Bauplan
San Francisco, US · Founded 2022 · Delaware corporation · 14 employees on LinkedIn · 3 known investors
Bauplan is an execution layer that lets AI agents and engineers build, validate, and ship data pipelines on production data using isolated branches, atomic publishes, and rollback. Pipelines are written as versioned Python and SQL code with quality checks, following a branch, run, inspect, and merge workflow.
Also known as Bauplan Inc. · Bauplan Labs
Founders & leadership
Bauplan was founded in 2022 by Jacopo Tagliabue, Mattia Pavoni, and Ciro Greco.
Investors · 3
Also in the syndicate · 2
Reported raises · per SEC filings
Form D private placements$4.1M disclosed across 2 of 4 rounds · 2022–2025
▶$600KofferedAug 2025 · Other TechnologyRule 506(b)
- Jacopo TagliabueExecutive Officer, Director
- Ciro GrecoExecutive Officer, Director
- Mattia PavoniExecutive Officer, Director
- Offering amount
- $600K
- Incorporated
- Corporation, Delaware, 2022
- Federal exemptions
- 06b
▶$100KraisedMar 2025 · 1 investors · Other TechnologyRule 506(b)
- Jacopo TagliabueExecutive Officer, Director
- Mattia PavoniExecutive Officer, Director
- Ciro GrecoExecutive Officer, Director
- Offering amount
- $3M
- Amount sold
- $100K
- First sale
- Mar 2025
- Incorporated
- Corporation, Delaware, 2022
- Federal exemptions
- 06b
▶$4MraisedJan 2023 · 2 investors · Other TechnologyRule 506(b)
- Mattia PavoniDirector
- Jacopo TagliabueDirector
- Ciro GrecoDirector
- Offering amount
- $4.5M
- Amount sold
- $4M
- First sale
- Dec 2022
- Incorporated
- Corporation, Delaware, 2022
- Federal exemptions
- 06b
Source: SEC EDGAR Form D. Amounts as filed; amended filings shown once at their latest values.
Company profile
researched Aug 2026Bauplan builds a serverless data execution platform that runs Python and SQL pipelines directly on Apache Iceberg tables held in the customer's own object storage, without servers or clusters to manage. The platform models data state as branches and commits, so changes can be made on an isolated branch, validated, published atomically to main, and rolled back if they fail. Pipelines, models, tables, environments and quality checks are all defined as code, versioned in the user's repository and driven through a CLI and Python SDK rather than a GUI.
The product is positioned as an execution layer for both engineers and AI agents working on production data, providing isolation, transactional guarantees and rollback so that AI-generated changes can be run against real data before publication. A consistent control loop — branch, run, inspect, merge — is exposed to humans and agents alike, with an MCP server and "Skills" for agent-driven automation. Stated use cases include building and maintaining data pipelines, safe table ingestion, diagnosing and fixing pipeline failures by replaying runs against the state that produced them, and data exploration and profiling across isolated branches.
Bauplan integrates with adjacent tooling rather than replacing it: ingestion and ELT tools (Fivetran, Estuary), warehouses and query engines reading Iceberg (BigQuery, Snowflake), orchestrators (Airflow, Dagster, Prefect, Temporal, Orchestra, DBOS), and analysis or app layers (Jupyter, marimo, Streamlit, Metabase).
Founding story
Bauplan was founded by Ciro Greco, Jacopo Tagliabue and Mattia Pavoni, serial entrepreneurs who, according to the company, went from early stage to acquisition to IPO in under five years. The founders spent roughly a decade building data and AI systems at scale and started the company on the view that data pipelines increasingly need to be built and operated like software — versioned, testable and deployable — as AI applications, agents and open table formats such as Iceberg reshape data tooling. The name comes from evolutionary biology, where a Bauplan denotes the structural or ground plan shared by groups of organisms, chosen to reflect an interest in the structural optimization of complex systems.
Business model
Bauplan sells a managed, serverless data platform that executes on data stored in the customer's own object storage; a third-party profile describes a subscription model with tiered pricing plans that scale with usage. Prospective users can sign up or book a demo through the website, and the company has worked with enterprise design partners.
Subscription-based access to the platform, with tiered pricing plans according to organizational needs, per a third-party company profile.
Traction
Bauplan reported use by enterprise design partners at the time of its seed round, including the European broadcaster MFE-MediaForEurope. Published case studies include Mediaset, which replaced Airflow and EMR with Temporal and Bauplan and reduced dashboard refresh times from 60 minutes to 5 minutes, and Trust & Will, described as a lakehouse migration to Iceberg. The company also cites a user that built a threat detection system in which a generative agent calls Bauplan's APIs to generate, validate and execute analytical queries without humans in the loop.
Latest developments
Recent output includes Bauplan Skills for safe agent automation on customer data, work on trustworthy AI in the agentic lakehouse covering concurrency and governance, evaluation and optimization of data engineering workflows, an Estuary integration for CDC into Iceberg with branches, guidance on giving coding agents branch-scoped access to production data, and a workflow in which Bauplan, Claude and Linear turn stakeholder questions into reviewed pipelines. The company also presented at AI Council 2026 on Git-for-everything, semantic layers and open source.
▸Full profile — market position, technology, go-to-market, geography, history, risks & controversies
Market position
Bauplan operates in the data lakehouse and data engineering infrastructure market, competing with and integrating around orchestration tools and warehouse-centric platforms. It is an early-stage, seed-funded company backed by Innovation Endeavors and South Park Commons, with an emphasis on open table formats (Iceberg) and agent-oriented workflows.
The company positions Bauplan as a lakehouse execution layer rather than an orchestrator: scheduling, retries and task ordering remain with tools such as Airflow or Dagster, which call Bauplan, while Bauplan supplies fast pipeline execution, Git-style data branching, atomic publishing and rollback. Its design principle is a small set of primitives already familiar to developers — functions, packages, tables, commits and branches — replacing the split between separate runtimes and interfaces for interactive exploration, pipeline development, and orchestration at scale. Being fully code-native, with no GUI or hidden state, is presented as the property that also makes the platform usable directly by LLM agents.
Technology
The platform runs pipelines as serverless functions in a managed runtime optimized for large-scale data processing, with native abstractions for tabular formats including I/O, caching, projections and filter pushdowns. Data is stored as Apache Iceberg tables on S3, providing schema evolution, partitioning and time travel, and every run and artifact is tracked through a Git-style commit log for reproducibility and rollback. Pipelines are ordinary Python and SQL functions with declarative environment specifications (Python version, pip dependencies) and expectations used as quality gates before merges. The query engine was moved from DuckDB to Apache DataFusion for an Arrow-native architecture, and the company has published research on zero-copy Arrow pipelines, FaaS scheduling simulation, and formal semantics for branching, merging and rollback.
Go-to-market
Direct sales motion via demo requests and self-serve sign-up on the website, supported by developer documentation, CLI/SDK quick starts, video walkthroughs and a YouTube channel. The company publishes a steady stream of engineering, research and case-study content, contributes papers to academic venues such as VLDB, AAAI and IEEE BigData, and partners with ecosystem vendors (for example Estuary for CDC and Temporal for workflow orchestration).
Data engineers and data teams that want to ship production data changes without operating a heavy data platform, plus the AI agents they deploy to build, test and fix pipelines. A funding report describes the target as infrastructure and data science teams in medium-sized and large enterprises with data-intensive machine learning and AI use cases, particularly in B2B software, media, financial services and healthcare technology.
Geography
Headquartered in San Francisco (595 Market St, Floor 10, San Francisco, CA 94105) with a second location in Queens, New York (28-07 Jackson Avenue). Coverage describes the company as San Francisco- and New York-based, and a third-party profile notes products available primarily in North America and Europe with European expansion plans.
History
The company introduced its serverless data platform publicly and announced a $7.5M seed round led by Innovation Endeavors on April 16, 2025. Over the following period it published case studies (Mediaset, Trust & Will), shipped an MCP server in August 2025 as a step toward an "agentic lakehouse," migrated its query engine from DuckDB to DataFusion in November 2025, and subsequently released Skills for safe agent automation alongside integrations such as Estuary CDC. Alongside product work, the team published research on Git-for-data formal semantics, zero-copy Arrow pipelines, FaaS scheduling simulation, and safe untrusted agents, including a GitLake paper accepted at the DASHSys workshop at VLDB 2026 with Columbia University, the University of Wisconsin-Madison and Carnegie Mellon University.
Risks & controversies
Third-party database entries differ from the company's own materials on basic facts — for example, one profile lists a 2023 establishment date and names Ihab Ilyas as a co-founder, while the company's site and funding coverage identify Ciro Greco, Jacopo Tagliabue and Mattia Pavoni as the founders and list Ihab Ilyas among investors. Executive titles also vary across sources, with the about page listing Jacopo Tagliabue as CTO and Mattia Pavoni as COO.
Compiled by commissioned research from 8 cited public sources — announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed — not independently audited.
Competitors · 3
by search overlap
Companies competing with Bauplan for the same Google search keywords, organic and paid, via search-intersection analysis.
Timeline · 8
launches, deals, and filingsBauplan published a preprint of GitLake, a Git-for-data system for the agentic lakehouse, accepted at the DASHSys workshop at VLDB 2026 in Boston, in collaboration with Columbia University, University of Wisconsin-Madison and Carnegie Mellon University.
Bauplan published an integration with Estuary for building change-data-capture pipelines into an Iceberg lakehouse using branches.
Bauplan introduced Skills, described as safe automation for AI agents working on customer data.
Bauplan researchers had a paper on safe, untrusted agents for big data infrastructures accepted at the S2AI workshop at IEEE BigData 2025.
Engineering post describing Bauplan's move from DuckDB to Apache DataFusion, citing Arrow-native design and the DataFusion community.
Bauplan announced an MCP Server, described as a first step toward an agentic lakehouse.
Bauplan raised $7.5M in seed funding led by Innovation Endeavors, with participation from Wes McKinney, Aditya Agarwal and Chris Re. The company said it would use the funds to expand operations and development efforts.
$7.5M source ↗
Bauplan published an introductory post describing a serverless data platform that treats pipelines, models and tables like software: versioned, testable and usable by agents, with pipelines written as Python functions running on Iceberg tables in object storage.
Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.
Legal entities · 1
corporate structureIn the news
▸Research sources · 8
primary sources listed
- Bauplanbauplanlabs.com · web
8 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Bauplan do?
- Bauplan is a serverless Git-for-data lakehouse that runs Python and SQL pipelines on Apache Iceberg tables in object storage.
- Who founded Bauplan?
- Bauplan was founded by Jacopo Tagliabue, Mattia Pavoni, Ciro Greco in 2022.
- Who are Bauplan's investors?
- Bauplan's investors include Innovation Endeavors.
- How much funding has Bauplan raised?
- Bauplan has disclosed $4.1M raised across 2 of its 4 known rounds.
- Where is Bauplan headquartered?
- Bauplan is headquartered in San Francisco, US.