Find the right AI company.

Concept Loom gives coding agents a structured teaching workflow with prerequisite maps, concealed-answer checkpoints, local Markdown notebooks and saved learning progress across sessions.

Yomiyasu refines Japanese technical and business writing through agent-guided structural editing, with local lint and diff tools that flag wording, formatting and possible meaning changes for review.

EasyRead turns research PDFs into annotated, bilingual reading workspaces with original-page alignment, equations, tables, citations and AI discussion through local or user-selected models.

Operator Memory gives coding agents durable Markdown specifications, decisions, standards and indexes across private project, shared project and user-global partitions.

Magnitude runs open-weight models locally through a desktop app, headless CLI and compatible inference APIs, compiling and tuning kernels for the machine's hardware.

ai-memory gives coding agents shared, self-hosted memory and typed handoffs, with Git-backed Markdown as the durable source and a rebuildable retrieval index.

CodeGraph builds a locally synchronized graph of symbols, calls, imports and routes so coding agents can inspect dependencies and change impact through MCP.

Alibaba hybrid deterministic-plus-agent CLI reviews diffs/files with line comments, built-in rules, codebase search and parallel bundles.

Agent memory layer with retain, recall and reflect, mental models and knowledge pages; self-hosted, embedded or managed.

Cross-platform agent development environment runs Codex, Claude, OpenCode and Pi side by side in isolated worktrees with mobile monitoring and diff review.

Knowledge platform combines cited RAG, multi-step agents, generated wikis, memory, skills/MCP, sandboxed tools, RBAC and audit logs.

Local desktop suite for voice cloning/design, dubbing, dictation, transcription and audiobooks across 646 languages with local API/MCP.

Operational-assurance auditing for deployed agents, with self-hosted inspections and paid suites covering purpose, authority, workflows, evidence and governance.

AI coworker analyzes and operates LinkedIn Ads/ABM from app or MCP, grounded in live ads and CRM data.

Mac/Windows memory app records screen and meeting audio locally, indexes on-device and exposes searchable context to Claude Code, Codex and Cursor.

Vectorless document retrieval builds a hierarchical page/tree index that an LLM reasons through, locally or with managed OCR/storage.

Self-hosted harness joins Claude Code, Codex and Pi into persistent named teams with durable roles, queues, direct messages, TUI, MCP and tmux workspaces.

AI-native B2B marketing automation with per-person playlists, account/buying-group records, campaigns, approvals, MCP/API and cross-campaign traffic control.

Agent acceptance testing turns outcomes into checks, operates real web/CLI/native surfaces and publishes screenshots, recordings and logs.

Local 0.6B decision model returns calibrated distributions for boolean, choice and score questions without text decoding.

Native macOS share extension converts merged WeChat chats into local archives and routes them to Codex, Claude, Obsidian or another app.

Local desktop app installs, updates and repairs ten AI CLIs while managing providers, skills, MCP servers, prompts and sessions.

Runtime for agent fleets in kernel-isolated sandboxes with declarative policy, credential injection, observability and runtime checks.

Stable Astro CMS with built-in admin, MCP/OAuth, API/CLI, bundled agent skills and permissioned sandboxed plugins.