factorai is a local agentic development environment for supervising Claude Code and Codex CLI sessions across projects. It provides per-agent terminals, scheduled routines, transcript search, file inspection and Git changes and graph views. Project grouping and a separate project shell help developers follow several active tasks while reviewing the resulting commits.
Filament Agentic Chatbot adds configurable AI agents to a Laravel/Filament application. Agents answer from approved files and web sources, read scoped application data and call selected APIs or MCP tools. Optional Playbooks define multistep workflows, while proposed database writes require visitor confirmation and staff can take over in an inbox.
ellô-wp connects compatible AI clients to authorized WordPress and Elementor sites through MCP. It builds editable native Elementor pages, updates text and templates and manages selected WooCommerce, media and plugin settings. Draft copies, role-based permissions and write logs support inspecting changes before publication.
Chunk is a macOS time-blocking app with an optional local MCP server for planning through Claude Desktop. Tasks, recurring templates and calendar events share a timeline; Claude can read the schedule and create, edit or remove items while the app synchronizes supported calendars.
Brethof-brain adds project memory, standing rules and searchable session history to compatible AI agent harnesses. A client supplies relevant context to later sessions, while curated records and a decision graph capture what changed. Local and encrypted-hosted memory editions are documented.
BCD connects approved Macs to an AI assistant through MCP so it can read permitted files, run commands and inspect results remotely. Users can work from another device while the Mac performs the task. Device-specific permission controls distinguish read access from writes, process execution, UI and clipboard actions.
Awakado is a macOS menu-bar utility built for AI coding-agent sessions. It follows supported agents through hooks or process activity, keeps the computer awake while work is running and reports waiting, stalled and completed states. Pro adds lid-closed operation and waking for a Claude Code usage-limit reset.
Auto-Allow is a macOS utility that clicks configured approval buttons in selected AI applications through Accessibility. It uses exact button labels and per-app custom phrases, a separate ghost pointer, local dry-run history, a pause shortcut and timed sessions. It supports AI workflows rather than generating model output itself.
NotFair connects AI clients to advertising, search, analytics and selected business platforms. An assistant can inspect live campaign data and prepare keyword, bid, budget or content changes, while the connector stages each write for approval and logs its result. Supported platforms include Google, Meta, X, LinkedIn, Reddit and TikTok Ads.
Defang Station provides an always-on cloud workspace for Claude Code and Codex signed in with your existing accounts. Agents keep files and context between sessions, publish previews at a reachable URL and can respond to configured signed webhooks. Defang Deploy provides a separate handoff into your own cloud account.
Voyp makes phone calls from a written brief, such as booking a table or rescheduling an appointment. Its assistant introduces itself as AI, conducts the conversation and returns a summary and transcript. Users can rehearse a conversation by chat, supply missing details during a call or transfer it to their verified phone number.
A4F deploys agentic AI systems for companies to automate sales around the clock, personalize customer experiences, and govern operations without losing control. The Spanish marketing site positions it as scalable enterprise agent infrastructure.
Aampe sells agentic infrastructure that learns customer behavior to automate personalized marketing treatments and experimentation without manual modeling. Teams use it to orchestrate actions across data agents, treatments, and internal workflows for lifecycle and CRM programs.
Ability.ai is an applied AI lab building sovereign agentic systems for production operations. It offers managed agent fleets, the open-source Trinity runtime and Cornelius cognitive core, and partner programs for agencies, with agents for outbound, support, content, recruiting, research, and back-office workflows.
Abstra automates end-to-end finance workflows such as accounts payable, reconciliation, and tax processing by connecting ERPs, banks, and document portals with AI-driven execution. Teams describe processes in natural language while Abstra builds controlled flows with audit trails and human exception handling.
AC0 builds practical AI assistants that plug into existing business tools such as Intercom, Aircall, SMS, and order systems to automate follow-ups and handoffs. The public site showcases workflow examples like surfacing leads that need a second sales call directly in Slack.
Activepieces is an open automation platform for building AI agents, flows, tables, and MCP integrations with a visual builder and cloud or self-hosted deployment. Teams use it to connect apps, run chat-to-automation workflows, and embed white-label automation in their products.
Adaptional builds custom AI for insurance claims and underwriting that reads, extracts, and analyzes documents so teams focus on decisions instead of manual data prep. The homepage highlights AI-powered continuous claims review.
Adri AI provides AI agents for ERP migration projects, helping consultants explore client business processes and catch critical details. Plans cover SAP ECC and S/4HANA learning indexes, simplification lists, token limits, and optional live system connections.
Advanced Metal Research (AMR) builds closed-loop robotic welding systems and an open robotics stack for American manufacturing. Its public materials center on ROSIE welding intelligence, the War Machine cell, and GradientOS, an in-house open-source robotics controller for experimentation and deployment.
Aemon is an AI R&D engineer that takes a technical problem and evaluation spec, studies state-of-the-art research, runs experiments in your codebase, and delivers validated breakthroughs through self-accelerating research loops.