One memory. Every AI you use.

Why flat-rate, not pay-per-query?

Becki's value compounds with use. Token-metered pricing punishes the people who get the most out of her. And they're the people who'd evangelize her hardest. Flat rate keeps the incentives aligned.

How does Becki actually save me money on tokens?

Connected AI tools (Claude, Cursor, Codex, Zed, Gemini CLI, Raycast) call Becki's MCP server for context instead of you re-explaining yourself each session. Pre-extracted decisions are ~5-10× denser than chat history. Across 4+ tools, this typically reduces input-token usage by 30-50%. Concrete savings on your Claude Pro / ChatGPT Plus / API bills.

What's the session sweeper?

A background process that watches your AI-tool session logs (Claude Code, Codex CLI, etc.) and harvests decisions, dead-ends, commitments, and open-loops directly from the conversation, even if the agent never explicitly called becki_ingest. Both Core and Studio run a daily digest pass. It's off by default; you opt in from Settings (Studio) or via becki-mcp init (Core).

What's "Bring your own AI provider key" (BYOK)?

Studio only. Plug in your own API key for Anthropic, OpenAI, or Google in Settings, and Becki uses it for extraction (commits, sessions, meeting transcripts) and synthesis (Ask Becki responses) instead of routing through our backend. Pick the provider that fits your existing contract. If you already pay OpenAI for Codex / GPT-5 / o-series, or Google for Gemini, BYOK lets your Becki spend land on the same bill. Embeddings and the rest of the vault pipeline stay on our backend, so the experience doesn't change. Core uses our backend for all LLM calls (included in the $15/mo). BYOK is on the roadmap.

Core vs Studio. Which one do I want?

Core ($15/mo) is the MCP daemon. It runs in the background on Win / Mac / Linux, ingests your AI tool sessions, git activity, and anything you write to it, and serves it all back to Claude, Cursor, Codex, and any other MCP client. No app, no meeting capture. If you live in your editor and chat tools, Core is enough.

Studio ($25/mo, Mac only) is everything Core has, plus the parts that only work when Becki is on your machine and in your meetings. It records your calls on-device, so the audio never leaves your Mac, and captures the commitments you made, the asks you're still waiting on, and what's blocking you. Then it nudges you before a deadline lands, and when someone still owes you a reply you forgot about. You also get a daily briefing that preps your next meeting from your calendar and prior context, the NeuraVault 3D graph to see and move through everything Becki holds, and bring-your-own-key for Anthropic, OpenAI, or Google so the AI spend lands on a bill you already pay.

Pick Core if you want one memory across your editor and chat tools. Pick Studio if you also want Becki in the room, catching the commitments you'd otherwise lose and following up on them for you.

Is my data used for training?

No. Becki's underlying AI providers all operate under no-train policies. Your data is never used to train any model. Audio is transcribed locally on your Mac and never leaves the device. Vault content is encrypted at rest in your private account.

What happens to my data if I cancel?

Becki keeps your vault for 60 days after cancellation in case you come back. After 60 days, everything is purged. You can also export everything as JSON from Settings → Export anytime.

Will pricing change later?

Our default is to keep existing subscribers at the price they signed up at when we raise list pricing. If costs ever force us to reprice existing accounts, we'll give at least 60 days notice and the option to lock in another year at your current rate before any change.

Why no lifetime deal?

Lifetime deals trade years of revenue for one-time cash and lock us into supporting users we can't fund long-term. We'd rather build a real business we can keep showing up for. Annual is the largest discount we'll do.

Can I get isolated infrastructure?

On Enterprise, yes. Your data lives on a dedicated, single-tenant deployment in the region of your choice. Custom retention and redaction policies can be applied at the deployment level. Studio and Team accounts share the multi-tenant backend, with row-level isolation per account.