FOMER AI turns capable models into dependable products: assistants grounded in your own knowledge, automation with people in the loop, and integrations that fit the systems you already run.
AI products we build ourselves, each designed to do one job well.
RAG assistant · our own product · 2026
An assistant that answers questions over customer feedback, sales call notes and documents, and shows which document each answer came from.
Outfit recommendation app · 2025
An AI-assisted app that suggests what to wear based on your wardrobe, the weather and the event you’re going to.
Where AI earns its place in a product or a process, and how we make it reliable there.
New products where AI is the core of the experience, designed around what models do well.
Workflows that take repetitive work off your team, with people kept in the loop where it matters.
Search, recommendations and decision support built on your own data, with measurable quality.
Connecting leading models to the software you already run, securely and within budget.
Early ideas tested quickly in the open. The good ones become products.
Focused work on continual learning and reinforcement learning. The code is open.
A study of a model that learns new tasks without forgetting old ones: a multi-level memory and replay buffer, tested with ResNet18 and ViT on CIFAR-10/100.
Continual learning
A driving agent trained with PPO in the highway-env simulator that learns to change lanes and move safely through traffic.
Reinforcement learning
Fast memory, replay, LwF distillation, EWC and BiC compared side by side in one PyTorch setup.
Continual learning
Model-agnostic by design. We choose the model per task and measure it before it reaches your users.
Tell us about the process or product you have in mind. We will help you find where AI makes a real difference, and where it does not.
