EveryDev.ai
Subscribe
Home
Developers

4,173+ AI companies

  • Radar
  • Trending
AI Tools by Topic
  • AI Coding Assistants
  • Agent Frameworks
  • MCP Servers
  • AI Prompt Tools
  • Vibe Coding Tools
  • AI Design Tools
  • AI Database Tools
  • AI Website Builders
  • AI Testing Tools
  • LLM Evaluations
Follow Us
  • X / Twitter
  • LinkedIn
  • Reddit
  • Discord
  • Threads
  • Bluesky
  • Mastodon
  • YouTube
  • GitHub
  • Instagram
Get Started
  • Users
  • Rate Tools
  • About
  • Editorial Standards
  • Corrections & Disclosures
  • Community Guidelines
  • Advertise
  • Contact Us
  • Newsletter
  • Submit a Tool
  • Start a Discussion
  • Write A Blog
  • Share A Build
  • Terms of Service
  • Privacy Policy
Explore with AI
  • ChatGPT
  • Gemini
  • Claude
  • Grok
  • Perplexity
Agent Experience
  • llms.txt
Theme
With AI, Everyone is a Dev. EveryDev.ai ©
    1. Home
    2. Developers
    3. EverMind AI

    EverMind AI

    EverMind is an AI research and infrastructure company building long-term memory for intelligent agents. Its stated goal is to make AI remember, understand, reason, adapt, and evolve through a persistent, customizable memory layer.

    Visit Website

    At a Glance

    2Tools Listed
    5Products
    64Tool Views
    10Capabilities
    Discussions
    San Mateo, CaliforniaHeadquarters
    2025Est.
    50Employees
    Focus Areas
    Autonomous Systems
    Multi-agent Systems
    Agent Memory
    Retrieval-Augmented Generation
    Connect
    Latest News
    EverMind introduces EverMe, a personal AI digital twin built on a shared memory layer across agentsSep 8, 2026
    EverMind launches Raven Agent, an open-source self-improving harness for L3-level digital lifeJul 22, 2026
    Markets
    • AI developers and makers
    • Researchers
    • Individual users
    • Startups and product teams
    • +3 more

    AI Tools by EverMind AI

    (2)
    View Raven
    Raven tool icon

    Raven

    Open Source AI Agent Orchestrator

    Agent HarnessMulti-agent SystemsAutonomous Systems
    View EverOS
    EverOS tool icon

    EverOS

    Open Source AI Agent Memory OS

    Agent MemoryMulti-agent SystemsRAG

    Discussions

    No discussions yet

    Be the first to start a discussion about EverMind AI

    Latest News

    09/08/2026

    EverMind introduces EverMe, a personal AI digital twin built on a shared memory layer across agents

    desmoinesregister.com
    07/22/2026

    EverMind launches Raven Agent, an open-source self-improving harness for L3-level digital life

    usatoday.com
    06/26/2026

    Skill Hub launches as a measured foundation for community-powered agents

    evermind.ai
    04/17/2026

    EverMind introduces self-evolving Agent Memory in EverOS

    evermind.ai

    Products & Services

    5
    EverOS
    April 14, 2026 (global public beta; predecessor EverOS research results were published September 30, 2025)

    A self-evolving, multimodal, cross-agent and cross-platform memory operating system available as cloud service and self-hosted/open-source infrastructure. It provides layered memory extraction, mRAG, persistent user and agent memory, Cases, Skills, temporal reasoning, and memory management.

    EverMemOS
    December 9, 2025

    An open-source, self-organizing memory operating system using engram-inspired MemCells and MemScenes / evolving memory structures to provide durable, coherent, continuously evolving memory for agents.

    Raven Agent
    July 22, 2026

    An Apache-2.0 open-source self-improving multi-agent orchestration ecosystem ('harness of harnesses') with built-in Raven-Research, Raven-Code, Raven-Design, and Raven-Oncall agents; it turns runs into Cases and reusable Skills.

    EverMe
    September 8, 2026

    A personal AI digital twin and memory hub whose private, traceable memory can be shared across agents; it organizes a user's digital footprint into knowledge, skills, and a persona that reflects communication style, tone, and values.

    Market Position

    EverMind positions EverOS as an agentic memory operating system rather than a conventional vector-store or retrieval plugin: it emphasizes structured, temporal, multimodal, self-evolving memory, portability, and user control. Its own comparison material identifies Mem0, Zep, MemOS, MemU, Cognee, Graphiti, and LightRAG as relevant alternatives; EverMind differentiates with Cases-to-Skills learning, multi-agent memory, and cloud/self-hosted portability.

    Leadership

    Founders

    DY

    Deng Yafeng

    CEO of EverMind and vice president of Shanda Group; public biographies describe him as a Tsinghua University graduate with more than 20 years of AI-algorithm and product experience.

    Executive Team

    DY

    Deng Yafeng

    Chief Executive Officer

    Also a vice president of Shanda Group; public descriptions identify him as a Tsinghua University graduate with more than two decades of AI algorithm and product experience.

    Board of Directors

    DZ
    Di Zhao
    Member of Technical Staff, OpenAI
    ZT
    Zhe Tang
    Senior Solutions Architect, AWS
    GL
    Guohao Li
    Founder, CAMEL-AI

    Founding Story

    EverMind was incubated by Shanda Group as a long-term-memory initiative. Its core vision grew from the observation that contemporary AI is optimized for instantaneous, static 'spatial' information while intelligence needs a continuous, time-based memory structure; the company set out to build the memory layer for future intelligence and to give agents durable identity and continuity.

    Business Model

    Revenue Model

    Open-source/self-hosted software is offered alongside a managed EverOS Cloud service with subscription access; EverMind has also described custom deployments for strategic partners and a personal version for hardware/MCP contexts.

    Pricing Tiers

    Pro
    $25/month

    EverOS Cloud plan with 8 Memory Spaces, 250,000 MCU per month, 500,000 Retrieval API calls per month, top-up packs, self-evolving Skills, and priority email support.

    Target Markets

    Industries & Segments
    • AI developers and makers
    • Researchers
    • Individual users
    • Startups and product teams
    • Enterprise and strategic partners
    • Agent-platform and workflow developers
    Use Cases
    • Multi-agent systems and agent orchestration
    • Personal AI assistants, chief-of-staff agents, and digital twins
    • Long-running coding, research, design, and workflow-automation tasks
    • Company knowledge bases and institutional memory
    • Customer-support intelligence and professional assistants
    • Personalized AI companions and emotionally aware applications
    Notable Customers
    • Tanka

    Quick Facts

    Headquarters
    San Mateo, California, United States
    Founded
    2025
    Entity Type
    Inc.
    Employees
    50
    Total Funding
    Backed and hosted by Shanda Group
    Investors
    Shanda Group
    Office Locations
    San Mateo

    History & Milestones

    September 8, 2026

    Introduced EverMe, a personal AI digital twin based on a private memory foundation shared across multiple agents.

    July 22, 2026

    Launched Raven Agent, an open-source self-improving agent harness built on EverOS for orchestration, memory, skills, and multi-agent workflows.

    April 14, 2026

    Announced the EverOS brand upgrade and global public beta, introducing self-evolving agent memory, mRAG, HyperMem, and a Skills Evolution Engine.

    March 18, 2026

    Released Memory Sparse Attention (MSA), an end-to-end trainable sparse latent-state memory architecture aimed at scaling long-term memory beyond 100M tokens.

    December 9, 2025

    Announced the release of open-source EverMemOS; the announcement reported 92.3% LoCoMo accuracy and 82% LongMemEval-S accuracy and said EverMemOS powered Tanka.

    Key Capabilities

    10
    Persistent long-term memory across sessions, agents, devices, and platforms
    Self-evolving Agent Memory that records Cases and distills recurring patterns into reusable Skills
    Multimodal ingestion and retrieval (PDFs, images, Word documents, spreadsheets, presentations, emails, HTML, text, and URLs)
    mRAG hybrid retrieval combining semantic, sparse-keyword, and multimodal alignment approaches
    Temporal knowledge tracking, contradiction resolution, profile updates, and lifecycle controls
    Traceable, inspectable, editable, exportable memory with source provenance

    Integrations & Partnerships

    Platform Integrations

    • Claude Code
    • OpenClaw
    • Hermes Agent
    • Codex
    • MCP
    • OpenAI SDK
    • Anthropic SDK
    • Discord

    Key Partnerships

    Incubated by Shanda Group
    EverMemOS powers Tanka
    EverMind's public beta included a Global Developer Community Co-Creation Plan for applications and plugins

    Connect

    Website
    evermind.ai
    GitHub
    EverMind-AI
    X / Twitter
    evermind
    LinkedIn
    ai-evermind
    Discord
    gYep5nQRZJ

    AI Topics

    5

    EverMind AI focuses on these topics:

    Autonomous Systems(1)
    Multi-agent Systems(2)
    Agent Memory(1)
    Retrieval-Augmented Generation(1)
    Agent Harness(1)
    Back to all developersSuggest an edit