Tanka Launches EverMemOS AI-Native × Memory-Native OS

Tanka launches EverMemOS, an OS that gives organizations virtually unlimited, persistent, and searchable memory, breaking through the limitations that have held AI back.

For the first time, teams can compound intelligence and value over time.

EverMemOS addresses a core challenge for enterprise AI: not simply mitigating information loss but creating a truly persistent and evolvable AI memory system. This is critical for modern organizations that are overwhelmed by scattered documents and communications, which results in critical decisions getting buried and lost knowledge when key people leave.

As Bessemer Venture Partners noted in their 2025 State of AI report, Memory and Context are the new moats for AI applications – yet most collaboration tools lack true memory. Powered by EverMemOS, the company takes a fundamentally different approach: transforming every interaction into persistent, retrievable organizational memory, proactively surfacing insights and driving execution, thereby ensuring every detail becomes a catalyst for smarter, faster growth.

Architecture of EverMemOS, which transforms scattered data into unified, retrievable memory to power AI-native collaboration and context-aware execution.
(Credit: Tanka)

Click to enlarge

Inspired by neuroscience, the firm’s EverMemOS is a unified memory OS that integrates all your internal and external data, turning scattered information across Slack, Gmail, Notion, and WhatsApp into a single source of truth. Its Dynamic and Static Memory Architecture mirrors how human memory works, balancing instant recall with deep, structured long-term knowledge. Meanwhile, Behavior-Driven AI continuously learns from team actions and decisions, proactively suggesting replies, tracking follow-ups, and orchestrating collaboration.

EverMemOS’ edge has enabled it to achieve a Locomo score of 92.3%, the highest reported to date on this emerging benchmark for long-context memory systems, thereby setting a new global standard for AI memory systems and cementing Tanka’s leadership in semantic memory and intelligent agents. This breakthrough is powered by atomic memory structuring for precise extraction, memory purification for noise-free reasoning, enhanced answer generation via chain-of-thought logic, and a next-gen recall mechanism leveraging embedding and reranking.

As Kisson Lin, CEO, Tanka, puts it: “EverMemOS is the answer to the most fundamental question every startup faces: how do you preserve and leverage everything you learn as you build? For the first time, founders have an AI cofounder with true long-term memory, one that remembers every detail, learns continuously, and proactively drives execution. This is not just enhanced productivity – it’s a new era of memory-powered competitive advantage.“

Working closely with the startup community the firm understands that companies move at breakneck speed, pivot often, and face constant change, which is why the company has dedicated itself to rolling out a purpose-built memory system that ensures every lesson, decision, and breakthrough is remembered, leveraged, and never lost. The result is a transformative impact on how teams collaborate and retain knowledge.

To support broader adoption and encourage innovation, Tanka will soon open-source EverMemOS. By sharing its core memory infrastructure, the company aims to make it easier for developers to build AI-native tools that remember, learn, and adapt – unlocking new possibilities for teams and organizations WW.

Technical Overview: Tanka’s Unique Value Underpinned by Three Pillars of Innovation
Building on the company’s proprietary architecture, the firm delivers three pillars of unique value – demonstrating the power of EverMemOS as a platform:

• Ecosystem Memory: Complete, Connected Knowledge
  • Tanka unifies all your apps and data sources – Slack, Gmail, Notion, WhatsApp, and more – so every insight, decision, and file is captured and organized in one place. For founders, this means nothing gets lost, even as teams grow, pivot, or change.
• Collective Memory: Interactive, Proactive Collaboration
  • AI assistant not only remembers every conversation, but also actively engages with your team – proactively following up, assigning tasks, and generating smart, context-rich replies. This ensures your team stays aligned, moves faster, and organizational knowledge is always accessible and actionable.
• Individual Memory: Deep Thinking to Tangible Results
  • Empowers founders to turn deep, independent thinking into real outcomes – helping you develop ideas into detailed plans and shareable deliverables, like business plans or websites. With memory-driven support at every step, your insights become actionable results.