Future Present Labs is now on Hugging Face. We’re building a home there for our model work, and we’ll be releasing our first fine-tuned model later. It isn’t available for public download yet. Today is about opening the door and sharing where we’re headed.
At FPL, we work across manufacturing, computing, and software development. Those aren’t separate worlds to us. A physical part starts with an idea, becomes a design, moves through software and machines, and ends up in someone’s hands. The tools connecting those steps matter as much as any individual step.
We’re excited about what AI can make possible across that whole process.
From Better Tools to Better Work #
In manufacturing, we’re interested in AI that helps people move from intent to a clear, buildable specification: making technical knowledge easier to find, improving the handoff between design and production, and reducing the time spent translating between documents, systems, and people.
That doesn’t mean treating a model’s answer as permission to run a machine. Materials, tolerances, inspection, and human judgment still matter. The opportunity is to give skilled people better tools and more time for the work that needs their expertise.
In computing, we’re excited about systems people can understand, adapt, and run on infrastructure they control. Models become more useful when they fit the job and the hardware, rather than requiring every workflow to fit one service.
In software development, we see room for much more than faster code generation. We want better ways to investigate problems, connect existing tools, test ideas, and carry a project from an experiment into something reliable enough to use every day.
These are directions we’re working toward, not a claim that one model solves them all.
Our First Fine-Tune Is on the Way #
We’ve been working on our first fine-tune and the tooling around it. One lesson has already become clear: the weights are only part of the result. Data preparation, reference selection, inference settings, evaluation, and the actual experience of using the model all matter.
Sometimes more processing makes the result worse. Sometimes a smaller, faster setting is the better choice. And sometimes an automated check misses something a person notices immediately.
We want to share that work honestly: what we built, what helped, what didn’t, and where the limitations remain. When the model is ready for release, we’ll publish a separate announcement with the details needed to evaluate and use it.
See You There #
We’re excited for an AI future that reaches beyond a chat window: into the workshop, into the systems running our businesses, and into the software we build together.
Hugging Face gives us a place to participate alongside the people building that future. This is our first step there, with more to come.
Follow Future Present Labs on Hugging Face for the upcoming fine-tune release and future updates.