Research
Inside χ-Bench’s healthcare workflows
A benchmark for complete healthcare workflows across providers and health plans.
Read the storyNews & Blog
Research, engineering, and ideas from Actava.ai.
Research
How Cura 1T learns through evaluation
How evaluation failures guide training data for a healthcare foundation model.
Read the storyProduct
Training specialized models with Model Factory
How verified workflow experience supports training and checkpoint selection.
Read the storyResearch
29 articles
Research2 min read
How GeoGNN preserves the geometry of text embeddings
Geodesic message passing uses the geometry of text embeddings to preserve meaning as neighboring nodes exchange information.
Research2 min read
A shared Slack workspace for collaborating agents
Slack channels, messages, and agent descriptions form a communication layer for human–agent and agent–agent work.
Research2 min read
Inside Enterprise Deep Research’s steerable investigation
A planning agent coordinates specialized search, enterprise tools, visualization, and reflection while a research session develops.
Research2 min read
Learning a cost-aware model policy with xRouter
A reinforcement-learning router chooses model calls using a reward that combines task success with inference cost.
Research2 min read
Organizing generated reasoning tools with ToolLibGen
Semantic clustering and code-review agents consolidate task-specific tools into interfaces that support retrieval and reuse.
Research2 min read
Building verifiable reinforcement-learning tasks with Webscale-RL
Domain-aware generation and verification turn pretraining documents into question-and-answer tasks for reinforcement learning.
Research1 min read
CoDA’s training recipe for a compact diffusion coder
A staged training pipeline develops a 1.7B-parameter diffusion coder with bidirectional context and confidence-guided sampling.
Research2 min read
Where UserRL assigns credit in a conversation
Standardized user-interaction environments support studies of reward design, training initialization, and transfer across user simulators.
Research1 min read
What UserBench measures about clarification
A controlled travel-planning environment tests whether an agent uncovers hidden preferences and selects suitable options.
See what ACTAVA can do for you
Bring one workflow your team still runs by hand. We'll show you the agent that runs it.