MZLA Technologies Corporation, the Mozilla Foundation subsidiary behind Thunderbird, announced Thunderbolt, an open source, self-hostable AI client for organizations that want to run AI on their own infrastructure.
This project is funded by investment from Mozilla and standalone productseparate from Thunderbird, was built by another team within MZLA focused on enterprise AI products.
Provided below Mozilla Public License 2.0Thunderbolt provides an AI workspace where users can interact with AI through chat, search, and research, connect to corporate data, and choose models and tools that fit their needs.
runs natively Linux, window, macOS, iOSand androida web client will also be available.
Points to note are…
you should know that Thunderbolt ships with telemetry turned on by default.
According to the project’s telemetry documentation on GitHub, PostHog is used to collect usage data, including chat activity, model selection, settings changes, and location information.
This can be turned off in settings, and the project states that no personally identifiable information (PII) is collected without your explicit consent.
Who is it for?
of Target audience This may be an organization with strict data residency or compliance requirements. So consider healthcare providers, law firms, and financial institutions that can’t afford to handle sensitive internal data flowing through third-party AI services.
In terms of competition, Thunderbolt is a direct challenge to Microsoft Copilot, ChatGPT Enterprise, and Claude Enterprise. In the open source space, it ranks alongside tools like Open WebUI and LibreChat, both of which offer self-hosted AI front ends.
In announcing Thunderbolt, Ryan Sipes, CEO of MZLA Technologies Corporation, added:
AI is too important to outsource. With Thunderbolt, we give organizations a sovereign AI client, allowing them to decide how AI fits into their workflows based on their infrastructure, data, and conditions.
What can you expect?
lightning strike Connect to Frontier models From Anthropic, OpenAI, and Mistral, handle local inference through Ollama, and accept custom providers using workspace offerings. chat and search mode.
We can also accommodate scheduled workcompile briefings, track topics over time, and initiate actions when set conditions are met.
deepset’s Haystack integration ties clients to enterprise agents and RAG pipelines in the same architecture, while support for MCP (Model Context Protocol) is in preview and ACP (Agent Client Protocol) is in active development with a target date of April 2026.
How do I get it?
Visit Thunderbolt.io to get started with Thunderbolt. If your organization is interested in enterprise deployment, professional support, or custom development, please contact our team.
The source code can be found on GitHub.
Besides that, the FAQ states: Thunderbolt version for regular users There are plans, but no release date has been decided yet.
