Exogram is a personal AI memory vault shared across ChatGPT, Claude and Gemini. Free plan available; Personal Pro is $19/month. Read the full review.
Researched and fact-checked by Bartu Cavusoglu · Updated October 26, 2026
Exogram is a personal AI memory vault and workspace that keeps a permanent, encrypted record of your work, life, and relationships, so you do not have to re-explain yourself in every chat. It organizes what it learns into a Personal Neural Network covering three spheres: Personal, Social, and Professional. The stated goal is to let you speak naturally without prompt engineering.
Memory is not tied to one vendor. Exogram works across Claude, ChatGPT, Gemini, and other models, and it reaches coding tools through a FastMCP server. It was built by Richard Ewing, an independent technologist in Seattle, who says every AI conversation began with total amnesia. The company says its memory verification ledger was first used in production on CareerWin.ai.
Exogram describes six phases. It saves facts and decisions to an encrypted vault, identifies the people and projects involved, and links facts into a multi-hop graph. When facts change, a bi-temporal ledger updates current state while keeping history. For each request it pulls only the needed facts, using Reciprocal Rank Fusion that combines dense vector similarity with BM25 search.
A reflex layer called Laya resolves intent in about 30 milliseconds and returns Living Tiles, while frontier models handle deeper reasoning. Memories progress from chat to candidate to verified fact, backed by SHA-256 provenance hashes and conflict handling (ADD, UPDATE, MERGE, DELETE). Prepared work stays reviewable, and consequential external actions remain under your explicit authority.
The core feature is the Personal Neural Network, a graph of people, events, and decisions that is searched across multiple hops rather than returning only top text chunks. It sits on an AES-256 encrypted vault with a local SQLite database. The multi-provider swarm routes subtasks across Claude, ChatGPT, Gemini, Perplexity, Kimi, Llama, and local Ollama models using the same memory.
Living Tiles turn requests into drafts, checklists, metric snapshots, and status trackers. The Memory Porter imports existing context from other assistants, and the FastMCP server shares the vault with Cursor, Claude Desktop, Claude Code, and Windsurf. A separate governance layer evaluates agent actions through a REST API or MCP server, with signed execution tokens and an audit chain.
Exogram is a personal AI memory vault and workspace that stores your facts, decisions, relationships, and preferences in an encrypted Personal Neural Network. It shares that memory across ChatGPT, Claude, Gemini, and MCP-compatible tools so you do not repeat yourself.
The Everyday Life plan is free forever at $0 with no credit card required. Personal Pro costs $19/month and can be cancelled online at any time. Enterprise is custom-priced on an annual basis.
Yes. The free plan includes the 3-sphere Personal Neural Network, the Memory Porter, pooled free AI models, unlimited bring-your-own-key, an encrypted vault, and a Living Knowledge Graph preview.
Yes. You copy the Memory Porter prompt into ChatGPT or Claude, then drop the resulting markdown into Exogram. The site says this takes about 60 seconds.
Exogram connects to ChatGPT through a Custom GPT, and to Claude Desktop, Claude Code, Cursor, and Windsurf through MCP. It also works with agent frameworks such as LangChain, CrewAI, AutoGPT, and OpenClaw through a REST API or MCP server. Gemini support is listed as coming soon.
The site states that all plans include a zero-training commitment, so private memories are not used to train models. All plans also include AES-256 encryption at rest, TLS 1.3 in transit, and full data export.
Yes. The site lists full data export portability on all plans, and Personal Pro includes 1-click cryptographic JSON and markdown export. Personal Pro can be cancelled online through Dashboard, then Billing, with no penalty.
Exogram suits professionals, founders, consultants, and developers who use several AI assistants and want one persistent memory. Organizations needing isolated vaults and SSO can use the Enterprise plan.