Verified answers·Never hallucinated

The AI data analystyou can actually trust.

Ask anything about your data. Get the answer, the chart, and the exact SQL + Python behind it — verified against your data, with no hallucinated numbers. From CSV to board-ready decision in 60 seconds.

✓ No credit card✓ Free forever tier✓ Every number shows its SQL
Revenue by Product — Jan–Jun 2024

REVENUE BY PRODUCT

Laptop
₹19.8L
Phone
₹15.4L
Tablet
₹9.1L
Accessories
₹5.2L
Cables
₹2.7L
142ms · rule engine
SELECT product, SUM(revenue) AS total FROM df GROUP BY product ORDER BY total DESC
0
chart types, auto-picked
0
data connectors
0s
CSV → decision
0%
answers show their SQL

Connects to the data you already have

  • PostgreSQL
  • MySQL
  • Snowflake
  • BigQuery
  • Redshift
  • Google Sheets
  • REST API
  • CSV · Excel · Parquet

The wedge no analyst tool gives you for free

Cohort retention,
vs a ~15% D2C benchmark.

Every D2C analysis surfaces an M0→M6 retention curve with a commonly-cited industry benchmark hairline. If you’re below it, the report tells you. Auto-derived from your orders + customers tables — no SQL needed.

COHORT RETENTION · M0 → M6benchmark 15%
0%25%50%75%100%M0M1M2M3M4M5M6

⚠ M1 retention 18% — above benchmark, but the dropoff to 12% by M2 is the real problem.

Specialized pipeline · not chat-with-CSV

Five specialized stages.
Each doing what it's best at.

Generic AI tools throw one model at every problem. Workliq routes each stage to the engine that natively excels — deterministic maths where the number has to be exact, then a reasoning model for analytics, an executive writer for narrative, a refiner for polish, and a fast chat layer for follow-ups. Cost-aware: simple queries skip stages.

STAGE 01

Data

deterministic engines

DuckDB · scikit · SHAP · ARIMA. No LLM. Zero hallucination.

STAGE 02

Analysis

reasoning engine

Root cause. Confidence ranking. Statistical reasoning.

STAGE 03

Executive

narrative engine

Board-grade narrative. Decision card. 90-day plan.

STAGE 04

Refine

polish engine

CEO-tier polish. Active voice. <22 word sentences.

STAGE 05

Chat

fast chat engine

Sub-second follow-ups. Interactive UX. Suggestions.

Cost stays ₹0 on free tier — most queries skip 3 of the 5 stages.

How the routing works →

Capabilities

Six analyst-grade engines.
One simple chat.

Drop a CSV. Ask in plain English. Get the SQL, the Python, the chart, and the consultant-grade insight — in 60 seconds.

Transparent

Verified SQL + Python

Every answer shows the exact query and pandas code behind it.

Rigorous

Statistical tests

t-test, ANOVA, chi-square, normality — with effect sizes.

Predictive

ML + forecasting

Segmentation, regression, SHAP, and ARIMA out of the box.

Proactive

Anomaly detection

Spikes and outliers flagged before you think to ask.

Connected

14 data connectors

Postgres, MySQL, BigQuery, Snowflake, Sheets, REST, and more.

Enterprise

Audit + workspaces

Immutable logs, RBAC, and shareable reports for teams.

Try the live demo

Pre-loaded with retail sales — no signup needed

Metrics vs diagnosis

A number is not an answer.

Knowing refunds rose 12% doesn't tell you which segment moved, how much of the change it accounts for, or what to do on Monday. Everything below is in the product today — verifiable on the live demo.

What a dashboard leaves you with

The questions a metric raises

  • · Refunds are up — which products, which cohort?
  • · Is this a real move or normal variance?
  • · How much of the total change does it explain?
  • · Is this number even correct?
  • · What is the one thing to do next?

Answering these is the analyst's actual job — and where the hours go.

What Workliq returns

1 diagnosis. 3 ranked actions. Rupee tradeoffs.

Diagnosis · 85% confidence

PMF erosion — 4 co-moving symptoms

60%
22%
12%

Next move: 30-customer cohort interview · ↑ ₹2.4L upside · 30-day reversibility

Why WorkLiq

Trust every number you ship

Checked Two Ways

Every number is recomputed by a second, independent method — deterministic SQL cross-checked against a pure-pandas pass. Answers come back marked Verified or Disputed. When two methods disagree, you see both, not a confident guess.

Says What It Doesn't Know

Each answer carries its confidence and the specific reason: sample below 30 rows, a model wrote the query, a summary claim it couldn't verify. Reports end with a "what we could not determine" section most tools never write.

Answers "Why", Not Just "What"

Revenue moved? It decomposes the change by segment, ranks what actually accounts for it, and states the counterfactual: had this segment held flat, the move would have been X instead of Y. Plus what-if projection on your own data.

Scientist-Ready

Picks the correct statistical test and tells you why it picked it. Compares several models on held-out data and names the trade-off. Segmentation, regression with SHAP, backtested forecasting.

How it works

From raw data to a clear decision in 60 seconds

2
3
4

01 · Connect

Bring your data

Upload CSV, Excel, Parquet, or a multi-table ZIP — or connect Postgres, MySQL, BigQuery, or Google Sheets.

s

from upload to a board-ready answer

Who it's for

See yourself in the product

Data Analysts

Stop writing the same SQL queries.

Ask in English, get verified results. See the exact SQL and pandas code behind every answer.

Startups

Get analyst-grade insights without hiring a data team.

Upload your CRM export or ops CSV. Board-ready insights in minutes, not days.

Researchers

Run real statistics without writing the code.

t-test, ANOVA, chi-square, regression — each with effect size and a plain-English read. Export-ready for your paper.

Students

Learn with real data — see the code behind every answer.

Every result shows the SQL and Python. The fastest way to actually learn data analysis, not just get an answer.

Pricing

Simple, honest pricing

Start with the demo — no account needed.

Switch to annual — save 20%

Starter

₹999/mo
  • 5 datasets
  • 3,000 credits / month
  • SQL + Python output
  • Anomaly detection
  • Basic insights
  • All chart types
Get Started
MOST POPULAR

Pro

₹2,499/mo
  • Unlimited datasets
  • 12,000 credits / month
  • Full ML suite (regression, clustering)
  • ARIMA forecasting
  • Statistical tests
  • Email digest reports
  • PDF + Excel export
  • Shareable reports
Start Pro →

Team

₹8,999/mo
  • Everything in Pro
  • Team workspaces + RBAC
  • 60,000 credits / month
  • 14 connectors (DBs, Shopify, Stripe, Meta, Google, GA4)
  • Scheduled reports
  • Audit log (CA-ready)
  • Priority processing
Start Team →

INR today · USD & global card payments rolling out. A fraction of US-priced alternatives.

Early access

Built with feedback from early users

WorkLiq is in early access. We ship weekly based on direct feedback from finance teams, audit firms, and operators across India. Want to shape the product?

Join the early-access program

Stop guessing.
Start knowing.

Upload your first dataset free. No credit card, no setup fees.

SQL transparency · deterministic results · built in India 🇮🇳

We use essential cookies for login and security. Optional cookies improve product analytics (Sentry session replays, page-view counts). You can decline — the product still works. Privacy details.