Why DataMind

Solutions

Company

DataMind Universe

Free trial

The whole OS. Two months. Free.

Every app, on your own servers, with your own data.

  • Your data stays on your servers On your servers
  • Installed in about 20 minutes 20-minute install
  • No card, no feature limits No card

One quick check, then it sends.

Your request goes to our sales team. How we handle it (opens in a new tab)

Ask DataMind

DataMind:

Hello, Winner.

Tell me who you are and what slows you down. I will show you how DataMind OS fixes it.

This chat is not stored. Answers run on zero-data-retention model endpoints. Privacy

DataMind OS vs Palantir Foundry

Everything Foundry does. Installed by your own team.

DataMind OS covers the work teams use Foundry for: integrating every source, modeling it in a governed warehouse, controlling access down to the row, analyzing it with AI agents and pushing results back into operational systems.

01Verdict

The short answer.

DataMind OS covers everything teams use Palantir Foundry for, and adds much more on top: DataPrism BI, Slides and Oda, in one Intelligent Data OS that your own team installs on your own servers, at one published company price.

Palantir Foundry today. Data integration and operations software from Palantir, built around the Ontology (objects, links, actions and functions over integrated data) and sold with AIP, Palantir's AI layer, as managed SaaS or a Palantir-managed on-premises appliance.

  • On your servers.

    Installed on your own Linux servers in about 20 minutes, in any cloud or data center.

  • One OS, one login.

    Connectors, Lake, Warehouse, DataPrism BI, Slides and AI Minds share one set of access rules in the database.

  • Free builders and viewers.

    Report and dashboard builders, Mind builders, viewers and subscribers are free for an unlimited number of users.

02Side by side

DataMind OS and Palantir Foundry, row by row.

14 topics. Every Palantir Foundry fact cites a numbered source, checked October 2026.

DataMind OS compared with Palantir Foundry, checked October 2026
Topic DataMind OS Palantir Foundry
Deployment Installed by your own team on your own servers: one x86_64 Linux server with Docker (Ubuntu, Debian, RHEL, Rocky or Alma), physical or virtual, in any cloud or data center. The browser Installer runs the install in about 20 minutes on the recommended hardware, sizes memory limits to the host, and updates itself with an automatic rollback; a new release installs from the browser every month.3, 4 Foundry SaaS runs on AWS, Azure, Oracle and GCP in the United States, Canada, the European Union, Japan, Australia and Brazil. Foundry On-Premises is a Foundry Appliance fully managed by Palantir through Apollo.1, 2
Price One published company plan, on the pricing page, with one Privileged seat included (integrations, the data warehouse and access permissions). Everything else is free for an unlimited number of users, every app is in the plan, 85M Kartlos Premium and 850M Kartlos Standard tokens a month are included, and DataMind meters no compute, storage or queries.8 No published price. Usage is metered in compute-seconds and gigabyte-months, and LLM tokens convert to compute-seconds per model. The UK G-Cloud 14 rate card (May 2024) lists a single-organization Foundry license at £3,000,000 a year including £3,000,000 of usage, and implementation at £150,000 per person per quarter.1, 5, 6, 7
Time to production About 20 minutes from a bare Linux server to a running DataMind OS, with a demo semantic model and report on day one. The Lake Builder Mind imports up to 1,000 tables at once under approval, Report Builder AI builds a dashboard page from a screenshot in minutes, and the two-month free trial runs the whole OS on your own servers.3, 11 AIP Bootcamps go from zero to a use case in hours or days alongside Palantir engineers, and a free AIP Developer Tier gives individuals a trial account. A Palantir implementation partner estimates 6 to 12 weeks to a narrow first production use case.9, 10
Getting data in 22 built-in source types, from MSSQL, Oracle, Postgres, MySQL, MariaDB, Snowflake, BigQuery, Redshift and MongoDB to S3, SFTP file drops, RabbitMQ and signed webhooks; old SQL Server and Oracle versions connect as they are. Connect Any API has AI build a connector for any REST, OData or GraphQL API from its docs, ERP OData services included, and a connector agent writes and tests custom Python connectors. No fee per connector or synced row. Data Connection syncs batch and streaming data from relational databases, FTPS, HDFS, S3, SFTP and other source types, and exports datasets back out as files, table rows or streams. Since August 2026, Foundry Worker runs ingest and export without compute charges.12, 13, 14
Storage and SQL Lake and Warehouse tables are ClickHouse tables on your own servers, columnar and ZSTD-compressed, and each table shows its real compressed size. Charts, exports and subscriptions query them live, and each user has a matching database account, so DataGrip, DBeaver, Power BI and Excel connect under the same object, column and row rules. The browser SQL Editor adds schema-aware autocomplete and AI next-step suggestions. Data stays in its original format, and transformed data defaults to Iceberg and Parquet, reachable through REST, JDBC and S3-compatible access. SQL runs in a Spark SQL dialect on the Furnace engine and the Ontology SQL engine, from SQL Studio, Arrow Flight SQL or the SQL REST API.15, 16
Transformation and modeling A Warehouse that builds itself: dimensions, facts, dictionaries, static tables and views, DDL generated on Deploy, nine load patterns with generated load SQL, SCD Type 2 history as a switch, a duplicate-key check on every load and a Test Run on real data before a table changes. Generate SQL with AI works in a diff, and the Warehouse Builder Mind builds facts and dimensions in chat, under approval. Pipeline Builder for no-code pipelines, Code Repositories in Python, Java and SQL on a Git service, Code Workspaces with Jupyter and RStudio, and Global Branching, which reached TypeScript v2 and Python functions and Developer Console in October 2026.15, 17, 18, 14
Orchestration Built in, with no DAG to maintain. The Warehouse run plan builds itself from dependencies, runs independent steps in parallel, waits on other workflows through gates and refuses circular dependencies. A restart reruns only unfinished steps, temporary errors get up to 10 automatic retries, any run replays on a timeline, failure alerts reach Email, Slack, Discord or Google Chat, and a failed workflow can wake a Builder Mind to investigate. Builds run on schedules with time or event triggers and retries, and exports are scheduled from Data Lineage like any other job.13, 19
Lineage Lineage the engine writes: saving a transformation, view or dictionary runs ClickHouse's own EXPLAIN QUERY TREE and records every table it reads. Lanes run from sources through Lake and Warehouse to Writeback, semantic models, Minds and Slides, filtered to what each viewer may see, and every table lists its dependants before a delete. Data Lineage is an interactive graph with search by project, table or column name and the code that generated each dataset, and metadata services expose lineage across projects, datasets, Ontology elements, agents and models.20, 15
Access control Object, column and row rules compile into per-role ClickHouse views, so the same limits hold in the SQL editor, DataPrism, AI Minds and external tools. Minds read only the columns granted to them or run as the asking person, columns marked as personal data reach AI as tokens, one Audit button downloads a full permissions workbook, and the request and AI audit logs stay in your own database. Markings and classification-based access controls propagate to derived data, and restricted views and object and property security policies give row, column and cell-level security. Since October 2026, administrators can allowlist the markings a session may carry into LLM requests.21, 22, 14
Data back into operations Writeback sends rows from a Lake table, a Warehouse table or your own SQL into an existing table in MSSQL, Postgres, ClickHouse, Oracle, MySQL, MariaDB or BigQuery, or as JSON messages to a RabbitMQ exchange, by hand or on a schedule, with column mapping and a Sync Log. Writebacks appear in lineage beside their sources.23 Ontology Actions write user edits back to objects, webhooks reach operational systems, and Data Connection exports files, table rows and streams to external systems on a schedule.5, 13, 15
BI and dashboards DataPrism, in the plan for every user: semantic models with SQL measures and calculated columns, 130 measure recipes, up to 50 tests per measure, 15 visual types plus custom interactive visuals you create with AI on the VDK, Compare mode, 3D Maps, bookmarks and stories, row-level security with View as, subscriptions up to 24 times a day under each recipient's rows, and a native streaming download of any visual, even a billion rows as CSV or TSV.26, 27 Contour and Quiver for analysis and dashboards, Workshop for operational applications, Notepad for reports, and out-of-the-box connectors for Power BI and Tableau.15, 24, 25
Presentations DataMind Slides: Slides AI builds and edits decks, a Mind turns its analysis into a deck inside the chat, and charts, tables and text bind to live SQL and refresh on open. Decks export to PowerPoint with native charts that carry an embedded workbook, existing .pptx files import as editable decks, and decks email on a schedule. Notepad and AIP Analyst export PDF, and since October 2026 organizations can require watermarks on AIP Analyst PDF exports.28, 14
AI agents More than 20 purpose-built AI agents share more than 250 tools. AI Minds answer by writing and running SQL on governed tables, every step a card with its SQL and result; the Lake Builder and Warehouse Builder Minds do the data engineering under Manual or Auto-accept approval; Report Builder AI, Model AI, Custom Visual AI and Slides AI build reports, models, visuals and decks; Swarms run many agents in parallel, and Benchmarks score Minds on your own questions. AIP Analyst for conversational analysis, AIP Logic for LLM-backed functions, AIP Assist, AI FDE for building in plain language, and Palantir MCP for agent access to the Ontology.6, 15, 21
AI models and data handling Kartlos Premium and Kartlos Standard, DataMind's built-in model tiers backed by frontier models DataMind curates, are always on, with 85M and 850M tokens a month in the plan; administrators can allow more catalog models, billed per token. The AI gateway sends calls only to zero-data-retention endpoints and denies provider data collection, a Region setting keeps AI chat on EU endpoints, and every AI request is logged in your own database. Model-agnostic, with models from OpenAI, Anthropic, Google, xAI, Meta, Mistral, NVIDIA and open models from DeepSeek, Moonshot, Z.ai and Qwen. Third-party providers retain no prompts or completions and do not train on them, regional endpoints serve the US, UK and EU for some models, and tokens bill as compute-seconds.6, 14, 29

Checked . Sources at the foot of the page.

03Further

Where DataMind OS goes further.

All of it in the same plan, on the same servers, under the same rules.

  1. One published price

    One company plan on the pricing page, every app included, with no compute-seconds or gigabyte-months to meter.

  2. Installed by your own team

    The browser Installer sets up DataMind OS on one Linux server in about 20 minutes, sizes memory to the host and updates itself with an automatic rollback.

  3. DataPrism BI

    Semantic models with SQL measures and calculated columns, up to 50 tests per measure, row-level security with View as, and subscriptions under each recipient's rows.

  4. Report Builder AI

    Rebuilds a Contour or Quiver board from its screenshot on your semantic model in minutes, and Model AI writes the measures in SQL.

  5. Slides

    Slides AI builds decks bound to live SQL, and decks export to PowerPoint with native charts that carry an embedded workbook.

  6. Free builders and viewers

    Report and dashboard builders, Mind builders, Oda users, viewers and subscribers are free for an unlimited number of users: 1,000 report developers and 10,000 viewers cost nothing.

  7. AI Minds with the SQL shown

    Every answer is a chain of cards with its SQL and result, Swarms run many agents in parallel, and Benchmarks score Minds on your own questions.

04Switch

Switch from Palantir Foundry.

The systems Foundry reads from connect to DataMind OS on day one, Foundry's own exports bring over what lives only in Foundry, and both run side by side until each use case moves.

  1. Connect the same source systems

    The systems Foundry reads from connect straight to DataMind OS: SQL Server, Oracle, Postgres, MySQL, MariaDB, Snowflake, BigQuery, Redshift, Athena, MongoDB, S3 buckets and SFTP file drops are among the 22 built-in sources, SAP comes in through its OData services with Connect Any API, and any other REST, OData or GraphQL API connects the same way.

  2. Bring over what lives only in Foundry

    Foundry's own Data Connection exports carry it out: a file export writes a dataset's Parquet files to your S3 bucket, which DataMind OS loads as a file source, and a table export inserts rows into a database table that DataMind OS reads as a Lake source.

  3. Rebuild the pipelines as tested SQL

    Pipeline Builder and Code Repositories logic becomes ClickHouse SQL transformations: Generate SQL with AI drafts each one in a diff, Import from SQL builds the columns, the Warehouse Builder Mind builds dimensions and facts in chat, and Test Run checks every insert, update and delete on real data before a table changes.

  4. Rebuild dashboards from screenshots

    Contour and Quiver boards come over page by page: Report Builder AI rebuilds each from its screenshot on your semantic model in minutes, and Model AI writes the measures in SQL from plain language.

  5. Keep operations fed

    Results that Foundry pushed into operational systems move to Writeback, which appends modeled rows to existing tables in MSSQL, Postgres, ClickHouse, Oracle, MySQL, MariaDB or BigQuery, or publishes them to a RabbitMQ exchange, on a schedule.

  6. Give analysts AI Minds

    People who asked AIP questions get AI Minds on the same governed tables, each answer showing its SQL, and with Execute as User each person sees only their own rows.

  7. In parallel

    Foundry keeps running while DataMind OS loads the same source systems, so both read the same data during the move and each use case cuts over when its reports, Minds and writebacks are ready. The two-month free trial runs the whole OS on your servers with your own sources before any contract.

Q and A

Questions, answered.

What teams ask before they move from Palantir Foundry.

What is a good Palantir Foundry alternative?

DataMind OS covers the work teams use Foundry for: integrating data from every source, modeling it in a governed warehouse, controlling access down to the row, analyzing it with AI agents and pushing results back into operational systems. On top it brings DataPrism BI, Slides and Oda, in one Intelligent Data OS on your own servers at one published company price.

How much does Palantir Foundry cost compared with DataMind OS?

DataMind OS publishes one company plan on its pricing page, with one Privileged seat included and everyone else free. Foundry publishes no price and meters usage in compute-seconds and gigabyte-months; the public reference point is Palantir's UK G-Cloud 14 rate card from May 2024, which lists a single-organization license at £3,000,000 a year including £3,000,000 of usage.

Can our own team install and run DataMind OS?

Yes. The browser-based DataMind Installer sets up DataMind OS on one Linux server in about 20 minutes on the recommended hardware, sizes memory limits to the host and updates itself with an automatic rollback. Administrators apply monthly updates and restart services from inside DataMind OS, and builder agents do much of the data engineering under approval.

How do we move from Foundry to DataMind OS?

Connect the source systems Foundry reads from as Lake sources on day one, bring Foundry-only datasets over through Foundry's file and table exports, rebuild pipelines as tested SQL transformations with Generate SQL with AI and the Warehouse Builder Mind, and rebuild dashboards from their screenshots with Report Builder AI. Foundry keeps running until each use case is ready.

Can DataMind OS write results back into operational systems?

Yes. Writeback sends rows from a Lake table, a Warehouse table or your own SQL into existing tables in MSSQL, Postgres, ClickHouse, Oracle, MySQL, MariaDB or BigQuery, or publishes them as JSON messages to a RabbitMQ exchange, by hand or on a schedule, with a Sync Log for every run.

Is our data stored or used to train AI models?

No. DataMind's AI gateway sends model calls only to zero-data-retention endpoints and denies provider data collection, so nothing a Mind sends is stored by DataMind or the model provider. Columns marked as personal data reach the model as tokens, and every AI request is logged in your own DataMind OS database for your auditors.

Can we prove value without a long engagement?

Yes. The two-month free trial runs the whole OS on your own servers with no card, a demo semantic model and report are ready on day one, the Lake Builder Mind imports your source tables under approval, and Report Builder AI turns a screenshot of the dashboard you want into a working page in minutes.

05Sources

Where the facts come from.

Checked . Product names are trademarks of their owners.

  1. Palantir Foundry Plans
  2. Apollo: Introduction
  3. DataMind docs: Install on a fresh host
  4. DataMind docs: System requirements
  5. Resource Management: Usage types
  6. Compute usage with AIP
  7. Palantir Platform: Foundry and AIP pricing document, G-Cloud 14 (May 2024)
  8. DataMind OS pricing
  9. Getting started with Palantir (Developer Tier, AIP Bootcamp)
  10. Palantir Foundry implementation guide (partner estimate)
  11. DataMind docs: Report Builder AI
  12. Data Connection: Overview
  13. Data Connection: Exports overview
  14. Foundry announcements, October 2026
  15. Architecture center: Interoperability
  16. SQL in Foundry: Overview
  17. Pipeline Builder: Overview
  18. Foundry announcements, May 2026 (Global Branching GA)
  19. Data integration: Schedules
  20. Data Lineage: Overview
  21. Foundry announcements, April 2026 (AIP Analyst GA, object and property security policies)
  22. Classification-based access controls
  23. DataMind docs: Writeback
  24. Analytics: Dashboards (Contour, Quiver)
  25. Application building: Workshop
  26. DataMind docs: Export reports and data
  27. DataMind docs: Email subscriptions
  28. Analytics: Exporting outputs
  29. AIP security and privacy
  30. BD Emerson: Palantir cost (partner estimate)

Try it beside Palantir Foundry.

Two months, the whole OS, on your servers. Free.