How Assemblē Protects What Matters Most 

You lived through the earliest days of AI. You may remember the horror stories of accountants at Fortune 500 companies sharing proprietary financial data with public AI models, only to discover their confidential information could be used for training. Or the stories of creative professionals finding their distinctive styles replicated within seconds after users prompted systems to generate content ‘in the style of’ their copyrighted work.

Sometimes a new frontier really is the Wild West.

While we were excited about the benefits that AI could have for our work, we knew we needed some thoughtful guardrails. With this in mind, our first all-employee communication about AI was a gentle reminder that while we dove into these new tools, our first duty to ourselves and our clients was protecting our intellectual property.

We placed protection of proprietary information—both ours and our clients’—and security in general at the top of our requirements list for the internally developed tool that became Assemblē.

Assemblē is a self-hosted application, allowing us to set it up on secure environments where we can maintain our own firewall security. User access is password protected, and a variety of user roles limit individual users to only the parts of the system that they need for their work.

Because we trust that only our authorized users will have access to any part of our system, Assemblē is able to support our highly collaborative internal process. If someone is working on a proposal and wants to share their chat outputs with someone else, they can send a link to the chat and the other logged in user will be able to read the history and enter the chat.

Our system maintains the full-text chat history of every conversation, and surfaces key facts about the user and conversation so we can intelligently supply the assistant with the level of detail it needs to answer the question. If an employee moves on to another role, we can revoke their access but keep their chat history to remember our thought process around a certain proposal.

The key difference from public AI platforms? At every step, we maintain complete control of the security barriers protecting our data – no third-party access, no training on client information, no compromises.