SuperAPI AI Directory
Partner with us中文AI newsFavorites

AI agents

2,964 results · Page 47 of 83

Formant

AI agents

Formant is an operating-layer platform for physical AI that connects robots, operators, enterprise systems, and AI agents through a governed layer, providing monitoring, intervention, evidence retention, and orchestration capabilities. The platform has been running in production for nearly a decade, covering 644 production organizations, and supports scenarios such as remote inspection, centralized expert support, robot fleet observability, remote support, global service operations, teleoperation, remote diagnostics, and distributed systems. Formant uses a software-plus-services model, with forward-deployed engineers assisting with implementation to help robot deployments succeed in real-world environments. However, the materials do not provide specific pricing, charging models, supported robot brands, or detailed usage steps.

Read more

friday

AI agents

Friday is an AI-powered application generation tool, positioned as a developer assistant that can help users quickly create complete Node.js projects through unlimited prompts. It provides one core prompt for setting up the application foundation and allows adding any number of sections, each corresponding to a prompt and representing a specific part of the application. Friday uses GPT-4 to provide AI assistance and has been tested and optimized for GPT-4-32k to achieve better speed and results. Generating the application foundation requires 2 small requests, and each user-provided section requires 1 request. Friday uses esbuild in the background to process each created application. The project uses the MIT license.

Read more

future-os

AI agents

FutureOS is an open-source AI agent project, primarily written in Rust, designed to let the same agent work collaboratively across terminals, desktops, mobile devices, and chat applications. It adopts a local-first architecture, provides configurable file access approval and operating system sandboxing, and supports driving terminal UI, desktop applications, mobile devices, CLI, and IM bots through a gRPC backend, sharing sessions, memory, and skills. The built-in loop control plane supports long-running (e.g., 24 hours or more) persistent goals, event-sourced state, and verification gates. It can be deployed via a one-line installation script, supports macOS, Linux, and Windows, and after installation runs future init for initialization. In terms of models, it supports FutureOS-hosted models and custom providers, API keys are stored in ~/.future/agent/auth.json, and custom models can be configured in ~/.future/agent/models.json. Optional skill packs can also be installed, loaded from ~/.future/agent/skills/. Terminal and CLI clients connect via local IPC, and a TCP mode is also optional (expose with caution).

Read more

GAAI-framework

AI agents

GAAI (Governed Agentic AI Infrastructure) is a .gaai/ folder that can be placed into any Git project, using Markdown, YAML, and bash to turn AI programming tools into a reliable agentic software delivery system. Through its dual-track mechanism of Discovery (defining what to build) and Delivery (autonomously executing until acceptance criteria pass), it adds a governance layer to AI programming agents, preventing scope drift and context pollution. No SDK, no packages to install, no external services, and compatible with tools such as Claude Code, Codex CLI, Gemini CLI, and Cursor.

Read more

GeniA

AI agents

GeniA is an open-source AI platform engineering team member, built on the function-calling capabilities of OpenAI or Azure, and can be integrated into a team's Slack channel to perform tasks on behalf of engineers in production environments. It can learn and use custom tools to help with engineering tasks such as deployment, troubleshooting, log summarization, build initiation, PR digestion, cloud resource optimization, security vulnerability checks, and service scaling, aiming to reduce MTTR, reduce boilerplate work, and reduce cross-team dependencies. GeniA requires an OpenAI API Key, uses the Apache 2.0 license, and is 100% open source.

Read more

GenoMAS

AI agents

GenoMAS is a minimalist multi-agent framework for automating scientific analysis workflows, with an official implementation for code-driven gene expression data analysis. It provides a typed message communication protocol and a notebook-style workflow, enabling agents to plan, write code, execute, debug, and backtrack multi-step tasks, and supports customizing agent roles, guidelines, tools, and action units for different domains. Its specialized implementation can analyze transcriptomic data from GEO and TCGA, identify genes significantly associated with traits while handling confounding factors, achieve a 60.38% F1 score on the GenoTEX benchmark, and discover biologically meaningful gene-trait associations. The project requires a Python 3.10 environment, supports multiple LLM providers, allows role-specific model configurations, and provides documentation on data preparation, environment setup, experiment execution, output structure, and troubleshooting.

Read more