# Emolysis: An entire data team. One download. > Emolysis is the agentic lakehouse for data engineering and analytics. Direct a team of agents from your laptop to investigate business questions, develop dashboards, and build and test pipelines for your data platform. An entire data team. One download. Made by Bandpey GmbH. Underneath is a local-first coding agent: it picks one of ten pipeline shapes, fills a template written for your platform, proves it with golden tests on DuckDB on your machine, and only then deploys it to Databricks, with Snowflake and Microsoft Fabric planned. Status: in private development; early access by invitation; nothing is released yet. Website: https://emolysis.com/. Short index of pages: https://emolysis.com/llms.txt. ## Status Emolysis is in private development, and early access is by invitation. Nothing is released yet: there is no download, no public release and no release date. Accounts open with early access. The website describes Emolysis in the present tense, as the product it is; this status applies to all of it, and platform statuses (in development, planned) apply to each platform. The sample dashboard on the website renders the same chart specs the agent produces, on synthetic data. ## About Emolysis Emolysis is the agentic lakehouse for data engineering and analytics, a product of Bandpey GmbH in Munich, Germany. Emolysis puts a team of agents at your direction from your laptop: to investigate business questions, develop dashboards, and build and test pipelines for your data platform. Underneath is a local-first coding agent that builds data pipelines from tested templates for ten pipeline shapes, proves them with golden tests on DuckDB on your machine and deploys them to Databricks, with Snowflake and Microsoft Fabric planned, and turns a business question into a gold table and a shared dashboard. Status: in private development; nothing is released yet. About page: https://emolysis.com/about. ## Company: Bandpey GmbH Bandpey GmbH is a German data and AI services company in Munich. It works with enterprises in retail, telecommunications, energy, financial services, manufacturing and pharma, from data infrastructure and platform architecture to data science, MLOps and generative AI, and it makes Emolysis. Its co-founder and CEO is Roozbeh Bandpey. Website: https://bandpey.com. - Address: Gmunder Str. 20b, 81379 Munich, Germany. - Register: Amtsgericht München (Munich Local Court), HRB 285782. - Imprint: https://emolysis.com/legal/imprint. ## Founder: Roozbeh Bandpey Roozbeh Bandpey is an entrepreneur and AI engineer based in Munich, with more than 20 years of hands-on software engineering. He is the co-founder and CEO of Bandpey GmbH, the company that makes Emolysis, and its managing director. Bandpey is a German data and AI services company. It works with enterprises in retail, telecommunications, energy, financial services, manufacturing and pharma, on everything from data infrastructure and platform architecture to data science, MLOps and generative AI. Emolysis, the agentic lakehouse for data engineering and analytics, comes out of that work, and Roozbeh is building it. It is in private development, and nothing is released yet. His career runs through data platforms and machine learning in production. At Accenture he was a Distinguished Engineer and Head of Data and MLOps for Germany, Austria and Switzerland, and he founded the firm's Databricks Center of Excellence. Before Accenture he led machine learning deliveries at Reply, worked on natural language generation for a voice assistant at Samsung Electronics, and did research at the University of Stuttgart and at Robert Bosch GmbH. He has been a Databricks Solutions Architect Champion, an invitation-only program, since November 2021. Since September 2026 he is also the founder and CEO of FACTHORY, an AI-native operating system for industrial work. He writes about generative AI, AI agents, GPUs, AI infrastructure, cloud and distributed systems. Career, newest first: - Since 2026, FACTHORY: Founder and CEO. An AI-native operating system for industrial work, founded in September 2026. - Since 2023, Bandpey GmbH: Co-founder and CEO. A German data and AI services company in Munich, working with enterprises from data infrastructure and platform architecture to data science, MLOps and generative AI. Since June 2023. Bandpey makes Emolysis, the agentic lakehouse for data engineering and analytics. - Since 2021, Databricks: Solutions Architect Champion. An invitation-only program for solutions architects, since November 2021. - 2021 to 2023, Accenture: Distinguished Engineer, Head of Data and MLOps for Germany, Austria and Switzerland. Databricks partner lead for EMEA strategic alliances. He founded Accenture's Databricks Center of Excellence, was the first Databricks Champion in Accenture EMEA, trained and certified more than 100 professionals, and led the work behind Accenture's 2023 Databricks Partner of the Year award. Before that, as Engineering Manager for Data and AI, he led MLOps and data and ML platform work for insurance clients. - 2019 to 2021, Reply: Technical Team Lead and Lead Machine Learning Engineer. Led machine learning deliveries: MLOps, microservices on Kubernetes, and ML pipelines for autonomous driving, predictive maintenance and computer vision. - 2017 to 2019, Samsung Electronics: Technical Team Lead and Senior Software Engineer. Natural language generation for a voice assistant. - 2017, University of Stuttgart: Research assistant. Emotion analysis of tweets, combining conditional random fields with convolutional neural networks. - 2015 to 2017, Robert Bosch GmbH: Research software engineer. Research software engineering. - Earlier, Software development: Software developer and founder. Software developer roles, and an e-commerce marketplace he founded in 2013. Areas of expertise: Data engineering and data platforms; Databricks; MLOps and DataOps; Machine learning engineering; Natural language processing; Generative AI and AI agents; AI infrastructure and GPUs; Solution architecture; Cloud and distributed systems. Founder page: https://emolysis.com/about/roozbeh-bandpey. Public profile: https://www.linkedin.com/in/roozbehbandpey/. ## What Emolysis is Emolysis is the agentic lakehouse for data engineering and analytics. Direct a team of agents from your laptop to investigate business questions, develop dashboards, and build and test pipelines for your data platform. Business users want answers to the questions that matter. Data teams need the definitions, transformations and pipelines behind those answers. Emolysis brings these needs into one development workflow, with the person directing the work and reviewing the results. Our approach combines local development and testing with platform-specific pipeline templates. Production pipelines run on the customer's data platform. Databricks is the first design-partner focus; Snowflake and Microsoft Fabric are planned. Emolysis gives people the capabilities of a data engineering and analytics team through a downloadable product. Underneath is a local-first coding agent, built for data engineering leads, platform teams and analytics engineers who run production data on Databricks, Snowflake or Microsoft Fabric. Most production pipelines follow about ten known shapes, and each shape has a correct form on each platform. Emolysis picks the shape that fits your request, fills a template written for your platform, proves the result with golden tests on DuckDB on your machine, and only then deploys it as your platform's native unit. The same engine turns a business question into a gold table and a dashboard you can share. Emolysis does not replace Databricks, Snowflake or Microsoft Fabric as your production platform, or their catalogs, and it is not a hosted service that runs your production jobs. Production runs on your platform, under your account. ## How it works 1. Ask. Start from a business question, such as monthly revenue by region, or name the pipeline you need. Emolysis finds the sources that can answer it: the tables in the platform you connect, sampled to your machine. 2. Confirm the definition. You confirm what the metric means: its measures, grain and joins. The agents send your request, with the schema details they need, to the AI model provider you choose. 3. Build and test. Emolysis fills the template for your platform and runs its golden tests on DuckDB on your machine. A failing test sends it back to the template, not to production. 4. Dashboard, then deploy. You see the dashboard on your machine first. A green run deploys as your platform's native unit, such as a Databricks Asset Bundle, and publishes the dashboard when you ask for one. ## Pipeline shapes Each shape has a template per platform, with named slots, test fixtures and expected outputs (golden tests). Status: templates are written for Databricks, Snowflake and Microsoft Fabric and pass their golden tests on DuckDB; none has passed on a real platform yet, and none is released. - Ingest (batch) (https://emolysis.com/shapes/ingest-batch): Land files, extracts and database tables in bronze tables, typed and checked on load. - Incremental append (https://emolysis.com/shapes/incremental-append): Append only the rows that arrived since the last run, tracked by a watermark. - CDC merge (https://emolysis.com/shapes/cdc-merge): Apply inserts, updates and deletes from a change feed to a target table. - SCD Type 2 (https://emolysis.com/shapes/scd2): Keep the full history of a dimension, with a validity range for every version. - Snapshot (https://emolysis.com/shapes/snapshot): Copy a source in full on a schedule and keep dated copies for a set retention period. - Backfill (https://emolysis.com/shapes/backfill): Re-run a pipeline for a past date range, one partition at a time, without touching the rest. - Late-arriving (https://emolysis.com/shapes/late-arriving): Reprocess the right time windows when facts arrive after the period they belong to. - Dedupe / conform (https://emolysis.com/shapes/dedupe-conform): Turn raw bronze rows into clean silver rows: one row per key, with consistent types and names. - Aggregation / gold (https://emolysis.com/shapes/aggregation-gold): Build the metrics tables that dashboards and reports read, at a declared grain. - Reverse ETL (https://emolysis.com/shapes/reverse-etl): Push curated gold data back into business applications and operational databases, safe to re-run. ## Platforms Connecting a platform loads a platform profile: dialect, table format, merge idiom, incremental tools, deployment unit, compute, sign-in method, known traps and how to sample. Only the templates for your connected platform are offered. - Databricks (https://emolysis.com/platforms/databricks): Deploy target (in development). The first target. Writes for Delta tables and Unity Catalog and deploys as Databricks Asset Bundles. - Snowflake (https://emolysis.com/platforms/snowflake): Deploy target (planned). Writes for streams, tasks and dynamic tables, with Snowflake's own identifier rules. - Microsoft Fabric (https://emolysis.com/platforms/microsoft-fabric): Deploy target (planned). Writes for lakehouses in OneLake and deploys notebooks to your Fabric workspace. - SQL Server (https://emolysis.com/platforms/sql-server): Source (planned). Planned as a source: operational tables and change data from SQL Server, read for your pipelines. - PostgreSQL (https://emolysis.com/platforms/postgresql): Source (in development). Reads tables from PostgreSQL as a source: mirrors its catalog and samples tables into DuckDB on your machine. - DuckDB (https://emolysis.com/platforms/duckdb): Local engine (local dev loop). The local development loop: every pipeline runs and passes its tests on DuckDB before it touches your platform. ## Where you use it Emolysis is one engine with one set of commands, reached the same way from every surface. Status: none of these are released yet. - Desktop app (https://emolysis.com/integrations/desktop-app): The full engine with a workspace view, for macOS, Windows and Linux. In development. - emo CLI (https://emolysis.com/integrations/cli): Every Emolysis command from the terminal, for scripts and CI. In development. - MCP server (https://emolysis.com/integrations/mcp): Emolysis inside any editor or coding agent that supports MCP. In development. - GitHub App (https://emolysis.com/integrations/github): Checks on every pull request: lint, golden tests, cost estimate and a preview link. Planned. ## Data handling How Emolysis handles your data: - Your platform credentials stay on your machine, in its credential store. They are not sent to Emolysis. - Deployed pipelines and queries against your platform run there, under your account; local tests run on your machine. - To write code, the agent sends prompts to the AI model provider you choose. A prompt can include table and column names, schemas and sample values from your data. - A dashboard leaves your machine when you publish it, and only then. Shared dashboards have three visibility settings: anyone with the link, members of your organization (the default), or only you. You can revoke a link or set it to expire. - With your own model key, prompts go to that provider under your key and its terms. With a paid plan's included usage or usage credits (on Enterprise, metered usage), prompts and responses pass through Emolysis's metering service on the way to the model provider. That service is designed not to store or log their content; it records token counts for billing. - When you sign in, Emolysis receives your account identity from the provider you choose: GitHub, Microsoft or Google. Emolysis's services keep your account, organization, plan and usage records. They are designed never to connect to your data platform and never to receive your platform credentials. Data from your tables reaches them only in prompts that pass through the metering service, which is designed not to store or log them, and in dashboards you publish. - The GitHub App is planned. If you install it, it reads the contents of the repositories you choose, to run its checks, and writes check results and comments on their pull requests. - Product telemetry is documented before the first release. The website loads Google Analytics and Microsoft Clarity, and sets their cookies, once a visitor chooses in its cookie banner, whose every choice includes them; the signed-in account and shared dashboards never load them. - Emolysis has not been through a third-party security audit and holds no compliance certifications. Details: https://emolysis.com/security. ## Plans Plans and prices are planned; nothing can be bought yet. Every plan accepts your own model key from the model provider you choose. Requests made with it are billed by your provider and do not count against your plan's limits. - Free: bring your own model key, billed by your provider; the local engine and DuckDB, one platform connection, the core templates and the MCP server. - Pro: included model usage, every platform connection, the full template library, web dashboards and sharing, and the planned GitHub App with pull request checks. - Max: 5x or 20x Pro's included usage, priority at busy times, and early access to new templates and connectors. - Team: per seat, with included model usage for each seat, a shared catalog, an organization workspace, audit export and single sign-on. - Enterprise: no included allowance; model usage at standard metered rates within spend limits that admins set, or an organization key; support terms, and options for running Emolysis in your own environment (planned). Planned prices in US dollars: https://emolysis.com/pricing. ## Company and contact Emolysis is made by Bandpey GmbH, Munich, Germany (https://bandpey.com). - Email: hello@emolysis.com (early access, security reports, press and general questions). - GitHub: https://github.com/emolysis (public repositories). - Imprint: https://emolysis.com/legal/imprint. - News and release announcements: https://emolysis.com/news. --- Page: https://emolysis.com/. Emolysis is a product of Bandpey GmbH, Munich, Germany. It is in private development, and nothing is released yet.