Slowave
Slowave is an open-source, local-first persistent memory layer for AI coding agents and other AI tools. It gives supported clients one shared memory across sessions, using local embeddings and deterministic mechanisms so useful context can strengthen while stale context fades, without a separate LLM or hosted memory service.
At a Glance
- Individual developers and AI-assisted programmers
- Teams using AI coding agents
- Developers and researchers building MCP-compatible agent workflows
- Organizations needing local-first or privacy-sensitive agent memory
- +1 more
AI Tools by Slowave
(1)
Slowave
Local Memory Layer for AI Agents
Discussions
No discussions yet
Be the first to start a discussion about Slowave
Latest News
Slowave v0.20.3 released with MCP Registry metadata
Slowave v0.20.2 released with hard-delete memories preview
Slowave v0.20.1 released with hard-delete memories from the dashboard and README
Slowave v0.20.0 released with shared multilingual retrieval
Products & Services
AGPL-3.0-or-later open-source package for brain-inspired long-term memory for AI agents. It can be installed with pipx or pip and configured with slowave setup; the project is classified as beta and requires Python 3.11 or newer. Sources: https://www.slowave.ai/; https://pypi.org/project/slowave/; https://github.com/slowave-ai/slowave/blob/main/pyproject.toml
A local service exposing the five-tool lifecycle activate, remember, recall, feedback and commit, with background consolidation of memories and procedures. Source: https://slowave-ai.mintlify.app/
A local browser dashboard for inspecting and controlling memories, procedures, retrievals, activity, system health and the memory graph; users can trace evidence and forget or restore memories. Source: https://www.slowave.ai/
Market Position
Slowave positions itself against static RAG stores, transcript or Markdown-file memory, LLM-generated summarization memory and tool-specific memory silos. Its differentiators are local SQLite storage, inspectability, cross-tool sharing, deterministic/local-embedding memory maintenance, feedback-shaped salience and no extra LLM API calls in the memory loop. The project materials do not name specific commercial competitors.
Leadership
Executive Team
mrsalty
Visible maintainer and package author
Named as the author in the package metadata and credited for the recent GitHub release commits and repository development. The public materials reviewed do not provide a personal name, résumé or corporate title. Sources: https://github.com/slowave-ai/slowave/blob/main/pyproject.toml; https://github.com/mrsalty; https://github.com/slowave-ai/slowave/releases
Founding Story
Slowave's project materials identify the problem it was designed to address: AI-agent context ends with a session, forcing users to restate decisions, constraints and failed approaches when they start a new session or switch tools. Its initial vision was an adaptive, inspectable memory layer shaped by agent feedback rather than a static retrieval store, transcript, summary, or tool-specific silo.
Business Model
Revenue Model
The documented offering is free/open-source software under AGPL-3.0-or-later. COMMERCIAL.md says organizations needing proprietary products, closed-source hosted services or embedded commercial offerings can contact the author for a separate commercial license; no hosted subscription or usage-based service is documented.
Target Markets
- Individual developers and AI-assisted programmers
- Teams using AI coding agents
- Developers and researchers building MCP-compatible agent workflows
- Organizations needing local-first or privacy-sensitive agent memory
- Open-source contributors and researchers evaluating adaptive memory systems
- Carrying coding decisions, constraints, preferences and lessons across AI-agent sessions
- Switching between AI coding clients without losing project context
- Reusing successful multi-step solutions as procedures
- Reviewing, auditing, correcting, forgetting or restoring agent memory locally
- Privacy-sensitive or offline-oriented agent workflows that avoid sending memory to a hosted service