Verified analytics engine

Ask your database anything. Trust every answer. Verified before you act. Reports in seconds. Decisions, not data.

Thor turns plain-English questions into self-checked SQL, source data, and dashboard-ready insights — so business teams stop second-guessing the numbers and start shipping decisions.

Start 7 days free trial See a demo

Thor is powered by eSapiens proprietary database reasoning and self-checking technology. Read the THOR paper →

Ask in plain English

Generate report + SQL + data

Build dashboards instantly

Customize with rules

Structured data flow

Ask

Generate

Dashboard

Customization

Ask

Ask the way your team already thinks. Start from a business question and let Thor translate intent into structured analysis.

What changed in revenue last quarter? Break it down by product line. Compare it to the previous quarter. Flag the anomaly.

Thor is preparing the answer path. Loading schema hints, business rules, and query context before generation.

Ask Thor Why did SMB revenue dip while enterprise stayed stable?

Revenue QoQ Anomaly Product line

Routing

Structured query detected Thor classifies intent before anything else runs.

T2S ✓

RAG

Chat

Confidence94%

Context loading

Schema + rules joined

sales_factschema

Revenue rulerule

Prior Q contextmemory

Soft-delete filterglobal

Generate

Generate report, SQL, and data in one pass. Thor builds the answer, the read-only query, and the source rows together.

Report Revenue fell 8.4% QoQ. SMB contraction led the drop.

SQL Read-only Schema-aware generation.

Data Rows attached Evidence stays visible.

SELECT product_line, SUM(net_revenue)
FROM sales_fact
WHERE status IN ('paid', 'delivered')
GROUP BY product_line

Self-check score

Rated before delivery

80/100

✓ Rows returnedpass

✓ No exec errorpass

↻ Score 62 → retryfix

Guardrail

Write-proof by design.

Blocked at the gate

INSERT UPDATE DELETE

Always allowed

SELECT ✓

No write path exists. Schema stays intact.

Dashboard

Turn validated insight into a real dashboard. Good answers can become persistent views your team revisits every day.

Net revenue $2.18M Quarter view

QoQ -8.4% Validated change

Flagged SMB Main anomaly

Live dashboard block

Revenue dip by product line

$2.18MThis quarter

−8.4%vs last Q

Evidence trail

Proof stays attached

📄Report narrativelinked

⌗SQL queryviewable

⊞Source rows (3)attached

Rules

Customize rules and train Thor on your tables. Save metric logic, filters, joins, and training context so future answers stay aligned.

Revenue logic Paid + delivered only

Active

Soft delete filter Applied across all tables

Global

Stripe join path orders -> payments by stripe_id

Saved

Train tables Schema memory

sales_fact Revenue source of truth

trained

customers_dim Segment and lifecycle attributes

trained

payments_log Settlement and delivery status

linked

Rule impact

Rules compound over time

Revenue logic51×

Soft-delete filter112×

Stripe join27×

Schema memory

Tables stay teachable

sales_facttrained

customers_dimtrained

payments_loglinked

              
                ?
              
              Ask in plain English
              Start from intent — no SQL, no backlog, no analyst needed.

              
              See the proof
              SQL, source rows, and reasoning travel with every answer.

              
                80
                
              Self-check weak output
              Thor retries broken or low-scoring SQL before it reaches users.

              
              Generate dashboards
              Promote validated answers into persistent, reusable views.

              
              Teach Thor your rules
              Save metric logic and filters so future answers stay on point.

Trusted by data teams

trusted by top teams.

Real data teams — what changed after they moved their analytics to Thor.

"We used to wait two days for ad-hoc queries. Now the team gets answers in minutes — and can see exactly the SQL that produced them. That transparency changed how we trust AI-generated data entirely."

Sarah R. Head of Analytics · @saasstartup

"The self-check loop is the thing no other tool has. Before Thor, we'd ship a dashboard only to find the SQL was wrong. Now we catch it at generation time, not production time."

Marcus K. VP Engineering · @fintechplatform

"Our product and ops teams now answer their own data questions. Saved rules mean everyone uses the same revenue definition. No more spreadsheet discrepancies in board meetings."

Jamie L. Chief Operating Officer · @ecomco

"We work with sensitive patient data across a dozen hospital systems. Thor's read-only guardrails meant we could roll this out to clinical staff without a single compliance review escalation."

Priya S. Data Team Lead · @healthanalytics

"I handed Thor to our three data engineers on Monday. By Wednesday they'd deprecated two internal BI tools and had dashboards running from natural language. The setup was shockingly fast."

Daniel W. CTO · @growthstartup

Core flow

ask. validate. generate. reuse.

One path from plain-English question to shared, checked dashboard output — question routed, SQL generated and self-verified, then promoted to a live view your whole team can act on.

Start free trial

Start with a prompt. Save shared questions everyone follows.

Thor checks logic, retries weak SQL, and keeps evidence visible.

Turn a trusted answer into a dashboard the team can keep exploring.

Thor saves reusable rules, and your team can add or edit them anytime.

Animated preview ask starts with a real team question.

Quick start seeds the first ask. Shared questions keep the team aligned.

ask

quick start

ask any business question.

Start with Quick start, then share the asks the team should watch together.

Why did pipeline coverage fall in smb this week?

quick start shared ask team thread

starter Revenue dip watch

shared Visible to sales + finance

Technology deep dive

technology. deep dive.

Thor learns the database deeply, checks every answer path with SQL visible for QC, then turns strong answers into flexible team-ready output.

Database learning

Thor builds context before it answers.

ThunderScan, schema, and skills become one usable model.

ThunderScan schema map saved skills

scan AI readiness + schema health

schema columns + relationships

skills user + system logic

suggested quick start Why did pipeline coverage fall in smb while enterprise stayed flat?

Self-check loop

Thor checks the path before it ships.

Each run stays reviewable before delivery.

route schema + joins

run read-only SQL

check score + retry

proof SQL and rows stay attached

review Customer can QC before ship

result score

pass ready after one retry

proof kept

sql

rows

retry

Flexible delivery

One checked answer can become team-ready output.

Export, share, or pin it right away.

checked answer

Coverage shift detected

checked ready to deliver

doc export

ready to send

dashboard

save as tile

team share

post to team

connector version-aware

Concrete support by engine.

mode read-only

Safe analytics only.

ready now 5 engines

More warehouse support coming.

Compatibility

connect. around Thor.

Thor already connects to the databases teams run every day, with more warehouse support on the way.

Supported now

PostgreSQL MySQL SQL Server Oracle Aurora

More to come

Security & compliance

enterprise-grade protection.

Thor connects read-only by design — no write paths, no schema changes. Every query is encrypted in transit and the full SQL evidence trail stays visible for internal audit at every step.

Read-only only

Query mode

access SELECT only

risk no writes

Read-only by construction

Thor runs analytical questions without schema mutation, update paths, or hidden write risk.

transport Encrypted in transit

review SQL and proof remain visible

Encrypted and reviewable

Connections stay encrypted while the answer path remains inspectable for internal review and audit.

retention Zero-copy posture

answers Computed on demand

Zero-retention by default

Answers are generated live when needed, without creating shadow stores of customer data behind the scenes.

Rules & skills

rules. skills. remembered.

Define your revenue logic, table relationships, and default filters once. Thor applies them to every future question automatically — keeping answers consistent across teams without re-explaining anything.

Shared definitions

Revenue, pipeline, and retention stay defined the same way across every team ask.

Join memory

Approved joins, scopes, and warehouse paths are remembered instead of re-explained.

What counts as expansion revenue?

source

skills

saved joins

scoped filters

memory save logic once.
apply it to every ask.

logic

permissions

shape

output

consistent metric names

safe default scopes

dashboard-ready output

Scoped filters

Soft-delete, region, and customer-safe filters can be attached once and reused everywhere.

Skills and output shape

Teams can save preferred phrasing, dashboard blocks, and answer formats close to the engine.

Get started

Ask your first question in minutes.

Connect the warehouse, let Thor learn the model, and move from question to checked dashboard output in one path.

What do you want Thor to answer first?