Simulan

SAPI is Simular's API and command-line tooling for programmatically controlling its cloud computers. The video describes it as exposing the capabilities available in Simular's web portal through terminal commands and an API, so developers can integrate cloud-computer control into their own products and workflows.

Simile is a company and simulation platform that builds behavioral foundation models and synthetic "digital twins" of people to simulate populations and human decision‑making. The platform accepts scenario specifications, surveys and experimental designs as inputs and produces simulated populations, individual-level decision trajectories and aggregated outcome statistics that can be queried for policy analysis, product testing, forecasting and similar use cases. Simile is published from the vendor website and presents itself as a simulation platform for human behavior.

Semwright is an open-source semantic layer and broker that gives AI agents typed interfaces to real software. Agents discover application capabilities, resolve named objects, and request typed operations with explicit inputs; policy checks authority and scope before execution, while references, jobs, readback, and execution evidence make object identity and results inspectable. It is designed to replace pixel-based interaction with application-native operations and supports Figma, Chromium, Blender, and MLT video through drivers. The project exposes the broker through MCP, a CLI, and an SDK, and can be installed on Linux, macOS, and Windows. The page states that the initial release starts with read-only permissions and that its independent security review is incomplete.

Agent S is an open-source computer-use agent framework from Simular. It accepts natural-language tasks, observes a real desktop through screen and accessibility signals, and operates ordinary applications by clicking, typing, scrolling, and executing optional local code rather than relying on per-application APIs or scripts. The project has progressed through three generations: accessibility-tree grounding with an LLM, a vision-based manager-worker architecture using multiple foundation models, and Agent S3, which scales computer-use attempts through parallel rollouts and selection. Its workflow can combine model-driven exploration with reusable automation code, replay saved skills, and recover when the interface changes. The framework supports macOS, Windows, and Linux, works with models from OpenAI, Anthropic, Gemini, OpenRouter, vLLM, and open-weight providers, and is distributed through the `gui-agents` Python package. An optional local coding environment can execute Python and Bash with the current user's permissions, so the repository warns that this capability should be used only in trusted environments.

Simo is a System 1.5 judgment model from Temprl Labs. It accepts typed questions about text, screenshots, or video and returns calibrated probability distributions and the values needed for software to act. Applications can use numeric thresholds to branch on its results instead of parsing generated text.