This is BLUF: short missives from the founder of Lovarys. We aren't merely thought leaders. We deliver value.
Don't put your book of business or client matters, money, or reputation at risk: OWN Your Intelligence.
Our entry-level offerings are the best fit for small to medium-sized businesses, provided your owners are credentialed at US Federal or State-level, such as tax and wealth advisors. Book time with our Founder (or a partner) to learn more.
We have something new in the works, and it runs on Framework Desktop machines: our Big Box (larger-SLM capable)
The target market is anyone who has feels "left out in the cold" by OpenAI Enterprise, or Anthropic's Project Glasswing. Any model YOU can't run today, on-premise isn't intelligence you own, and what you own can't go away overnight. If you've been cut-off from LLMs by usage-based billing (token limits, or policy) then you can still finish the mission.
Our founder is SAM.gov registered (CAGE code: 19NW0) and we can bid on contracts directly, or as/with a sub-contractor for AI services, hardware, or software consulting. Lovarys partners with Nvidia via the Inception program for start-ups.
The shortage (and long lead-time) for Apple's M4 Pro hardware has limited our ability to offer flagship units directly via Kickstarter to consumers. We DO continue to work with businesses for local-first inference services, and I'll announce a new offering next month for the public sector.
TBA: We are planning to launch a Kickstarter.com campaign (soon) to bring the Box to more of you!
We are seeking early beta testers for our first iPhone app, which grants exclusive access to our custom models:
Based upon `gpt-oss:120b` tuned for tax questions, on my 128 GB mac Studio (named "everest" affectionately)
All we need is: any @gmail.com address associated with Apple, so that we can invite you via TestFlight.
Interested? Reach out! Click the "Reach our Founder" button, at the top of the page.
We're at the Linux Foundation Member Summit in Napa, California today to announce our Mark Two, AI demo unit: an Apple M4 Pro compute node combined with NVIDIA peripherals for CUDA accelerators. If you're here at the event, ask around to meet our founder and see a demo of our hardware prototype.
In addition to the Mark Two, we have "bricks" (turn-key encrypted data storage for IP law / eDiscovery) and Yubikeys linked to your professional license.
If you can figure out how to get invited: we now have an easter egg on this website, hint: for Konami fans.
We're conducting some initial market research to help us expand our knowledge base. Here is a non-exhaustive list of open-source projects that will serve a key component for Lovarys in 2026:
Apple containers, similar to Docker, for secure process isolation: https://github.com/apple/container
Under the covers, our AI models are loaded from GGML with llama.cpp: https://github.com/ggml-org/llama.cpp
Ollama for OpenAI-compatible APIs (edit, Jan 2026: also supports Claude now, Anthropic's ChatGPT analogue)
Weaviate for RAG (on the box) for "discovery oriented" workflows: chat with your data, finding "needles in haystacks"
Two secure libraries for org-scoped MCP with private tools registry, tracking Arcade.dev and Daytona releases
Please register your interest and tell us how you're using tools like these at work, or in your professional life.