Pre-seed  ·  India

Market research is dead. Market intelligence has arrived.

JanPulse turns live communities — students, workers, consumers — into continuous, hyperlocal intelligence. AI-native conversations in people's own language, anchored to a living profile, fused with validated behavioural frameworks.

Market Research + Our AI = Market Intelligence
Policy Expertise + Our AI = Policy Intelligence

The raise — $200K pre-seed Read the measured evidence See the technology

Founders

Ashwin Srivastava

Ashwin Srivastava

Co-Founder, JanPulse
Founder & CEO, Sapio Analytics — 11 million+ low-income lives transformed. Creator of a sovereign Livelihood AI model proven at nation scale. Patents, IP and people-management frameworks used by hundreds of thousands of leaders.
IIT BombayMumbaiCo-author, The AI Native State LinkedIn →
Prashant Nikam

Prashant Nikam

Co-Founder, JanPulse
Civil services officer with direct exposure to population-scale data. Author and creator of consciousness frameworks that read human beings scientifically — the layer beneath demographics.
MBA CranfieldMSc LSECo-author, The AI Native State LinkedIn →

Built on the Livelihood Intelligence experience of Sapio Analytics.

$200K
Pre-seed round
2
Use cases in discussion
~$40B
Addressable market
4
Revenue lines
The AI Native State, by Prashant Nikam and Ashwin Srivastava
New from the founders · 2026

The AI Native State

Rebuilding Government for the Age of Intelligent Machines

Prashant Nikam & Ashwin Srivastava

AI will reshape every government on earth. The only question is whether it rebuilds them, or just speeds up what is already broken. Modern governments hold more data, technology and expertise than ever, yet citizens remain invisible to the systems meant to serve them — and policies fail despite good intentions. The book argues for institutions that can see, remember and learn.

  • Written by a former Indian civil servant of 23 years and the creator of a sovereign Livelihood AI model proven at national scale.
  • The thesis JanPulse is built on: a state that cannot hear its citizens cannot serve them, whatever it spends on technology.
Read or buy the book → Amazon.in

Paperback ₹590 · Hardcover ₹750 · Notion Press, Amazon.in and Flipkart.

The category shift

Market Research is dead. Market Intelligence has arrived.

Research was an event that ended in a PDF. Intelligence is a product that compounds.

Yesterday — Market Research

  • One-off surveys, stale by publication
  • Aggregated national and district averages
  • Static questionnaires, fixed samples
  • Human panels, high cost per insight
  • Reports die in PDF. Findings don't compound.

Today — AI-Native Market Intelligence

  • Continuous AI conversations — an always-on pulse
  • Hyperlocal, down to ward, campus and cohort
  • Adaptive dialogue that builds rich profiles
  • Every response feeds a compounding model
  • Insight becomes a product that compounds, not a report.

We are building the category. Pulse on people, multiplied by AI.

How it works

The AI-native intelligence loop

01

Engage

AI-native conversations with real people — students, workers, consumers — in their language, at their moment.

02

Profile

Every dialogue enriches a living profile: livelihood, aspiration, behaviour, spend, sentiment, context.

03

Model

Profiles flow into models that fuse pulse-data with validated research frameworks and prior intelligence.

04

Deliver

Hyperlocal, contextual intelligence — piped as briefs, dashboards and API into partner products.

The loop compounds: every conversation makes every future insight sharper. This is our moat.

The moat

Base Intelligence × Custom Research = Enhanced Data

Pulse on People

Continuous AI-native engagement across live communities — students, workers, consumers.

Validated Frameworks

Livelihood, consumer and policy frameworks baked in from Sapio's institutional work.

Base Intelligence Layer

Every past conversation, every profile, every model output — reusable across new studies.

Enhanced intelligence

Any new study starts smarter

Faster

Prior profiles pre-answer half the study before it starts.

Cheaper

AI-native capture beats field-panel unit economics 5–10×.

Deeper

Frameworks + base data lift raw responses into structured intelligence.

Defensible

Competitors can copy the survey; they can't copy the compounding profile graph.

Our foundations

Making the invisible visible. Making the inaudible heard.

Every organisation has an invisible population — informal workers, non-customers, employees whose real concerns never reach leadership. JanPulse builds on population-intelligence methodologies developed for public-sector initiatives in Maharashtra and Uttar Pradesh, designed to discover the realities conventional systems miss.

Public intelligence · Livelihood

Mumbai Suburban Skill Census

Maharashtra

Administrative records capture education, occupation and residence. They rarely capture aspiration, the barriers in front of it, nearby employer demand, or the specific pathway that would change a livelihood. Treated as a one-off report, that gap is invisible; multiplied across thousands of households, it becomes an institutional intelligence problem.

The method: adaptive questionnaires, multimodal listening, contextual inference, response-quality signals, and continuity after the census closed.

Economic intelligence · Informal economy

Uttar Pradesh Unorganised Economy Census

Lucknow · Gorakhpur · Meerut · Varanasi

Establishment-centric statistics miss household trading, home-based production, informal services and seasonal work. Two people recorded under the same occupation can face entirely different customer networks, costs, supply chains and credit needs.

The method: household-first discovery, hyperlocal context, rich economic profiles, derived estimates with stated uncertainty, permissioned access.

Read the two case studies → The book behind the method

Together they form the methodological pedigree for JanPulse: discovering missing voices, combining them with context, stating confidence, and turning signals into decisions.

Technology moat

Profile + Response. That is the whole trick.

Every data point we collect is anchored to a rich, living profile — so raw responses become predictive intelligence, not noise.

Demographics+ Location+ Behaviour+ Emotional state AI insight = profile + response
01

Implicit signal detection

Sentiment, hedging language and response timing read against validated valence–arousal models — capturing signal respondents don't state outright. This is the layer that is measured, published and open to inspection.

The measurement →

02

Behavioural science frameworks

Institutionally validated frameworks that structure every raw response into predictive behavioural signal.

03

Deep-tech statistical engine

Statistical machinery that fuses profile and response into insights that hold up in real policy, market and product decisions.

The unique marriage of behavioural and affective science with deep-tech statistical algorithms — paired with the pedigree of reaching billions across the developing world.

Evidence

The signal layer, measured — including where it fails

Most sentiment tooling returns a label with no way to check it. Ours returns numbers on a named psychological model (Russell's Circumplex), a second deterministic method as a control, and blind human ratings on the same 30 responses — scored by two independent raters.

Axis Model r 95% CI n Human ceiling Reading
Valence +0.96 [+0.92, +0.98] 30 +0.88 At the ceiling. The raters agree strongly, and the model matches what they can agree on.
Arousal +0.49 [+0.16, +0.72] 30 +0.71 Below the ceiling — this is model work, not ambiguity in the measure.
Hedging +0.81 [+0.64, +0.91] 30 +0.31 The raters agree only weakly on this axis, so the model's score is not claimed as validated. The rubric is being tightened and the axis re-rated.
Lexicon control +0.17 [-0.20, +0.50] 30 — A cheap deterministic word-lookup does not work here: not distinguishable from zero, and it contradicted the sign of the human mean on 14 of the 30 responses.

What the ceiling means. Two humans rated the same responses independently. How well they agree with each other is the ceiling: a model cannot be judged against an axis the raters themselves score differently. Valence 0.88 · arousal 0.71 · hedging 0.31.

What we are not claiming. n = 30 with two raters is an indicative measurement, not proof. A second pair of raters would tighten each ceiling. The LLM scorer is not fully deterministic, moving correlations by roughly 0.05 run to run. The lexicon control is licensed for research use and is not part of any shipped product.

Full measurement page → Rate the responses yourself →

The rating sheet at /rate.html is the same blind instrument the raters used: no model scores are shown, your name is required, blanks stay blank, and the ratings export as CSV that the harness re-runs against.

Never before possible

Decisions built on real pulse — not assumption, not hallucination

Micro-finance & lending

They don't know if a product will sell in a location — or what will make it sell.

JanPulse reveals what can sell in each locality, and how the consciousness and behaviour of the people shape what actually works.

Government

No tender, plan or policy should be designed without access to sovereign AI grounded in the citizens it will serve.

A foundational model reflecting unique socio-economic, cultural and linguistic profiling — for consultants, bureaucrats and tender design.

Product companies

Real surveys cannot be replaced by models that hallucinate and simulate.

JanPulse doesn't hallucinate: it gets into the mind of the respondent, knows their pulse, and responds on their behalf from continuous ground data.

Two use cases in discussion

Use case 01 · demo-ready

Education intelligence at scale

An educational institute with hundreds of thousands of students, turned into a live intelligence engine — ready to be initiated post funding.

  1. Deploy at scale — 100K+ students onboarded across campuses and cohorts.
  2. Train students in AI-native inquiry — they become skilled contributors of structured insight.
  3. Insights → intelligence — continuous conversations profile learning gaps, aspirations, spending and family context.
  4. Two intelligence products — for the institute, how to make each student succeed; for parents and partners, what to buy to unlock more student value.

Revenue: institute subscription + parent-facing intelligence product + brand data-licensing (edtech, publishing, D2C, financial services).

Use case 02 · demo-ready

Outbound India: language learners as an intelligence panel

A language training institute — students learning new languages to work abroad — becomes a live, contextual research panel. Ready to be initiated post funding.

  1. Language learners onboarded — high-intent, high-context, high-value cohort.
  2. Contextualised research — destination country, family, savings, remittance intent, product needs.
  3. Intelligence generated — behavioural, financial and cultural profiles built continuously across every learner.
  4. Products designed — consumer products, financing and remittance offerings designed on live evidence.

Buyers: remittance platforms · NBFCs & lenders · forex and travel cards · insurance · D2C brands · recruiters.

The pattern replicates

Pipeline

Gig & delivery workforce

Earnings volatility, credit appetite, brand affinity — for lenders, insurers, employers.

Pipeline

Rural & tier-2/3 consumers

Livelihood pulse for FMCG, agri-inputs, financial inclusion and welfare design.

Pipeline

Health & wellness cohorts

Patient panels for pharma, diagnostics and public-health institutions.

Pipeline

MSME operators

Real-time sentiment and demand signals for policy and B2B suppliers.

Both are demo-ready and sit on the same rails: every new deployment enriches the base intelligence.

Market size

A ~$40B addressable market. Two doors. One platform.

Mass-context market research across the developing world, plus ~1% of people-impacting government policy budgets.

Door 01

~$18B

Developing-world market research

Mass-context research spend across India, SE Asia, Africa and LATAM.

Buyers: consumer brands, financial services, edtech, D2C, pharma, agri.

Door 02

~$22B

Government policy-decision spend

Roughly 1% of major people-impacting government policy budgets across the developing world.

Buyers: government consultants, bureaucracies, tender designers, advisories.

TAM figures are the founders' estimates of addressable spend today, not bookings.

How we capture it, step by step

Starts month 1–2

Custom research + AI-powered ground reports

Client-briefed studies executed on top of the model — ground reports delivered from the field within weeks. Immediate revenue, immediate data deposits into the base intelligence layer.

Starts month 4

Continuous intelligence — AI tokens

A sovereign AI model grounded in citizen pulse, queried by researchers, consultants, brands and governments as a paid service. All four revenue lines go live here.

Step 1 funds the build. Step 2 compounds it into a category-defining intelligence layer.

Revenue model

Four revenue lines. One compounding data asset.

01

Deployment fees

₹15L–₹50L per engagement

Standing up JanPulse inside a partner community — institute, workforce, cohort. Covers integration, training and first intelligence delivery.

02

Intelligence subscriptions

Annual contracts

Recurring dashboards and briefs for the deploying partner and its stakeholder ecosystem.

03

Custom research, enhanced

Per study, premium

Client-briefed studies executed on top of base intelligence — priced higher because prior data pre-answers half the study.

04

Data & API licensing

Usage-based

Anonymised intelligence licensed to consumer brands, financial services, remittance, insurance and policy buyers.

Every rupee of revenue also deposits data into the base intelligence layer. Revenue and moat grow together.

Why now

Three shifts make this buildable today — and only today

AI-native engagement is finally natural

LLMs make continuous, multilingual, contextual conversation cheap. Gathering insight stops feeling like a survey and starts feeling like a chat.

Communities are digital by default

Institutes, training academies, gig platforms and workforces all run on mobile. Deployment inside a community is a distribution problem we can solve.

Buyers are ready to pay for evidence

Brands, lenders, remittance players, edtech and policy institutions are done with stale panels. They want live, hyperlocal, defensible intelligence.

Why us

Two founders who have already built what others are theorising

Ashwin Srivastava
Co-founder

Ashwin Srivastava

Sovereign AI · Behavioural science · Patents & frameworks

  • Creator of the world's first sovereign Livelihood AI model, proven at nation scale.
  • Founder & CEO of Sapio Analytics — 11 million+ low-income lives transformed.
  • Creator of a people-management framework used by hundreds of thousands of leaders across enterprises.
  • Developer of multiple patents and IP; author and designer of frameworks at the intersection of behavioural science and AI.
  • Co-author, with Prashant Nikam, of The AI Native State (2026).

IIT BombayMumbai LinkedIn →

Prashant Nikam
Co-founder

Prashant Nikam

Civil services · Consciousness science · Population-scale data

  • Government civil-services officer with direct exposure to billions of data points and a working understanding of large-scale populations.
  • Author and creator of consciousness frameworks that read human beings scientifically — the layer beneath demographics.
  • The bridge between sovereign data infrastructure and hyperlocal citizen intelligence.
  • Co-author, with Ashwin Srivastava, of The AI Native State (2026) — rebuilding government for the age of intelligent machines.

LinkedIn →

Sovereign data pedigree + consciousness science + behavioural AI — the exact stack JanPulse needs. Built by them before.

The ask

Raising $200K pre-seed to convert two demo-ready deployments into a category

A two-track strategy

Track 1 — 12 months to move JanPulse from two demo deployments to a piloted, revenue-generating AI-native market intelligence platform, using services revenue that starts in month 1 alongside the pre-seed.

Track 2 — once the AI model is initiated by month 3, an additional raise is planned that does not disturb Track 1, with funds generated by month 6 for a stronger foundational model.

Use of funds
Platform & AI build — conversational engine, profile graph, framework layer35%
Deployments — taking the two demo use cases into live delivery25%
Content & category — public voice, category creation, buyer education20%
Team & operations — founding hires across engineering and analysis15%
Legal, compliance & buffer — data governance and privacy foundations5%

We are a pre-revenue, pre-product company with access to the raw materials and servicing capabilities required to build the product and generate revenue, with the pedigree of products and businesses built by the founders. We invite partners who can build this into a $10B company* in the next ten years.
*The $10B roadmap is in progress and will be shared on request.

Pulse on people. Multiplied by AI. Delivered as intelligence.

A new form of market and policy research — backed by base intelligence, validated frameworks, consciousness algorithms, and a living pulse on people.

Email the founders Book a conversation Read the measurement

Ashwin Srivastava — Founder & CEO, Sapio Analytics  ·  Prashant Nikam — Co-Founder, JanPulse