Recallbase / Blog
Blog
Thoughts and practice around persistent memory for AI agents – context, teams, comparisons and setup.
Articles
What we built into AI memory from brain science
A forgetting curve with a formula, surprise as a measurable score, prospective memory, self-model - neuroscience in code.
Read →How Recallbase works internally
A memory architecture on one Postgres: hooks, distillation, bi-temporal graph, rank fusion, RLS.
Read →
Why your AI agent forgets everything every morning
Amnesia by design: the three escape routes – and why only one scales.
Read →
Your AI memory on your phone
Recallbase as a connector in the Claude app: check status, search knowledge, save notes – no installation.
Read →
Why a bigger context window is not a memory
More context is short-term memory. Why "load everything in" fails on cost, degradation and findability.
Read →How a shared team memory emerges
Turn many isolated AI sessions into one shared memory – without wiki upkeep.
Read →Comparisons & guides
Recallbase vs. CLAUDE.md
Why long instruction files tip over – and what a memory does differently.
Read →
Recallbase vs. claude-mem
Local plugin or team memory – where the line runs.
Read →Recallbase vs. Mem0
Finished service instead of SDK: different levels of AI memory.
Read →Recallbase vs. Wiki & Notion
Knowledge that writes itself, instead of being maintained.
Read →Memory for Claude Code
How session hooks let Claude Code remember for good.
Read →Integration guides
ChatGPT
Recallbase as a remote connector in ChatGPT – no installation.
Open guide →Codex CLI
Connect Recallbase to Codex CLI via MCP.
Open guide →Gemini CLI
Memory for the Gemini CLI via MCP.
Open guide →OpenCode
Project-scoped memory for OpenCode.
Open guide →Cursor
Give Cursor a memory via MCP.
Open guide →Try it yourself
Start for free, no credit card. In two minutes your agent remembers every session.
Start for free