Frequently asked questions.
What Becki does, how she connects to your AI tools, where your data lives, and what the waitlist actually gets you.
What is Becki?
Becki is a memory layer that lives between you and every AI tool you use. She captures the decisions, dead ends, and commitments from your meetings, your AI conversations, and your work. Then she makes that memory available to Claude Code, Claude Desktop, Cursor, Codex, Gemini CLI, Zed, VS Code, Raycast, claude.ai in your browser, ChatGPT in your browser (dev mode), Perplexity (Pro plan), and the claude.ai app on your phone. One memory. Every tool. Yours.
How does Becki actually work?
Becki runs as an MCP server. MCP is the open protocol every modern AI tool now speaks. When you ask Claude (or any other AI tool) a question, Becki feeds in the relevant context. The decisions you made. The dead ends you already ruled out. The commitments you've made. The open loops you're tracking. Nothing pasted, nothing managed. Just the memory showing up where you need it.
What tools does Becki connect to?
Desktop and CLI (one-click install): Claude Code, Claude Desktop, Cursor, Codex, Gemini CLI, Zed, VS Code, Raycast.
Web (one-time manual connector add): claude.ai (one click), ChatGPT (dev mode, a couple extra clicks the first time), Perplexity (Pro plan required, one-click connector).
Mobile: the claude.ai app on your phone.
Next on desktop: ChatGPT Desktop and Obsidian.
If a new AI tool ships with MCP support (Grok, whatever comes after), Becki picks it up the day it lands. The MCP protocol is the standard. Becki is the layer everything reads from.
My AI tools already have memory. Why do I need Becki?
They do. Each one keeps its own. The problem is they don't share. What you teach Claude isn't what ChatGPT knows. What you decide in Codex isn't what Cursor remembers. You're the one re-explaining your project every time you open a new tool.
Becki is the shared memory those tools read from. Including the dead ends. The path you ruled out in Cursor in March doesn't come back when you ask Codex in May.
What you teach one, the others remember.
Where does my data live?
Your data lives on your Mac. Markdown files in your Becki folder, audio recordings, local databases for meetings and commitments. That's the source of truth.
A retrieval-optimized copy of that content syncs to Becki's database in the cloud so cross-tool memory works fast enough to be useful mid-conversation. Every user's rows are isolated by your account credentials. Never sold. Never used to train AI. Never shared with anyone you didn't authorize.
Do you use my data to train AI or sell it?
No. Your data is yours. Not sold. Not aggregated for anyone else. Never used to train models. The whole point of Becki is that your memory belongs to you. If that posture ever changes, you'll know about it before it happens.
What platforms does Becki run on?
Becki the application runs on macOS today (Apple Silicon, macOS 14.4 or later). Your memory reaches off the Mac as of v0.10.0. claude.ai in your browser, ChatGPT in your browser (dev mode), Perplexity (Pro plan), and the claude.ai app on your phone can all read from your Becki vault through a remote MCP connection.
Windows for the full app is on the roadmap. iOS for capture is too.
Does Becki record my meetings?
Optionally, yes. Becki can capture system audio and your microphone during meetings, transcribe locally on your Mac (no upload to anyone's server for transcription), and extract structured memory. Decisions made. Commitments captured. Questions left open. Dead ends marked.
You can turn capture on or off per meeting. Audio and transcripts stay on your machine.
How much does Becki cost?
Pricing goes live at launch. Designed to be sustainable, never extractive. If you're on the launch email list, you'll know the day pricing publishes.
How does the waitlist work?
Drop your email at becki.io. You'll get an email the day Becki launches. That's it. No spam, no follow-up nudges. If you want off the list before then, email hello@becki.io and you're out.
Can I export or delete my data?
Yes. Your memory is already in Markdown files on your Mac. Copy them anywhere, archive them, version them in git. Delete any single entry from the Settings panel. Delete your whole account too, also from Settings. When you delete, it's gone from Becki's cloud retrieval index as well. Email hello@becki.io if you want written confirmation of deletion.
Does Becki support teams or admin controls?
Not today. Becki is single-user for now. Team workspaces and admin controls are on the roadmap for the quarter after public launch. If team support is a hard requirement for your decision, email bryan@becki.io and I'll scope it against your use case.
What's different about Becki vs other AI memory tools?
Two things.
One. Becki captures dead ends as first-class memory. The path you ruled out in Cursor in March is the path Codex won't re-suggest when you ask it the same question in May. Most memory layers only remember what worked. Becki remembers what didn't work too, and that's the part that actually compounds over time.
Two. Becki integrates via MCP across 11+ AI tools (desktop, web, mobile) without locking you into any of them. Your memory is the asset. The AI tools are interchangeable.