Multimodal interaction
Research into systems that can work with language, speech, visual inputs, and the surrounding environment in a more natural way.
Engineering intelligent systems that extend what people can perceive, understand, and accomplish. Human-directed AI, practical autonomy, and research grounded in real-world use.

The atlatl—an ancient spear-thrower—multiplied the reach and force of a human arm. Our name reflects that principle: the most valuable technology expands the capabilities of the person using it.
Atlatl Systems is building toward AI systems that are useful, understandable, and capable of acting with context. We approach autonomy as an engineering discipline: the right intelligence, the right tools, and clear human direction.
Our work connects perception, reasoning, memory, tools, and evaluation into coherent software architectures.
Research into systems that can work with language, speech, visual inputs, and the surrounding environment in a more natural way.
Coordinating models, software tools, and multi-step workflows with explicit goals, execution boundaries, and verification.
Exploring persistent state, retrieval, and context-aware responses so an assistant can build on prior work instead of starting over.
Designing systems around observable behavior, automated tests, recoverability, and clear evidence of what actually works.
V2 is Atlatl Systems' working development platform for exploring a multimodal personal AI assistant. Its architecture brings together reasoning, memory, speech, vision, tool use, and workflow automation in a locally oriented system.
Work is ongoing. Individual subsystems and demonstrations are at different stages of implementation and validation; V2 is not presented as a released commercial product.
Research directionsWe are exploring where human-directed intelligent systems could deliver practical value. The areas below represent directions of interest, not announced products or contractual engagements.
Helping decision-makers connect reliable evidence, domain context, and repeatable workflows—particularly in complex institutional settings.
Agent systems that can carry out bounded tasks, coordinate software, surface failures, and operate under clear human oversight.
Practical architectures that emphasize control, transparency, privacy-aware design, and observable system behavior.
Development status: This page describes current engineering interests and research objectives. Capabilities, availability, and use cases will be updated as projects are validated.
We favor engineering practices that make claims testable, progress measurable, and systems easier to improve.
Translate a clear problem into a working architecture that can be exercised and challenged.
Capture real execution evidence, investigate failures, and distinguish prototypes from validated capabilities.
Use observed results to guide iteration, expand reliability, and preserve what already works.
Atlatl Systems is establishing its public presence. Research and collaboration inquiries will be supported through dedicated contact channels once company email is configured.