Atomic Strata
Atomic Strata is building a configurable context-memory layer for AI applications, agents, teams, and organizations. Its stated mission is to make AI context durable, inspectable, correctable, governable, and composable across tools, models, agents, and organizations.
At a Glance
- AI application developers
- Coding-agent and developer-tool teams
- Enterprise AI and platform engineering teams
- Customer-support and service-automation teams
- +2 more
AI Tools by Atomic Strata
(1)
llmwiki
Markdown Wiki Compiler for LLMs
Discussions
No discussions yet
Be the first to start a discussion about Atomic Strata
Latest News
llmwiki 1.4.2 released with Codex-agent provider compilation fix
Atomic Memory CLI 0.4.0 released with hosted MCP support and cloud-first plugin setup
CEO Juan Bruce discussed Atomic Strata's enterprise AI context layer and vendor-lock-in thesis in a The Block interview
Published 'Agents Need Compiled Knowledge, Not Just Context,' describing llmwiki's compiled-knowledge approach
Products & Services
Open-source, Apache-2.0 AI memory infrastructure consisting of a configurable SDK and self-hosted Core engine. It is designed to let developers inspect, correct, customize, swap, and run memory on their own infrastructure.
Atomic Strata's managed hosted memory service with a console, API, CLI, configurable extraction and retrieval, operation history/audit capabilities, and organization/project management.
Custom deployment offering for dedicated performance, governance, private or on-premises hosting, RBAC/ABAC, enforced write-time policies, full auditing, key management, and custom compliance/SLA requirements.
An open-source knowledge compiler that turns papers, documentation, notes, transcripts, and web pages into a structured, interlinked, citation-traceable wiki with semantic retrieval, a local viewer, MCP access, SDK control, and Open Knowledge Format exchange.
Market Position
Atomic Strata positions itself against hosted, opaque AI-memory and tightly coupled model/context stacks by emphasizing configurability, portability, inspectability, correction, governance, and self-hosting. llmwiki is differentiated from conventional RAG by compiling sources into a durable, structured, citation-traceable artifact rather than repeatedly retrieving raw chunks at query time.
Leadership
Founders
Juan Bruce
Co-founder and CEO. His public profile describes experience at IDEO, BMW, and Team Downey; he also founded the early crypto venture fund Aenigma Capital and previously co-founded SuperNet.
Aileen Lei
Co-founder and CMO. Public professional materials describe her as a serial founder and AI builder with prior experience at Disney, Netflix, and Binance; she also founded Narra, an AI-native consumer entertainment platform.
Ethan Joffe
Chief Technology Officer and founding engineering leader. His public profile identifies MIT affiliation and an AI-innovation focus; he is the primary public maintainer of Atomic Strata's llmwiki compiler releases.
Executive Team
Juan Bruce
Co-founder and CEO
Entrepreneur and investor with public profile experience at IDEO, BMW, Team Downey, and Aenigma Capital; previously associated with SuperNet.
Ethan Joffe
Chief Technology Officer
MIT-affiliated AI and infrastructure builder; public maintainer of Atomic Strata's llmwiki and Atomic Memory open-source projects.
Founding Story
Atomic Strata says it was started by Stanford and MIT engineers who saw persistent AI context becoming production infrastructure while teams were rebuilding memory in fragments. The initial vision was a shared, configurable continuity layer that developers and enterprises could inspect, correct, govern, and run without being locked to a hosted black box.
Business Model
Revenue Model
Managed Memory Cloud subscriptions and usage-based memory operations, plus custom enterprise deployments and services. Atomic Memory and llmwiki are open-source foundations that can be self-hosted.
Pricing Tiers
10,000 memory operations per month; one project and one member; shared hosting; community support.
250,000 memory operations per month; three projects and two members; email-priority support; one month free.
1,500,000 memory operations per month; five projects and two members; priority support; one month free.
6,000,000 memory operations per month; ten projects and up to ten members; dedicated hosting marked coming soon; priority support; one month free.
Dedicated or private/on-premises environments, governance console, custom access control, write-time policies, full audit, key management, compliance, and custom SLAs.
Target Markets
- AI application developers
- Coding-agent and developer-tool teams
- Enterprise AI and platform engineering teams
- Customer-support and service-automation teams
- Research, technical-writing, and open-source-maintainer communities
- Organizations needing private, self-hosted, or compliance-oriented AI infrastructure
- Persistent memory for coding agents and terminal workflows
- Support and service agents that need durable customer context
- AI applications requiring user, team, project, or workflow memory
- Enterprise AI governance, auditability, provenance, and hallucination reduction
- Building durable knowledge bases from research papers, documentation, notes, and transcripts
- Open-source maintainers and technical writers creating citation-traceable reference wikis