Human-agent collaboration
How should people delegate work, inspect progress, and intervene?
We explore mixed-initiative interaction, adjustable autonomy, and meaningful human control.
About Us
Building the interaction layer for the agentic era.
Sentia Labs is an applied HCI research lab building tools for designing and evaluating human-agent interaction.
Our mission
Help designers shape how intelligent systems behave, and how people work with them.
Software is gaining a new kind of participant. We study how people and agents work together, and we build the tools that let designers prototype, observe, and evaluate that work.
Our thesis: design now mediates between people and agents.
As agents become participants in software, design must account for people expressing intent, agents taking action, and interfaces mediating between them. We believe the next generation of design tooling must make these interactions possible to prototype, observe, and evaluate.
The problem we are solving.
A polished interface does not tell us whether an agent understands a task, acts within the right boundaries, or helps someone reach their goal. Designers need ways to test behavior across different contexts, identify interaction failures, and preserve human agency as systems become more autonomous.
Our research agenda.
Three questions organize the lab’s work. Each one is a gap in how teams design for agents today.
Agent-interface interaction
How do agents perceive interfaces, interpret their affordances, and navigate workflows?
We study interface grounding, computer use, and recovery from errors.
Behavioral prototyping and evaluation
How can designers test experiences that respond differently to different users and contexts?
We explore simulation environments and human-calibrated evaluations of usability, task success, and agent behavior.
From research to tooling.
Our ambition is to turn these findings into practical design tools and evaluation infrastructure, helping teams move from designing screens to shaping intelligent behavior.
Our team brings experience across design, engineering, AI, and enterprise software, including building products inside Fortune 500 organizations, with advisors who have led large design organizations. Today that work ships in three parts, across two products.
Digital twins grounded in first-party data
Every twin is spawned from a team’s own interviews, research, and product behavior, never a stock persona.
Behavioral simulation of product decisions
Twins act on the principles psychology already trusts, from personality traits to loss aversion and habit, and work through real prototypes and live products.
A customer behavior model that learns from every release
Each company owns one model of how its customer segments decide, built from its own data. Every forecast is measured against what customers actually did once the change ships, and the model updates from the result. One model per workspace, isolated and encrypted.
Synthetic UXR
Get started todayDigital twins of your users, spawned from your own research and behavioral data, simulate how people respond to a design, flow, or price before it ships.
Explore Synthetic UXRDesignOps
Private BetaHuman-agent collaboration in practice: agents prepare the staffing, standards, and review work behind enterprise design, and people decide.
Explore DesignOpsHelp shape how people and agents work together.
Bring us the research or design question you are working through.
Meet Sentia