Assembly has been designed with every consideration for businesses, making data privacy and security central pillars of our product. If you have any questions, please contact support@askassembly.app
Assembly search can be deployed on premise (within a company’s network) or as an external cloud-based service. As such, we have detailed policies for both versions.
If a company chooses to use Assembly Search on the cloud (Assembly servers) through our web application or VS Code extension, Assembly will store code, which is used to improve the precision of our search results. The code itself is encrypted prior to being stored in a standard SQL database. We maintain this code in our database while you are signed into the Assembly web application or VS Code extension. If you choose to logout of either application, we will immediately delete all data related to your company’s usage including code, queries, logs, etc.
Note that if a company opts to use Assembly on-prem (on a company’s network), any data indexed by our system will never leave the company’s servers.
In terms of architecture, Assembly integrates with a company’s codebase, parses through its body, and creates vector representations of every single chunk of code. Applying this flow to our cloud-based version, we will store encrypted chunks of code on a SQL database, the vectorized representations of the code on a vector database, and use an LLM to generate summaries of code.
For our tool, we currently use the following database services and LLMs:
Assembly Search uses the Pinecone vector database to store embeddings of data. You may refer to the following for its policies
For its database, Assembly utilizes Supabase. You may refer to the following for its policies
In order to generate embeddings and summaries of code, Assembly Search uses the OpenAI embedding model and LLM. You may refer to the following for its policies
For its database, Assembly utilizes Supabase. You may refer to the following for its policies
If a company would like to deploy Assembly on premise, we will integrate with existing infrastructure used to host our service including SQL databases, vector databases, as well as self-hosted LLM’s (if applicable).
If Assembly is deployed on-premise, Assembly support will not have access to any data including code, logs, queries, that are usage related for a company. No data will leave said company servers.