Know who's
leaving before
they do.
The whole DataPlant customer-success platform, running on your own machine. Health scores, churn models, 360 views and an assistant that answers from your data on your own models.
Every screen is real.
Every number is computed.
Nothing here is a mockup. These are screen recordings of DataPlant 1.4.1 on one Windows machine, with a 40-account sample workspace and a local model.
The work happens before you open it.
A local job runner re-scores health, re-predicts churn, rebuilds cohorts and writes summaries whenever the data changes. This is its actual log.
- 01Re-score and re-predict after every upload or sync.
- 02Write the 360 synopsis for each account on your model, in the background.
- 03Send what's due: scheduled campaigns, digests, follow-up steps.
- Re-score scorecards after a data changescorecard.rescore_for_foundrySUCCESS · 5s
- Predict after a data changeoptmly.predict_for_foundrySUCCESS · 5s
- Rebuild cohort factcohort.rebuild_fact · accountsSUCCESS · 11s
- Generate KPI insight cardsmetrics.generate_insight_cardsSUCCESS · 20s
- Refresh dynamic cohortscohort.refresh_dynamicSUCCESS · 6s
- Recompute rollups over a time serieshierarchy.foundry_changedSUCCESS · 4s
- Write 360 synopsis · Globex Corporationentities.generate_360_synopsis · local modelSUCCESS · 1m 32s
- Re-check card alerts after a data changenotifications.evaluate_for_objectQUEUED
- Executive Review · insights digestcampaigns.run_due → outboxSENT
An assistant that reads your book,
not the internet.
Answers are grounded in your health scores, churn predictions, renewals and seed lists. Point it at the Meterless Runtime and the model runs on hardware you own.
Ten seconds each.
Short loops of one thing at a time. Click any clip to watch it large.
Drag to switch themes.
The same scorecard, the same numbers. Every screen ships in both.

Thirty-three screens. Zero mockups.
Filter by module, click to open full size, use the arrow keys to page through.
One installer. The whole stack.
The app starts its own PostgreSQL, boots the DataPlant backend and opens the interface. Your data lives in one folder you can see. AI is optional, and you choose where it runs.
- DBPostgreSQL 17, embedded, on a private port with its own credentials.
- AIMeterless Runtime, OpenAI, Anthropic, Gemini, OpenRouter or any OpenAI-compatible server. Or none: offline summaries still work.
- INCSV, Salesforce, HubSpot, Zendesk, S3, Slack, Google, Dynamics and Gmail Takeout imports.
From the cloud to your desk.
DataPlant began as a cloud platform for customer-success teams at thedataplant.com. Meterless DataPlant is that platform, rebuilt to run on one machine.
For teams, hosted
- Multi-tenant SaaS with team sign-in
- Hosted data and hosted AI
- Paid plans for teams
- Learn more at thedataplant.com
For you, on your machine
- Every module: Home, Insights, 360, Modeller, Foundry, Campaigns, Manage
- Your data stays in a folder on your PC
- Bring your own model, or run it locally with the Meterless Runtime
- Free · Windows installer · no account
Install it. Load a CSV.
Watch it score.
The installer bundles Python and PostgreSQL. First launch creates the database, then a workspace. Load the sample CSVs from the repo and the scorecards, churn model and insight cards build themselves.
SHA-256 setup.exe f5661a805f6babf675c5b7f22d739e11e6bf86887b21c5703e0f531133180f92