team
The strategic insight: OpenAI is building a universal brain. We're building 100 specialized Indian brains — one for healthcare, one for agriculture, one for manufacturing, one for every industry that matters. A universal brain speaks every language but understands none of them deeply. A specialized brain built on Indian data, for Indian problems, with Indian guardrails — that's irreplaceable.
06
Traction — Not Promises, Proof
While Others Pitch, We Ship.
2017
Founded — 9 years of building, shipping, and surviving. Not a pivot. Not a relaunch. Continuous execution.
50+
Team members across AI research, cybersecurity, enterprise engineering, and product design.
100+
Projects delivered across 12 industries — healthcare, fintech, edtech, e-commerce, government, rural.
7
Countries with registered delivery presence. India to the world.
Cehpoint AI Gateway
India's zero-friction, OpenAI-compatible AI gateway. No API keys, no cards, no metering. Tested across 32+ scenarios, 7 use cases, 4 languages. Live and serving developers across India. This is the foundation everything else runs on.
LIVE · FREE
HealthGPT — Flagship
Multi-modal clinical-intelligence AI. Medical imaging, document comprehension, FHIR-ready EHR, pre-visit automation. Deployed under hospital branding, on their infrastructure. The model for every industry-specific AI we'll build.
FLAGSHIP · DEPLOYED
Autonomous AI Agents
Named, multi-agent systems that do real work at scale — retrieval, reasoning, tool use, review. Human checkpoints on every consequential decision. The future of enterprise AI — and we're already shipping it.
LIVE
Cehpoint Code
Free terminal coding agent. No setup, no API key. Clone and run. Putting capable AI in front of every Indian developer and student — the bottom of the funnel that feeds everything above.
LIVE · FREE
Multimodal API Suite
Image generation, video generation, text-to-speech — all OpenAI-compatible. Commercial tier with SLA. Customer-requested features shipped monthly. Revenue engine #2.
LIVE · COMMERCIAL
Cybersecurity & Forensics
VAPT, SOC audits, AI red-teaming, court-admissible digital forensics. The same team that builds your AI secures it. No foreign company can offer this under Indian law.
LIVE · REVENUE
07
The Controlled Distribution Doctrine
Our AI Is Not Open Source.
And That's by Design.
THE CONTROVERSIAL POSITION
We believe — and this will make us unpopular in some circles — that open-sourcing powerful AI is irresponsible. When any model can be forked, there are no guardrails. No observability. No accountability. Models can be weaponized for deepfakes, weaponized for fraud, weaponized for radicalization. Meta's open-source strategy proved this: Llama was used for harm within weeks of release. We refuse to participate in that race.
HOW WE DISTRIBUTE INSTEAD
  • Organized distribution — Every deployment has guardrails, observability, and a human accountable
  • Purpose-built utilization — Different models for different use cases, with proper service wrapping
  • Layered access — Free tools for everyone, domain models for industries, frontier models for top 100 companies
  • Audit trail — Every interaction logged, every decision traceable, every deployment governed
  • Responsible scaling — We open-source what's safe (tools, frameworks) and control what's dangerous (models, weights)
WHY THIS IS A COMPETITIVE ADVANTAGE
  • Enterprise trust: The top 100 Indian companies will not adopt AI from a company that can't guarantee controlled deployment. We can. Meta can't.
  • Government eligibility: India's AI governance framework (coming 2026-27) will mandate traceability. Open-source models can't comply. Ours can.
  • Pricing power: Controlled distribution = value capture. Open-source = race to zero. We choose margins.
  • Liability protection: When something goes wrong, we know exactly which model, which deployment, which user. Open-source = no one's accountable.
  • Investor confidence: This is how you build a defensible, revenue-generating AI company — not a research project that gives away its value.
08
Democratization Layer
Free AI for Everyone.
Powerful AI for the Right Hands.
FREE — NO LIMITS, NO BARRIERS
Every Indian — from a street vendor in Imphal to a student in Chennai to a freelancer in Delhi — can use Cehpoint's free AI tools. No credit card. No API key. No metering. This is how we build a nation of AI-literate citizens.
  • 30+ free AI tools — idea validation, GST helpers, licence navigator, udhaar recovery, freelance pricing, SEO audit, security scan
  • CehpointGPT — free conversational AI for every Indian
  • Cehpoint Code — free coding agent for every developer
  • Reader API — free content extraction for builders
  • AI Gateway (free tier) — zero-friction AI access, no authentication
  • AI in Real Life — 3-month training program, building India's AI workforce
CONTROLLED — TOP 100 + INDUSTRY-WISE
Our most powerful models — the ones trained on proprietary Indian data, the ones that understand Indian law, Indian markets, Indian languages deeply — are distributed with purpose. To the organizations that can deploy them responsibly.
  • Top 100 Indian companies: Commercial frontier models with full SLA, custom fine-tuning, on-premise deployment
  • Industry-specific models: AgriAI for farming cooperatives, MedAI for hospital networks, IndustryAI for manufacturers
  • Government & PSU: Sovereign deployments with full data residency, audit trails, and Indian-law compliance
  • Academic institutions: Research access with responsible use agreements
  • Every deployment includes: Guardrails, observability, human checkpoints, and Cehpoint support
The economics of this model: Free tools create millions of users → users generate training data → data makes our models better → better models win enterprise contracts → enterprise revenue funds more free tools. The flywheel never stops. And unlike open-source, we capture the commercial value at every stage.
09
Commercial Strategy
India's Top 100 Companies.
Our First 100 Enterprise Clients.
India's top 100 companies — by revenue, these are the Reliances, Tatas, Adanis, Infosyses, HDFCs — collectively spend ₹5,000+ Cr annually on foreign AI and IT services. Not one rupee of that goes to an Indian AI company. We intend to change that.
BFSI — Banking & Finance
HDFC, ICICI, SBI, NABARD — AI-powered fraud detection, compliance automation, vernacular customer service. We already have FinTech delivery experience. This is our first vertical.
Healthcare Conglomerates
Apollo, Fortis, Max, Narayana Health — HealthGPT deployed across hospital networks. Multi-modal clinical AI in their branding. We've already proven this with nursing homes.
Manufacturing & Industry
Tata Steel, JSW, Mahindra — predictive maintenance, quality inspection, supply chain optimization. IndustryAI models trained on Indian manufacturing data. Zero foreign competition here.
Telecom & Tech
Jio, Airtel, TCS — autonomous agents for customer operations, network optimization, vernacular content generation at scale. Our multimodal API is built for this.
Government & Defence
Ministries, PSUs, state governments — sovereign AI deployments with full data residency. Digital India, e-governance, citizen services in 22 languages. Only an Indian company can do this.
Agriculture & Rural
IFFCO, NAFED, state cooperatives — crop intelligence, market pricing, vernacular advisory. Our rural digitalization practice gives us access no foreign company has. AgriAI is India's missing brain.
Revenue target from Top 100: ₹50Cr+ ARR by Year 3. Average enterprise deal: ₹40-50L. 100 clients × ₹40L = ₹40Cr. Plus annual maintenance, model updates, and expansion. The total addressable market for AI in India's top 100 companies alone is ₹500+ Cr annually. We're capturing the easiest 10%.
10
The Master Plan
How We Make American
and Chinese AI Irrelevant in India.
OpenAI and Google are spending ₹8 lakh crore+ on general-purpose models. They will always be smarter at trivia and coding essays. That's not the game. Here's how we win without outspending them:
1. Domain Depth > General Breadth
A model that knows everything about Indian crop diseases is more valuable to a farmer than GPT-5. A model that understands RBI compliance is more valuable to a bank than Claude. We build 100 specialized Indian brains. Each one is irreplaceable in its domain because it's trained on data that doesn't exist in Silicon Valley.
2. Integration Moat
Our AI doesn't just answer questions — it integrates with UPI, Aadhaar, ONDC, GSTN, IRDAI, FHIR, India-specific APIs. A foreign model can't do this. Every integration is a moat. Every workflow automated is a switching cost. After 100 integrations, no one switches to OpenAI because the migration would break their entire operations.
3. Regulatory Lock-in
India's DPDP Act and upcoming AI governance rules will require data residency, audit trails, and explainability. Our models are built for this from day one. Foreign models will need to rebuild. By the time they comply, we'll have 3 years of regulatory relationships and compliance certifications they can't fast-track.
4. The Infrastructure Play
We don't just sell models — we sell models + infrastructure + customization + deployment + support. The complete stack. A bank doesn't want an API key — they want a deployed, secured, maintained, audited AI system. That's what we deliver. OpenAI sells tools. We sell outcomes. And outcomes are 10× harder to switch away from.
The result: In 5 years, India's top 100 companies will run on Cehpoint AI for every domain-specific task. OpenAI will be used for generic queries — like using a Swiss Army knife for brain surgery. Our models won't compete with GPT. They'll make GPT irrelevant for Indian enterprise. That's the master plan.
11
Business Model
Six Revenue Engines.
One Compounding Platform.
35%
Enterprise AI (Top 100)
Custom AI, agents, RAG for India's largest companies. Demo-first, 40-55% margin.
25%
Commercial API & Cloud
AI API, Lifetime VPS, hosting. Recurring + usage. 65-80% margin.
20%
Controlled Model Licensing
Domain models licensed with guardrails. Industry-specific. 70%+ margin.
10%
Cybersecurity & Compliance
VAPT, AI red-teaming, forensics. Retainer + project. 50% margin.
5%
Infrastructure-as-Service
GPU sharing, deployment support, model hosting for partners. 60% margin.
5%
Training & Ecosystem
AI in Real Life, enterprise upskilling, partner program. Talent pipeline.
Blended gross margin: 45% today → 72% by Year 3 as licensing & API scale · Free tools = zero-CAC acquisition engine
12
Financial Projections
5-Year Forecast.
Conservative Assumptions. Aggressive Execution.
₹3Cr
FY26
(Current)
₹12Cr
FY27
₹38Cr
FY28
₹95Cr
FY29
₹200Cr
FY30
₹400Cr
FY31
Target
KEY DRIVERS
  • Enterprise: to 100 Top-100 clients by FY31, avg ₹45L deal (target)
  • API: to 500K commercial developers by FY31 (target)
  • Licensing: to 50 domain model deployments (target)
  • Cybersecurity: to 120 retainer clients (target)
  • Infrastructure: GPU-as-service for 200+ partners
PATH TO PROFITABILITY
  • Break-even: Month 22-26 post-investment
  • EBITDA positive: FY29 (Year 3)
  • Net margin: 20-24% by FY31
  • Cash-flow positive operations: FY30
  • Revenue per employee: ₹25L → ₹55L
13
Unit Economics
The Numbers Behind
Each Revenue Engine.
Enterprise AI (Top 100)
Avg Deal Size₹35-50L
CAC₹2-4L
LTV (3-yr)₹1.5-2.5Cr
LTV:CAC35-60:1
Gross Margin40-55%
Commercial API
Avg Rev/User/Mo₹2,500-8K
CAC (Developer)₹500-1.5K
LTV (18-mo)₹45K-1.4L
LTV:CAC30-90:1
Gross Margin65-80%
Controlled Model Licensing
License Fee₹20-80L
Annual Maintenance₹8-15L/yr
Deploy + Customization₹5-12L
LTV (5-yr)₹60L-1.8Cr
Gross Margin70-85%
14
Go-to-Market
The Flywheel That
Never Stops Spinning.
THE ENGINE
  • Free tools → 1M+ Indian users: Every student, shopkeeper, freelancer uses our free AI. Zero cost to us, massive data to us.
  • Users → Training signal: Real Indian usage patterns in 22 languages. Data no foreign company can access.
  • Signal → Domain models: We build AgriAI, MedAI, IndustryAI on real usage, not lab benchmarks.
  • Domain models → Top 100 clients: India's largest companies deploy models that actually understand Indian reality.
  • Revenue → Better infrastructure: Every enterprise rupee funds GPU clusters, better models, more free tools.
  • Repeat. Compounding. Unstoppable.
ACQUISITION CHANNELS
  • Bottom-up: Free API → developers adopt → bring to their companies (proven: Slack, Notion, GitHub all grew this way)
  • Top-down: Direct enterprise sales to Top 100. Demo-first: 24h working prototype kills long procurement.
  • Government: IndiaAI, Digital India, Smart Cities — aligned from day one. Only Indian company with full-stack sovereignty.
  • Partner channel: Professional Partner Program — resellers, agencies, integrators scale our distribution.
  • Academic: AI in Real Life creates users, talent, and enterprise leads simultaneously.
  • Global South: SE Asia, Africa — same model, same languages, same problems. India is the template.
15
Competitive Landscape
No One Else Is Building
What We're Building.
AI GatewayOwn ModelsEnterprise DeliveryCybersecurityControlled DistributionFree Tools for India22 LanguagesFull Infra
Cehpoint✓✓✓✓✓✓✓✓
Sarvam AI✗✓✗✗✗✗✓✗
Krutrim (Ola)✗✓✗✗✗✗~✗
TCS / Infosys✗~✓✓✗✗~✓
OpenAI / Google✓✓✗✗✗✗✗~
Meta (Llama)✓✓✗✗✗ (open)✗✗✗
Swipe the table to see every column
The only column that matters: Controlled Distribution. Every other player either gives away their AI (Meta, open-source community) or sells generic access (OpenAI, Google). No one — no one — is doing organized, guarded, industry-specific AI distribution for India with full infrastructure support. This is our category. We created it.
16
Technology Architecture
Five Layers. Every One Indian-Owned.
Not a Wrapper on Foreign Tech.
APPLICATION
HealthGPT · Autonomous Agents · Cehpoint Code · Free AI Tools · Industry-Specific Assistants — Products that solve real problems for real Indian users
DISTRIBUTION
Controlled Deployment · Guardrails Engine · Observability Dashboard · Human Checkpoint System · Audit Trail — The layer that makes us different from every open-source project
MODEL
Cehpoint Research Models · AgriAI · MedAI · IndustryAI · BhashaAI · RAG Pipelines · Evaluation Harnesses — Domain-specific Indian models, 22 languages, trained on Indian data
PLATFORM
AI Gateway (OpenAI-compat) · Multimodal API (Chat·Image·Video·TTS) · Agent Orchestration · Smart Router — Unified API layer, cost-optimized, zero-friction for developers
INFRA
GPU Clusters · Multi-Region Cloud · Edge Deployment · Lifetime VPS · OpenClaw · Data Sovereignty Layer — Indian-owned, Indian-controlled, Indian-compliant
IP advantage: Every layer owned by Cehpoint. The Distribution Layer is the moat — it's what makes our AI safe for enterprises and governments. OpenAI has no equivalent. Meta has the opposite. This is the layer that makes Indian enterprise adoption possible.
17
Investment Architecture
Three Ways In.
They Do Not Overlap.
Most decks offer one instrument and hope it fits everyone. Cehpoint is at a specific moment — a profitable operating business, incorporating as a private limited company — and that creates three genuinely different entry points. Read the one that matches your mandate; they are deliberately kept separate so nobody is sold the wrong thing.
Track 1 · Partner
No capital at risk
  • • Buy a commercial ORG licence
  • • Issue sub-licences to your startups, students or clients
  • • Revenue split on what you resell
  • • Live and built today
  • • For incubators, accelerators, universities, agencies
Track 2 · Profit share
₹5L – ₹50L
  • • 1% of profit at ₹5,00,000
  • • Scaling to 5% at ₹50,00,000+
  • • No equity, no dilution
  • • Only 10 slots, ever
  • • Open right now
Track 3 · Equity
On incorporation
  • • Real shareholding in the private limited company
  • • Board seat or observer by ticket size
  • • Standard instruments and diligence
  • • For institutional and serious strategic investors
  • • The mission round
Why there are three, stated plainly: Cehpoint trades today as a proprietorship, which is why profit-sharing is what we can legally offer this minute — and we are not going to pretend otherwise in a deck. The private limited company is being incorporated, and equity is issued from it. If you are here to own part of an Indian AI company, Track 3 is your track and the raise on the final slide is what funds that incorporation. If you are here to build a business on top of our infrastructure, Track 1 pays you without you writing us a cheque at all.
18
Track 1 · Partner & Sub-licensing
Our Customers Fund Our Infrastructure.
Then We Split What They Earn.
This is the part most investors miss. An incubator does not need to donate to Indian AI — it can buy capacity from us and resell it to the startups in its own cohort. Their subscription funds our GPU and engineering spend; their resale earns them margin. Capital goes in as revenue, not as a cheque, and both sides are paid. The hierarchy, sub-licence issuance, per-key quotas and usage roll-up are already built and running in production.
Partner typeThey issue sub-licences toWhat it does for themWhat it does for us
IncubatorTheir startupsEvery portfolio company gets working AI on day one, under the incubator's own brandDozens of live deployments from one relationship
AcceleratorTheir cohortA hard differentiator when recruiting founders for the next batchA new cohort of users every programme cycle
UniversityStudentsAI access for an entire campus without buying foreign seats in foreign currencyThe next generation learns on our stack, not OpenAI's
Agency / resellerTheir clientsSells AI to its client base without building any of itDistribution into SMEs we would never reach directly
EnterpriseBusiness unitsOne contract, governed keys per department, usage visible centrallyExpansion inside the account without a new sale each time
Swipe the table to see every column
ALREADY BUILT, NOT PLANNED
  • Parent-and-child licence hierarchy with a per-partner sub-licence ceiling
  • Each sub-licence gets its own API key, its own quota and its own grounding data
  • Usage rolls up to the partner: requests, keys and grounded projects across every child
  • The partner suspends or restores any sub-licence themselves — no ticket to us
WHY THIS COMPOUNDS
  • Zero acquisition cost per end user — the partner recruits them, not our sales team
  • Switching cost is theirs, not ours — the partner's own customers now sit on our keys
  • Revenue before capital — every partner is cash-positive for us from the first invoice
  • Structurally un-copyable — a foreign API cannot contract with an Indian incubator on these terms
19
Track 2 & Track 3 · For Investors
Profit Share Today.
Equity On Incorporation.
These two are kept strictly apart, because conflating them is how investors get misled. One is live and pays from operating profit. The other is ownership of the company we are incorporating, and it is the reason we are raising at all.
TRACK 2 — PROFIT SHARE · OPEN NOW
₹5L
Minimum ticket — and the entry point for a 1% profit share
5%
Profit share at ₹50,00,000, scaling linearly from 1% between the two
10
Total slots. When they are committed, this track closes permanently
  • No equity and no dilution — you are buying a share of profit, not of the company
  • Paid from an operating business, not from a projection — we have revenue today
  • Legally clean under a proprietorship, which is exactly why it is available this minute
  • Live in the investor portal — terms, quote and commitment are all self-serve
TRACK 3 — EQUITY · ON INCORPORATION
Pvt Ltd
Being incorporated. Equity is issued from that entity, not from the proprietorship
₹25–75 Cr
Pre-revenue valuation band for the seed round now open
  • Genuine shareholding with the rights that come with it — this is not a revenue-share dressed up as equity
  • Board seat or observer depending on ticket size, with monthly metrics
  • Standard instruments and full diligence — priced equity or SAFE, your counsel and ours
  • For the investor backing a mission, not renting a return: sovereign Indian AI infrastructure held by Indian shareholders
If you take one thing from this slide: the profit-share track is a good return on a business that already works. The equity track is a position in the company that intends to own India's AI layer. They are priced, papered and governed differently on purpose — and you are welcome to take either, or to start on Track 1 and never write us a cheque at all.
20
Participation · Beyond the Cheque
Capital Is One Contribution.
It Is Not the Only One.
India's AI opportunity will not be decided in Bengaluru boardrooms alone. It will be decided by whether the technology actually reaches a textile cluster in Tiruppur, a district hospital in Bihar, a co-operative bank in Maharashtra and a college in the North-East. That is a distribution problem, and distribution is built by people with local standing — not by an API. If you have reach, domain depth or credibility in a part of India, there is a defined way to put it to work here.
Regional Distribution Partners
Take a state or a cluster and become the face of Cehpoint in it — onboarding local businesses, running demonstrations, holding the relationship. Revenue-shared, not salaried. Best fit: established regional IT firms, MSME associations, chambers of commerce.
Domain Expertise
A domain model is only as good as the practitioner who corrects it. Doctors, chartered accountants, advocates, agronomists and manufacturing engineers shape and validate the vertical models in their own field. Contribution: judgement, not capital.
Language & Data Custodians
India's 22 scheduled languages need speakers who will curate, correct and defend how their language is represented. Universities, publishers and cultural institutions hold that material and that authority. Contribution: corpora, correction, legitimacy.
Deployment & Integration Partners
Systems integrators and IT service firms that already sit inside enterprise and government accounts, deploying Cehpoint under their own delivery. Contribution: existing trust inside accounts we would otherwise take years to enter.
Talent & Training
Colleges and training institutions that run the AI-in-real-life programme locally, turning free tools into an employable skill and a pipeline of people who already know our stack. Contribution: the workforce this industry does not yet have.
Policy & Institutional Access
People who can open the right door — state IT departments, PSUs, regulators, IndiaAI Mission programmes. In a market where the largest buyer is the state, access is a form of capital. Contribution: introductions and standing.
The structural point: a foreign AI provider can match our model. It cannot match a distribution network of people who are personally accountable in their own district. That network is the part of this business that cannot be copied from San Francisco — and it is built one contributor at a time, which is why every one of these paths is open.
21
Use of Funds
Building the First Structure.
Research Lab & Models — 40%
Cehpoint Research Labs: 15+ researchers. Domain models: AgriAI, MedAI, IndustryAI, BhashaAI. Training on Indian data. Open publications for reputation. Proprietary models for commercial distribution. The brain factory.
Infrastructure & Gateway — 25%
GPU clusters for training + inference. 10× gateway capacity. India's most reliable OpenAI-compatible API. Edge deployment. Multi-region redundancy. The backbone everything runs on.
Controlled Distribution Engine — 20%
Build the guardrails, observability, and deployment automation that makes enterprise AI safe. Agent platform at scale. Multimodal: vision, voice, video. Commercial API with SLA. The moat.
Team & GTM — 15%
Chief AI Scientist, Head of Research, 5+ ML engineers. Enterprise sales lead. Mumbai + Bengaluru offices. Scale partner program. AI in Real Life → 10 cities. Government partnerships team.
22
Research Initiative
Cehpoint Research Labs:
India's Problems. India's Models.
AgriAI
Crop disease detection, yield prediction, market pricing, soil analysis — trained on 10 years of Indian agricultural data. For 140M+ Indian farmers.
MedAI
Clinical diagnostics for Indian disease patterns, AYUSH integration, vernacular medical records, multi-modal imaging. For 1.4B patients.
IndustryAI
Predictive maintenance, quality inspection, supply chain optimization for Indian manufacturing standards. For India's ₹41 lakh crore manufacturing sector.
BhashaAI
Native understanding of all 22 scheduled languages — not translation, comprehension. Dialects, code-switching, cultural context. For 1.4B voices.
22
Languages in 3 years
8+
Domain foundation models
Open
Research publications
₹500Cr+
TAM per domain model
The infrastructure promise: For every domain model, we provide the complete package — infrastructure, model customization, deployment, guardrails, observability, and ongoing support. Clients don't just get a model. They get a deployed, secured, maintained AI system. We take responsibility for everything.
23
Risk Analysis
We Know Every Risk.
Here's How We Neutralize Each One.
GPU Access & Compute Cost
✓ Mitigation
Diversified: Nvidia + AMD + Cerebras. IndiaAI Mission GPU sharing. UAE G42 partnership (already in India). Smart routing across models = 60% cost reduction vs direct API. Our Gateway's intelligent load balancing means we never overpay for compute.
Foreign Model Dependency (Short-term)
✓ Mitigation
Gateway routes across all providers — zero lock-in. Open-source fallback (Llama, Mistral) for non-critical tasks. Every enterprise rupee funds own model training. Path to full model independence: 24-36 months. Even then, we route smartly — own models for India, foreign for generic.
Talent War with Big Tech
✓ Mitigation
Built in West Bengal — 40% lower cost than Bengaluru. Remote-first research roles across India. ESOP pool for early team. Mission narrative attracts purpose-driven engineers who don't want to optimize ad clicks at Google. We offer meaning.
Enterprise Sales Cycles
✓ Mitigation
Demo-first model: 24h working prototype bypasses 6-month procurement. Government contracts are longer but 10× larger. Free tools create bottom-up pull from developers inside enterprises. Top 100 strategy focuses on CTOs who can move fast.
Regulatory Shifts (DPDP, AI Governance)
✓ Mitigation
Compliance-first from day one — this is a feature, not a cost. Data localization = our advantage (foreign models can't comply). Cybersecurity practice ensures audit readiness. We're aligned with government, not fighting it. When regulation comes, we're already there.
US AI Companies Price-Dumping in India
✓ Mitigation
Our free tier already matches zero-price. Enterprise buys us for sovereignty, compliance, language, domain expertise, and on-premise deployment — none of which price can fix. And when US firms cut prices, it validates the market. We grow with the wave.
24
Proof of Impact
Real Problems. Real Results.
Not Hypotheticals.
Healthcare: Hospital Network, West Bengal
Deployed HealthGPT across 3 nursing homes. Multi-modal AI handles patient intake, medical record comprehension in Bengali and Hindi, pre-visit automation, and clinical document summarization.
40%
Admin time reduced
3×
Faster patient intake
22
Languages supported
FinTech: Regional Payments Platform
AI-powered fraud detection and RBI compliance automation. Reduced false positives while catching more real fraud. PCI-DSS compliant. Integrated with UPI verification systems.
65%
Fewer false positives
₹2Cr+
Fraud prevented/yr
PCI
DSS compliant
EdTech: Learning Platform, Tier-2 Cities
AI-powered LMS with personalised learning paths in 5 Indian languages, automated assessment, and vernacular content generation. Serving students who've never had AI-powered education before.
10K+
Students served
5
Indian languages
85%
Completion rate
Cybersecurity: Mid-Market Enterprise
Comprehensive VAPT + remediation. Discovered critical vulnerabilities that automated foreign tools missed — because we understand Indian network configurations and compliance requirements they don't.
47
Vulnerabilities found
12
Critical severity
100%
Remediated 30d
25
Returns & Exit · Track 3 (Equity)
How You Make Money.
Multiple Paths. One Direction: Up.
Everything on this slide describes the equity track and assumes the private limited company. Track 2 works differently and more simply: a profit share of 1% to 5% is paid out of operating profit as it is earned, with no exit event required and nothing to wait for.
EXIT SCENARIOS
  • IPO (Primary, Year 5-7): India's AI market will create IPO candidates. Cehpoint positioned as the infrastructure play — like India's answer to Palantir + Snowflake + CrowdStrike combined.
  • Strategic acquisition: Global tech entering India needs an indigenous partner with full-stack sovereignty. That's us. Premium acquisition target at ₹5,000Cr+.
  • Secondary sale: At ₹100Cr+ ARR, secondary market becomes liquid. Early investors can partially exit at 10-20× while maintaining upside.
  • Dividend model: Profitable platform business can distribute while growing. Patient capital gets paid while waiting.
RETURN MATH (SEED INVESTOR)
Seed (₹25Cr val)₹1Cr → 4%
Series A (₹150Cr)₹1Cr → ₹3Cr (3×)
Series B (₹600Cr)₹1Cr → ₹12-16Cr (14×)
Pre-IPO (₹2,000Cr)₹1Cr → ₹50-80Cr (60×)
26
Investor FAQ
Questions You're Thinking.
Answers That Should Give You Confidence.
"Why not just open-source everything like Meta?"
Because open-source AI gets weaponized within weeks — deepfakes, fraud, radicalization. Meta proved this with Llama. More importantly: open-source races to zero on pricing. We build a revenue-generating, defensible business. Investors don't get returns from research papers. They get returns from controlled distribution of valuable AI.
"Can you really compete with ₹8 lakh crore OpenAI?"
We're not competing. We're making them irrelevant for Indian use cases. A model that understands RBI compliance, UPI integration, 22 languages, and Indian manufacturing standards doesn't need to beat GPT at writing poetry. It needs to solve problems GPT can't even see. That's a completely different game.
"What's the real moat?"
Four layers: (1) Indian training data that doesn't exist in Silicon Valley, (2) integration with Indian digital infrastructure (UPI, Aadhaar, ONDC), (3) regulatory compliance built-in from day one, (4) 9 years of enterprise relationships across 12 industries. Each layer compounds. None can be replicated quickly.
"Isn't Birbhum a disadvantage for hiring?"
40% lower burn rate = longer runway. Remote-first means we hire from all of India. Our rural digitalization practice is a product differentiator — we understand non-metro India because we ARE non-metro India. Investors in Zoho (Tenkasi) and Freshworks (early Chennai) learned the same lesson: talent is everywhere, opportunity isn't.
"What if the government changes AI policy?"
Every policy direction benefits us. Data localization mandates = our advantage. AI sovereignty push = our thesis. Procurement preferences for Indian companies = our pipeline. We're aligned with every likely policy outcome. The only risk is if India abandons AI sovereignty entirely — and given the geopolitical climate, that's the least likely scenario.
"How is this different from Sarvam or Krutrim?"
They build models. We build models + products + enterprise delivery + cybersecurity + free user acquisition + controlled distribution. They need partners to reach enterprises. We ARE the partner. They burn cash on compute. We generate revenue from day one. They're research labs. We're a company.
"What's your IP strategy?"
All models, code, and products wholly owned. Enterprise clients get usage rights with guardrails, not IP transfer. Research published openly for reputation; core models and training data proprietary. Client data never used without consent. This is the model that makes investors money — not open-source goodwill.
"Cap table?"
Founder-promoted. Clean. No complex prior rounds. First institutional capital = first-in institutional pricing. Full disclosure in data room. We've kept it simple because we want the right partners, not the most money.
27
The 100-Year Vision
We're Not Building a Startup.
We're Building India's AI Institution.
2026–2028
Foundation. Research lab operational. First 4 domain models. 10K+ API devs. 50 Top-100 enterprise clients. Controlled distribution engine live. Seed + Series A.
2028–2032
Scale. Own GPU cluster. India-trained foundation models. 1M+ developers. Govt contracts across 5 ministries. SE Asia + Africa expansion. Series B.
2032–2040
Sovereignty. India's default AI platform. Full-stack independence. 500M+ daily users. IPO. 200+ researchers. The company India turns to first.
2040–2060
Global South. AI for 100+ developing nations. The "non-aligned" AI platform — neither US nor China. Digital public infrastructure for the world.
2060–2126
Legacy. 100-year institution. Cehpoint Research = India's Bell Labs. AI education for every Indian child. The company that made India an AI superpower.
"This isn't a company with an exit plan. This is an institution being built to outlast its founders. Every investment today is a stake in India's technological sovereignty for the next century. We are building the company that will make India an AI superpower — not by catching up to the West, but by building what the West cannot: AI that understands a billion Indians."
Impact
This Is Bigger Than a Company.
This Is India's AI Future.
1B+
Indians Served
AI that speaks their language, understands their reality — built in India, for India, by India
10M+
Jobs Created
Direct + partner ecosystem + downstream AI-native Indian businesses across every industry
100+
Nations Empowered
Global South AI platform — the alternative to US-China AI duopoly for developing world
We will build AI so deeply Indian that foreign models become irrelevant for Indian problems.
We will distribute it responsibly — controlled, guarded, accountable — not open-source chaos.
We will serve everyone — free tools for the student, powerful models for the top 100, domain AI for every industry.
We will provide everything — infrastructure, customization, deployment, support — one team, full responsibility.
And we will do it all from India, for India, starting now.
28
The Ask
We've Built the Engine.
We Need Fuel.
Raising: ₹25 Cr – ₹75 Cr (Seed + Bridge) · equity, from the private limited company being incorporated
  • ₹25 Cr: Research lab, 15 researchers, first 4 domain models, gateway infrastructure, controlled distribution engine v1
  • ₹50 Cr: + GPU cluster, commercial API launch, Mumbai/Bengaluru offices, enterprise sales team
  • ₹75 Cr: + SE Asia expansion, government partnerships, accelerate to 100 Top-100 clients
Also welcoming non-dilutive capital: government grants (IndiaAI, MEITY), research funding, CSR partnerships, problem-backed corporate investments. Every rupee has a structured path to impact.
What You Get — By Track
  • Track 3 (equity): real shareholding in the private limited company this raise incorporates — the only Indian company building the complete sovereign AI stack
  • Track 3: board or observer participation by ticket size, monthly metrics, pro-rata rights in subsequent rounds
  • Track 2 (profit share): 1% to 5% of operating profit for ₹5L to ₹50L, no dilution, only 10 slots — open today without waiting for incorporation
  • Track 1 (partner): resale margin on every sub-licence you issue, with no cheque written to us at all
  • Access to India's Top 100 enterprise pipeline and government relationships
  • Named recognition in research outputs and product launches
  • A stake in India's AI sovereignty — the defining technology opportunity of our generation
Cehpoint is not asking for permission to build India's AI future. We're already building it. We're inviting you to be part of it.
C E H P O I N T
Let's Build India's AI
Together.
AI-first. India-first. World-class.
Email
contact@cehpoint.co.in
Phone
+91-90911-56095
Website
cehpoint.co.in
AI Platform
ai-api.cehpoint.co.in
HQ: Bolpur, West Bengal, India · Delivery: India, UK, US
Confidential · August 2026
1 / 31