WHYSE
01 / 12

WHYSE

Perplexity for data science and analysis.

Powerful proprietary algorithms at scale.
Deep insights on big data in under 7 minutes.

The problem

Dashboards tell you what happened.
Nobody tells you why.

Analytics stops at the report. The moment a metric moves, teams fall back to guesswork and endless slicing — never quite trusting the number in front of them. That's data paranoia.

Our motivation

Let the data talk.

Eliminate data paranoia. Every answer is traceable to the exact rows that produced it — no black box, no guesswork. People finally trust what they see.

01

Traceable

Every number maps back to the source rows.

02

Reproducible

The same cut gives the same answer, every time.

03

Trusted

No paranoia — the data speaks for itself.

Versus Claude, ChatGPT & text-to-SQL

They generate SQL.
We run algorithms.

Text-to-SQL

Generates SQL from natural language. Still bound by what SQL can express — and still needs you to know what to ask.

WHYSE

Converts text to parameters, then calls purpose-built algorithms on the OLAP table. No code generation, no correction loop.

Text-to-SQL

Orchestrators generate, test, and fix SQL in a loop. Expensive, and accuracy still isn't guaranteed.

WHYSE

Known algorithms, known outputs. Cheaper, more accurate, no iteration.

Text-to-SQL

You choose the dimensions to slice by. If you don't ask the right question, you don't get the answer.

WHYSE

Point at a metric. We find which dimensions explain the move — no hypothesis needed.

The case

Why WHYSE?

Not one feature — a system. The engine, the explainable artifacts, the flexibility for your experts, and where it all lands in planning.

What you'd be licensing

One engine. Six capabilities.

01

Contribution Decomposition

02

Anomaly & Rare Events

03

Deficit & Surplus

04

Pattern Discovery

05

Temporal Drift

06

Predictive Segmentation

Drill-down, slicing, and filtering are built in — all on the same engine.

Explainability layer

Artifacts that make your agents smarter.

WHYSE emits structured, explainable artifacts — the drivers, the evidence, the decomposition. Your LLMs and agents consume them to summarise complex situations quickly, simply, and accurately, without inventing the details.

01

Structured output

Machine-readable evidence, not prose.

02

Explainable

Human- and agent-legible by design.

03

Agent-ready

The grounding a raw LLM lacks.

Who operates it

Rigorous by default. Flexible for experts.

WHYSE lets the data talk automatically — and hands the wheel to your subject-matter experts when they want it. Analysts can steer, constrain, and extend the analysis without giving up correctness.

MODE 01

Automatic

Point it at the data. It surfaces the drivers with zero setup — the data talks on its own.

MODE 02

Expert-guided

Analysts steer the search and add domain knowledge — with the same 100% numeric accountability.

Where it lands

Grounded storytelling and OKR planning.

Turn a metric movement into a clear, evidence-backed narrative — then set and pressure-test objectives against what the data actually supports.

01

Grounded storytelling

A metric shift becomes a narrative anyone in the room can follow and trust.

02

OKR planning

Objectives tested against the evidence, not a hopeful guess.

Integration

Three ways in. Zero rewrites.

Pick whichever fits your architecture. All three are the same engine underneath.

PATH 01

MCP Server

SaaS-based. Your agent calls each capability as a named tool. We host — zero infrastructure on your side.

PATH 02

Hosted Service

We run the engine, you call a REST API. Fastest path to production.

PATH 03

Embedded Library

Install as a dependency. Runs on your infrastructure — your data never leaves.

Proof of work

Three industries. One engine.

Same engine, no retuning between them. Nothing in it is domain-specific.

NYC Green Cabs

"How did fare contribution shift between January and February?"

U.S. Energy · EIA

"Which states and fuel types drove the Q3 production variance?"

DriveU · Transport

"Which booking type and zone shows the most anomalous trip outcomes?"

Team

The team.

Vijay Rajan Founder Veteran data scientist. Ex-Google, ex-Yahoo.
Ravi Tata Founder & Fractional VPE Data platforms at scale. Ex-Warner Bros., ex-Walmart, ex-Yahoo.
Girish Balakrishnan Investor & Advisor Stanford CS. Seed investor, enterprise SaaS.
Rajesh Natarajan Advisor Data science professor, Krea University. Ex-IIM Lucknow.
Dhruv Rekhawat Founding Engineer Builds AI-native systems from the ground up.
← → · N notes · P presenter