Bauplan

Bauplan is a data lakehouse for running Python and SQL pipelines directly on Apache Iceberg tables in a user's object storage, without managing servers or clusters. It uses serverless function-as-a-service execution for Python pipelines and supports Python, SQL, and mixed workloads on the same tables. Data branches, isolated runs, atomic publishing, and rollback provide controls for production work. The platform also supports AI agents building pipelines, exploring data, and managing a lakehouse through its MCP Server and Skills. Listed integrations include Airflow, Prefect, Dagster, Temporal, DBOS, Jupyter, Marimo, Streamlit, Snowflake, BigQuery, Metabase, Superset, Fivetran, and Airbyte. Bauplan says customer data stays in the customer's S3 bucket and offers PrivateLink and bring-your-own-cloud options. The free Sandbox is a shared environment where all data is public; the company recommends it for demos and toy projects rather than company data. Dedicated is $700 a month after a 14-day free period. Enterprise pricing is custom. Bauplan describes itself as a small engineering team based in New York and San Francisco.

Bauplan is designed for pipeline execution on object-storage data, with production safeguards and deployment options. The Sandbox is public, so assess the Dedicated or Enterprise options for private data use.

Bauplan suits developers and AI agents building, validating, and shipping data pipelines on production data. The public Sandbox is intended for demos and toy projects, not company data.

Bauplan is a pipeline execution platform for developers working with Apache Iceberg data in object storage. It is strongest for teams that want production safeguards and managed execution without running clusters. Private use costs at least $700 a month, so the public Sandbox is for evaluation rather than company data.

Overview

Bauplan runs Python, SQL, or mixed workloads against Iceberg tables in a user's object storage, with no servers or clusters for the team to manage. Its branches, isolated runs, atomic publishes, and rollback offer a more controlled route from pipeline development to production than an unisolated write-and-check workflow. That focus makes it a better fit for building and shipping data pipelines than for teams seeking a general analytics environment.

Serverless function-as-a-service execution handles running, isolation, and scaling Python pipelines. Integrations connect Bauplan with orchestrators such as Airflow, Prefect, Dagster, Temporal, and DBOS, as well as notebooks, data apps, warehouses, BI tools, and ingestion services. This breadth can help it slot into an existing stack, though it remains an execution layer rather than a replacement for every tool in that stack.

Key features

Production controls

Branches and isolated runs let developers validate changes away from production data; atomic publishes and rollback help control the transition and provide a way back. These safeguards are most useful when pipeline updates can affect downstream work. They do not make the Sandbox appropriate for sensitive or company data: all data in that plan is public.

Iceberg and mixed workloads

Using Apache Iceberg tables in external object storage lets teams use Python and SQL against the same data. Bauplan also supports streaming ingestion, SQL analytics, ACID transactions, and a governance catalog. This combination suits teams standardizing pipeline work around open tables; it is less compelling for teams that do not use Iceberg or already have an execution setup that meets their needs.

AI and deployment

Bauplan says AI agents can build pipelines, explore data, and manage a lakehouse through its MCP Server and Skills. Its deployment choices include a dedicated private instance, PrivateLink, and bring-your-own-cloud. SOC 2 Type II compliance and controls covering encryption in transit, production access, monitoring, incident response, and data retention and disposal support its production-data positioning. Dedicated offers SOC 2 Type II and GDPR compliance; Enterprise adds PrivateLink or BYOC. The platform says customer data stays in the customer's S3 bucket and can be processed through PrivateLink or within the customer's VPC using BYOC.

Pricing

Bauplan has a free Sandbox, a 14-day trial, and paid plans from $700 a month. The key dividing line is privacy: Sandbox data is public, while private use begins with Dedicated.

• Sandbox — free: A shared Bauplan environment with public datasets or the option to upload your own, full CLI, SDK, and API, and community Slack support. Because everything is public, it is suited to toy projects and demos, not company data.
• Dedicated — $700 a month: The first 14 days are free. It includes a dedicated private instance, data staying in your S3, 10GB memory, unlimited queries, users, and agents, private Slack support, and SOC 2 Type II and GDPR compliance. This is the entry point for teams needing private data handling, though the 10GB memory allocation may not suit every workload.
• Enterprise — custom pricing: Includes 50GB+ memory, unlimited queries and users, dedicated customer support, a forward deployed engineer, SOC 2 Type II and GDPR compliance, and PrivateLink or BYOC. It is aimed at organizations needing more memory, deployment flexibility, or closer support; pricing is custom.

The free plan gives broad product access but not data privacy. Dedicated removes that constraint and has no stated query, user, or agent caps, while Enterprise raises the memory floor and adds deployment and support options. The stated 14-day trial applies to Dedicated; no renewal terms are given.

Platforms

Bauplan is available through API and web. Its object-storage model and integrations with notebooks and orchestration tools make it relevant to teams building workflows across those environments.

Who it's for

Bauplan is aimed at developers and AI agents building, validating, and shipping pipelines on production data. It makes particular sense for teams using Iceberg that want managed execution plus controls for safer publishing, and for organizations that need a private instance or cloud-network deployment options. It is a poor fit for private-data evaluation on the free tier, or for buyers looking primarily for a broad analytics workspace.

Pros and cons

• Pros: Branches, isolated runs, atomic publishes, and rollback address risks in production pipeline changes.
• Pros: Python and SQL run against the same Iceberg tables in customer object storage, with serverless execution and no clusters to manage.
• Pros: Dedicated and Enterprise provide private deployment choices, with PrivateLink or BYOC at the Enterprise level.
• Cons: Sandbox data is public, making the free plan unsuitable for company data.
• Cons: Private use starts at $700 a month, a substantial entry point for teams that only need occasional pipeline execution.
• Cons: Dedicated's 10GB memory may be limiting for some workloads; larger memory is reserved for custom-priced Enterprise.

Alternatives

Data Lakehouse Platforms and Data Version Control Tools offer category-wide starting points for comparing lakehouse products or version-control-focused tools.

• AetherLake is worth considering if a free, self-hosted option fits; its license names Murat Özkul as licensor and identifies AetherLake as the project.
• Amazon SageMaker Autopilot suits readers who prefer pay-as-you-go pricing with no minimum fees or upfront commitments; its Autopilot-specific rate is not given.
• Databricks Notebooks may suit teams who want a free serverless workspace with limited compute and usage, or pay-as-you-go with no upfront costs.
• LavaLake is an option for readers seeking a free, no-time-limit Core plan with up to 16 vCPU on a single machine, or a Team plan at €249 a month.
• Oracle Autonomous AI Lakehouse may fit readers who want an always-free plan for unlimited time, subject to capacity limits.
• Starburst Data Platform is an alternative for ad hoc queries on up to three clusters under its free plan; Pro priced on request.
• Starburst Galaxy also offers a free plan for ad hoc queries on up to three clusters, with Pro starting priced on request.
• Apache Hudi is a free, open-source lakehouse option for Linux and self-hosted use.

Verdict

Choose Bauplan if your team builds Python or SQL pipelines on Iceberg and wants managed execution with safeguards for production changes. Its strongest reason to buy is the combination of isolated development, controlled publishing, and private deployment choices; its clearest reason to look elsewhere is the $700 monthly starting price for private use, especially if you only need a free environment or a broader analytics workspace.