JUDGMENTDIGITIZED.— Adversarial AI due diligence and decision support for venture and MSME evaluation
TRUTH BEFORE CAPITAL.
THREE SYSTEMS. ONE IDEA.
Big decisions still get made on instinct and a spreadsheet. We build engineering that reads the evidence, scores it the same way every time, and shows its working. One system for private capital, two for the public sector.
ZURVEK
LiveInvestment decision engine
You upload a deck, financials and a cap table. Five AI agents argue over the same company from five different angles. You get one score out of 100 in under 180 seconds, and every number traces back to the page it came from.
For investors, VC funds, family offices, accelerators and founders.
zurvek.com →CHETRA
In developmentPublic service performance OS
One live record per institution — building, staff, stock, vendors, spend. A fixed 0–100 condition score ranks what to fix first, and the fix is costed against money the department already has.
For education and health departments running government schools and hospitals.
chetra.tech →NIRNAE
In developmentForensic money-trail engine
A department sends the expenditure data it already holds. Nine detection agents return scored findings, the entities behind them, and the arithmetic behind every flag. Indicators for verification, not legal conclusions.
For ministries, treasuries, audit directorates, vigilance cells, PSUs and public sector banks.
nirnae.tech →INTELLIGENCE THAT
OUTPERFORMS INSTINCT.
Our flagship system, Zurvek, reads a pitch deck, the financials and the cap table. Five AI agents argue over the same company from five different angles. You get one score out of 100 in under 180 seconds, and every number points back to the page it came from.
VentureSense Technologies LLP is an Indian deeptech company founded in September 2025 and based in Noida, Gautam Buddha Nagar, Uttar Pradesh (LLPIN: ACR-5230), recognized by DPIIT Startup India and startinUP. We build three systems — Zurvek for investors, CHETRA and NIRNAE for the public sector. We are not affiliated with venturesenseai.com or any US-based entity using a similar name.
Data scales. Judgment does not. We close the gap.
Our flagship product Zurvek Engine is live at zurvek.com
Data Scales 10X.
Deals surfaced per year
Judgment Stays 1X.
Flatline — human review capacity
The Gap Is Closed.
Adversarial scoring, every deal
WHAT ZURVEK DOES, IN ONE SCREEN
Zurvek is an AI decision engine for unlisted companies. It reads a deal the way a diligence team would, and returns one score out of 100 with the reasoning attached.
You upload
A pitch deck, the financials or MIS, and a cap table. Add the CIN or company name and Zurvek pulls the Indian registry record itself. Nothing else to prepare — no data-room integration, no schema, no onboarding call.
Zurvek interrogates
The deck is broken into individual claims. Each claim is checked against the financials and outside sources, and whatever survives that is handed to five AI agents who argue over it from five different angles.
You get back
One score from 0 to 100, the ten risk areas scored separately, red flags with the page they came from, a geopolitical exposure read, and a memo you can export and take into committee.
What you can send
- Pitch deck (PDF / PPTX)
- Financials (XLSX / PDF)
- Cap table
- MCA21 / CIN
- GST
- Bank statements
THE ADVERSARIAL PROTOCOL
Five AI agents look at the same company and disagree on purpose. Most tools summarise a deck back to you. Ours argues with it, then shows you where each objection came from.
THE FORENSIC AUDITOR
Takes the mess you send — PDFs, spreadsheets, cap tables — and turns it into one clean, comparable set of facts.
THE FISCAL HAWK
Hunts for numbers that do not add up: odd burn, missing months, revenue that appears without a customer to explain it.
THE TECHNOLOGIST
Asks whether the technology is actually hard to copy, and measures the claims against sector norms and past deals.
THE LEAD INVESTOR
Weighs everything the other agents found and returns one score from 0 to 100, with the reasons written out. Deck to decision in under 180 seconds.
THE WAR ROOM
Works out what a war, a sanction or a closed shipping lane would do to this company — supply chains, chokepoints, cyber, money, people and rules.
GEOPOLITICAL & CONFLICT RADAR
G_R = Σ(W_i · C_i · S_i)
i ∈ {hw, log, cyb, fin, hum, reg}
Six vectors, weighted and combined into one exposure score
Physical & Commodities
Rare earth minerals, raw energy (oil/gas), semiconductor foundries, and physical manufacturing nodes.
Logistics & Chokepoints
Maritime straits (Hormuz, Malacca, Red Sea), closed airspace, and border checkpoints.
Cyber & Infrastructure
State-sponsored wiper malware, reliance on regional data centers near kinetic zones, undersea cable severing.
Financial & FX
Exposure to SWIFT banking bans, FX hyperinflation, capital controls, and asset freezing.
Human Capital
Conscription/draft risk for offshore engineering teams (Eastern Europe, Middle East) and mass displacement.
Regulatory & Sanctions
Secondary sanctions, ITAR/EAR export controls, and foreign government expropriation of IP.
⚠ The "Dark Data" Rule
If a startup operating in hardware, deep-tech, logistics, or global SaaS completely omits their supply chain, server locations, or offshore engineering geography — the Geopolitical Risk is automatically flagged as "ELEVATED".
Output — DEFCON Status
The output names the exposed vector in plain words. No hedging, no "may potentially face headwinds".
PROPRIETARY MOAT
THE VERDICT INTEGRITY LOOP™
WE DO NOT TRAIN ON DATA. WE TRAIN ON REALITY.
THE PREDICTION
SenseCore issues a verdict with cryptographically time-stamped logic chain. Example output: Score 63/100 — Governance Risk Detected.
THE REALITY CHECK
System polls sovereign registries — MCA Data, Banking Data. Did revenue collapse? Did founder exit? Did loan default?
THE EVOLUTION
If Correct: Verdict logic pattern reinforced — weighted higher in future analyses. If Incorrect: Pattern flagged and deprioritized. The system adapts based on real-world outcome data, not assumptions.
THE IMPOSSIBILITY
You cannot backfill a prediction you didn't make. This moat is absolute.
WHAT ELSE COMES WITH ZURVEK
Scoring a deal before you write the cheque is where Zurvek starts. These modules apply the same discipline to the money you have already put out.
Zurvek Vault
InstitutionalForensic intelligence on capital already deployed. Real valuation against book value, live burn against projection, governance reconstructed from MCA21 and ROC filings.
Zurvek CFO
InstitutionalA portfolio CFO mandate. Connected financials, sector benchmarks, and forward IRR projection under bear, base, and bull scenarios.
CENTRA
LiveThe in-product intelligence layer. Context-aware guidance across the evaluation surface, from ingestion through committee sign-off.
Geopolitical Conflict Radar
July 2026Synthesises 8–12 active global hotspots and maps each to its structural impact on capital flow, supply chain, and sector concentration risk.
Zurvek MCP
Coming soonModel Context Protocol access, so the engine can be queried directly from the tools an investment team already runs.
Zero-Inference Protocol
LiveEvery score, flag, and recommendation is traceable to its source document. Nothing is asserted that cannot be shown.
Product 01 — Zurvek // Pricing
WHAT ZURVEK COSTS
For Investors · VC Funds · Family Offices · Accelerators
Everyone starts with 5 free evaluations. After that, four tiers for investment teams, charged in INR with USD shown for reference. Applications reviewed on a rolling basis.
EXPLORER
FREEFirst contact with adversarial intelligence.
- ›5 free evaluations — every user, every segment
- ›Full risk engine access
- ›Audit trail included
- ›PDF export
- ›Single user
ANALYST
ACTIVE DEALFLOWActive deal flow. Individual investor or solo GP.
- ›50 Decisions / month
- ›₹320 per decision — less than one hour of analyst time
- ›Historical reports
- ›Audit logs
- ›Everything in Explorer
- ›Email support
- ›Deal memo export
- ›Central AI terminal
SYNDICATE
POPULARGrowth funds, family offices, small fund committees.
- ›300 Decisions / month
- ›₹117 per decision — cheaper than a single Bloomberg query
- ›Committee workspace
- ›Shared reports
- ›Everything in Analyst
- ›Portfolio-level insights
- ›Priority processing
- ›Deal pipeline board
INSTITUTIONAL
ENTERPRISEFunds managing >$10M AUM. IC-grade infrastructure.
- ›Unlimited decisions
- ›Zurvek Vault — portfolio forensics suite
- ›Zurvek CFO — portfolio CFO mandate
- ›Unlimited portfolio companies
- ›Everything in Syndicate
- ›Forensic audit on deployed capital
- ›Exit readiness & IRR projections
- ›Custom integrations
- ›Dedicated account manager
- ›SLA guarantees
- ›Audit-locked history
Custom pricing available for >$50M AUM funds
ORIGIN is Zurvek for founders — the same adversarial engine, run on your own company. Starts with 5 free evaluations.
Open ORIGIN →Product 02 — Public Sector
CHETRA — PUBLIC SERVICE PERFORMANCE OS
One live record for every government school and hospital, so a department can see which building is in the worst shape, fix it first, and pay for it out of money it has already been sanctioned.
Institution Register
One live record per government school and hospital — building, staffing, service delivery, stock, vendors and spend held against the same institution.
Condition Score
Deterministic 0–100 score across structural, utilities, sanitation, safety, equipment, accessibility and digital readiness. Fixed weights, no black box.
Priority Ranking
Institutions ordered by condition risk, population served, service criticality, structural and safety exposure, cost efficiency and equity.
Budget Allocation
Interventions costed and allocated against the envelope a department already holds — no request for additional money.
Evidence Verification
Completion is closed only against before-and-after evidence from the field, not against a paper certificate.
Reads, Does Not Replace
Designed to read slices of UDISE+, HMIS, payroll and procurement rather than replace them.
Build status
CHETRA is in development and does not claim to be complete. The public site runs on simulated demonstration data and is not a government record. The operational layer — attendance, staffing, stock, vendor performance and cost per student or patient episode — is still being built. AI is used only to convert inspection photographs and notes into structured findings; it never sets condition scores, priority ranks or budget allocations.
chetra.tech
Product 03 — Public Sector
NIRNAE — FORENSIC MONEY-TRAIL ENGINE
Send the expenditure data you already hold. Nine detection agents send back the payments worth a closer look, the firms behind them, and the arithmetic for every flag. Nothing to integrate, no change to your systems.
Threshold evasion — payments split to stay under sanction or tender ceilings
Vendor collusion rings — shared address, phone, bank account, tax-ID series or directors
Ghost billing and phantom works — vague line items, duplicate bills across years
Timing anomalies — March-end fund dumping, same-day sanction and payment
Geography mismatch — vendors far from the works site or outside jurisdiction
Beneficiary history — firms incorporated days before a large award, dormant entities revived
Head-of-account misuse — expenditure booked to the wrong scheme or budget head
Pass-through and circular routing — layered intermediaries adding no work
Consolidation — one severity score per entity across all corroborating findings
Stated limits
NIRNAE produces indicators for verification, not legal conclusions. It cannot confirm that physical work happened, cannot prove intent or establish liability, and cannot see records a department has not supplied. It does not replace a statutory audit, inspection or vigilance inquiry. False positives are expected and are dismissible from the attached figures.
Who it serves
nirnae.tech
IF YOU SIGN OFF ON MONEY, THIS IS FOR YOU
Private capital or public budget, the job is the same: decide with evidence instead of a feeling, and be able to show how you got there.
VC funds
Screen more deals without adding analysts. Every partner sees the same score, built the same way.
ZurvekFamily offices
A second opinion that does not depend on who introduced the deal.
ZurvekAccelerators
Rank an entire application cohort in an afternoon instead of a fortnight.
ZurvekFounders
Run the same engine on your own company before an investor does. Five free evaluations, like everyone else.
Zurvek for FoundersUnion ministries and state departments
Know the real condition of every school and hospital you fund, and what fixing it costs.
CHETRAFinance, treasury, audit and vigilance
Find the spending patterns worth a closer look, with the arithmetic attached.
NIRNAEPSUs and public sector banks
Scored findings on expenditure and counterparties, from data you already hold.
NIRNAEWE BUILD THE MACHINERY BEHIND BIG DECISIONS.
A fund partner and a district officer have the same problem. Too much paper, too little time, and a decision that has to hold up later. We build systems that read the evidence, score it the same way every time, and show the working.
The hard part is not the chat window. It is getting five AI agents to disagree usefully, fixing the weights so the same file always gets the same score, and refusing to state anything that cannot be traced to a source document. That last rule is why our output is defensible in a committee room. Our own model, the SenseCore SLM, is in development and targeted for post-raise deployment — it is not running today.
Built in Noida, used wherever money needs justifying. We do not predict the future. We organise the present so you can decide with your eyes open.
Applications reviewed on a rolling basis.