Research
We don't use external models or APIs. Our research focuses on building truly open workflow AI from the ground up—understanding and automating human work through observation and learning.
Research Focus
The core technologies powering Brend
Multimodal Video Ingestion
Our core differentiator: an AI engine that learns workflows by observing screen recordings. Brend watches how you navigate tools, fill forms, and complete tasks—then replicates that work at scale.
Workflow Understanding
Going beyond simple automation to truly understand the intent and context of human work. Brend learns not just what you do, but why you do it.
Multi-Workspace Orchestration
Executing complex queries and tasks across disparate data sources and applications simultaneously. Breaking down silos between your tools.
Voice AI
Natural language voice interactions for customer support, intake processes, and communication. Human-like understanding with machine efficiency.
Principles
Observation over instruction
Learn from watching, not from explicit programming
Understanding over mimicry
Grasp intent and context, not just surface actions
Openness over black boxes
Transparent models and reproducible results
Our Technology
Built from the ground up, owned entirely in-house
brend-assistant-v3
Our AI model, trained in-house by our research team. No external APIs, no black boxes—full control over capabilities and behavior.
Training Data
Accumulated and curated video, workflow recordings, and general purpose knowledge data for multimodal understanding.
buldak-server
Our own inference infrastructure, maintained and developed in-house for optimal performance and reliability.
Our People
Researchers from Stanford, Caltech, and University of Washington. Industry experience from Tesla, Dell, Pinterest, and Goldman Sachs.
Stanford University
Research
Caltech
Research
University of Washington
Research
Tesla
Industry
Dell
Industry
Industry
Interested in our research?
We're always looking for talented researchers and engineers to join our team.