ProLyt
ProLyt

Professional Analytical Toolkit

Turn your questions into visualized answers in minutes.

Book a demo Try it free

Reduce your weekly workload by up to

AI-assisted data cleaning
38% on data cleaning
is what data preparation and cleaning takes today¹AI suggests the fixes, every change is logged, the original table stays untouched. No more manual Excel cleanup.
Document search
9.3 h on searching
is what knowledge workers spend finding information today²Answers come with their source passage, instead of reading whole contracts, reports and policies.
Dashboards and decks
29% on creating reports
is what dashboards, reports and presentations take today¹Turn answers into live dashboards or presentations in a couple of clicks.
Natural-language querying
Weeks → minutes + a review
for a complex reportHalf of BI leaders surveyed wait multiple weeks for a complex report³. ProLyt drafts the query and the answer within minutes⁴; your analyst reviews the SQL instead of writing it.

Figures show the time these tasks take today; actual savings depend on the work. ¹ Anaconda, State of Data Science 2022 (about 3,500 respondents). ² McKinsey Global Institute, The Social Economy (2012). ³ ThoughtSpot BI survey, January 2015 (60 IT/BI leaders; a vendor survey). ⁴ Our banked token census, 2026-10-03: 18 runs of three multi-table questions on two hosted models took 11 s to 6.7 min each (median 16 s on gpt-oss-20b, 4.7 min on gpt-oss-120b on a slow provider day); the same questions ran in 7–33 s the day before. Speed depends on the model and the provider's load.

Built on measured research, not demos.
Token costs

Save on every prompt.

9× lower AI cost
per question, on the like-for-like test
3–9.7× fewer tokens
on measured data
$720 saved
per 1,000 questions on a premium model
Measured October 2026 against pasting tables into Claude Sonnet 5.5 and Opus 5.5. Read the measurements →
How it works

From question to checked answer

Scaffolding
Scaffolding: guards
Question
Router
Table search
Document search
AI model
Read-only SQL
Checked answer
read-only · number guard · completeness · join check
Deployment

Cloud, on-prem or hybrid.

Where ProLyt runs
Not in this modeProLyt-provided modelon every plan
Not in this modeOwn model accountOllama, OpenAI or Anthropic
AvailableIn-house inference servervLLM or Ollama on company GPUs
Not in this modeRented GPURunPod or Vast.ai, billed to the company
Company network: nothing leaves
Team
in the browser
ProLyt
app and guards
Company data
databases, files, documents
Open-source model
on company GPUs
stays inside
read-only SQL
What leaves the buildingNothing. App, data and model all run on company hardware, behind the company firewall.
App, database and model on company hardware
Runs fully air-gapped, no internet needed
Open-source models on company GPUs

Simple pricing

Compare with other tools →
Explore€0 forever15 questions / monthCSV, Excel & PDF uploadCharts + full SQL traceStart free
Analyst€29 / user / month300 credits / monthUse your own provider key (OpenAI, Anthropic, Ollama)Session memoryOne-click PPTX decksDashboards & workbenchComing soon
Team€940 / monthUp to 10 users3,000 credits / month per userShared team workspace100 MB uploads (CSV/Parquet; Excel 50 MB); documents 200 MBPriority supportComing soon
Business€2,400 / monthBilled annuallyUp to 25 usersEverything in TeamEditable metric glossaryContact us to start
On-premfrom €9,600 / node / yearYour hardware, your modelsAir-gap supportedOpen-source LLMsImplementation support

A question costs one credit, a deck export ten; a question that fails is refunded. Every plan gets the same safety rails: read-only SQL, validated charts, full trace. You pay for capacity and privacy, never for safety.

Compare

How ProLyt compares

ProLytWren AIJulius AIVanna AI
Entry price€29 per user per month (Analyst)$179/month, billed annually1€18/month (Plus)2$50/month3
Free plan✓———
Published on-prem price✓✗✗✗
Self-hosted option✓✓✓✓
Use your own model/key✓✓✗✓
Frontier models included without own key✗—✓—
Data cleaning built in✓———
Document search✓———
Dashboards✓✓——
Slide decks✓—✓—
Published token-cost measurement✓———

✓ yes · ✗ no · — not stated publicly. Marks follow each vendor's own public pages (sources below); a dash is a gap in what is published, not a judgement.

From public pricing and product pages, October 2026: 1 getwren.ai/pricing · 2 julius.ai/pricing · 3 vanna.ai · 4 public directory listings. Plans change; check each vendor for current terms.

Deterministic scaffolding

Numbers from the database, not the model

The AI model writes the query and the database computes the numbers. Fixed rules around the model check the answer before anyone sees it.

AI modelwrites the query
Databasecomputes the numbers
Scaffoldingchecks the answer, or refuses it
Table retrieval
Finds the tables a question needs before the model writes a line.
Read-only SQL
Queries can read data, never change it.
Number guard
An answer with figures but no query that read your data and no cited document is refused. Figures from documents that cannot be found in them are flagged.
Completeness check
A list built from only part of your documents says so, or is refused.
Join check
Warns when a join is likely to double-count.
Validated charts
Charts are drawn from the real result rows, never generated as images.

What scaffolding cannot prove: that the model chose the right query. That is why every answer shows its SQL and its sources.

Why ProLyt

Figures with no query and no cited document behind them are refused
Every answer shows its query and sources
Read-only: never changes the data
Runs on-premise, even air-gapped
Failed questions are refunded
Same safety rails on every plan