# Introducing Emolysis > We are building Emolysis, a local-first coding agent for data engineering and BI, starting with Databricks. What it will do and where it stands. _Update, 2026-10-23: Snowflake and Microsoft Fabric have connectors and templates, tested only on simulated services; their status, and the status of each source, is on the [platform pages](https://emolysis.com/platforms). The [GitHub App](https://emolysis.com/integrations/github) is planned. Pages now describe Emolysis in the present tense and mark what is built with a status label; this post keeps its original words._ _Update, 2026-10-18: we now describe Emolysis as the agentic lakehouse for data engineering and analytics; see [About](https://emolysis.com/about)._ Today we are introducing Emolysis, a coding agent for data engineering and BI. It is made by Bandpey GmbH in Munich, Germany, and it is in private development. Nothing is released yet. This post explains what we are building, why, and how we will talk about it while we build it. ## The problem we keep seeing Coding agents have become good at general software. Data engineering is a harder fit than it looks. A pipeline that reads well can still be wrong in ways that only show up on the platform: a merge that fails when a key appears twice in a batch, a slowly changing dimension with two current rows, a backfill that overwrites the wrong partitions, an identifier that Snowflake folds to uppercase. Each platform has its own dialect, table format, merge idiom and deployment unit. A general agent with a connection to your warehouse has to guess at all of that, and you find out on the platform whether it guessed right. At the same time, data engineering is more repetitive than general software. Most production pipelines follow a small number of shapes: landing files, appending incrementally, merging change feeds, keeping history, taking snapshots, backfilling, handling late data, cleaning into silver, building gold metrics and pushing results back to applications. Each shape has a known correct form on each platform. That is the opening we are building for. ## What Emolysis is designed to do Emolysis will work from three things. First, a platform profile. When you connect a workspace, Emolysis will load a structured description of that platform: its dialect, table format, merge idiom, incremental tools, deployment unit, sign-in method and known traps. The agent will write Databricks code for Databricks because it reads that profile, not because a prompt asked nicely. See [what a platform profile is](https://emolysis.com/faq/what-is-a-platform-profile). Second, templates for [ten pipeline shapes](https://emolysis.com/shapes). For each shape and platform there will be a template with typed slots, test fixtures and expected outputs. The agent will pick the shape, ask only for the slots it is missing, and fill the template instead of writing the pipeline from scratch. Third, golden tests on your machine. Before anything reaches your platform, the template's tests will run on [DuckDB](https://emolysis.com/platforms/duckdb) on your machine, against fixed inputs built to include the hard cases. A failing test will send the agent back to the template. Only a green run will deploy, and it will deploy as your platform's native unit, such as a Databricks Asset Bundle. The same engine will turn a business question into a tested gold table and a [dashboard you can share](https://emolysis.com/bi) with a link. ## Where you will use it Emolysis is planned as a [desktop app](https://emolysis.com/integrations/desktop-app), a command line called `emo`, and an [MCP server](https://emolysis.com/integrations/mcp) so you will be able to use it from any editor or coding agent that supports MCP. A [GitHub App](https://emolysis.com/integrations/github) will run lint, golden tests and a cost estimate on every pull request. The Free plan will be bring your own key. You will add a key from the model provider you choose, that provider will bill you directly at its own rates, and Emolysis will charge nothing. Pro, Max and Team will include model usage, so you will not need a key. The [pricing page](https://emolysis.com/pricing) shows the planned plans. ## Where it stands [Databricks](https://emolysis.com/platforms/databricks) is first, and it is in development. [Snowflake](https://emolysis.com/platforms/snowflake) and [Microsoft Fabric](https://emolysis.com/platforms/microsoft-fabric) are planned after it. SQL Server and PostgreSQL are planned as sources. The desktop app, the CLI and the MCP server are in development. There is no download, no sign-up and no release date yet. The sample dashboard on this site runs on synthetic data. ## How we will talk about it We will describe Emolysis in the future tense until each part ships, and we will say plainly what is missing. We will not publish numbers we cannot back, quotes from people who have not used it, or logos of companies that are not customers. We will also be exact about data. Emolysis is designed to be local-first: development will happen on your machine and production will run on your platform. That does not mean nothing leaves your machine. To write code, the agent will send prompts to the AI model provider you choose, and a prompt can include table and column names, schemas and sample values from your data. With your own model key, prompts will 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 will pass through Emolysis's metering service on the way to the model provider; that service is designed not to store or log their content. Dashboards will leave your machine when you publish them, and only then. The full list is on our [security page](https://emolysis.com/security), and it will stay there. ## What comes next The next milestone is the Databricks loop end to end: connect a workspace, ask a business question, and get a tested gold pipeline plus a shared dashboard. A small group of early users will try it on real work before anything is public; the [early access page](https://emolysis.com/early-access) explains who we are looking for. We will post here when that changes, and when each platform and plan becomes available. You can follow this page with the [RSS feed](https://emolysis.com/news/rss.xml). > Emolysis is in private development. Nothing described here is released yet. --- Page: https://emolysis.com/news/introducing-emolysis. Emolysis is a product of Bandpey GmbH, Munich, Germany. It is in private development, and nothing is released yet.