FAQ

What's free and what's Pro?

The full method is free. Every PrOACT step, the agent tools, the living decision document, the knowledge base, Markdown and print exports — no caps on decisions. We give away the recipe.

Pro is for decisions other people are part of: inviting collaborators who can comment, minting share links (whole document or single sections), and sealing decisions with zero-knowledge encryption. Current prices are on the pricing page; collaborators you invite participate free.

What does "bring your own agent" mean?

There is no chatbot inside Perspicuity. You connect the AI you already use — Claude (claude.ai, desktop, Claude Code), ChatGPT, or any MCP-capable agent — and it facilitates the method, using Perspicuity's tools to build the document. Your agent runs on your side with your subscription or API key: we never proxy its model calls and never see that key. Setup is one URL and a browser sign-in.

What does zero-knowledge sealing protect — and not protect?

When you seal a decision, its most sensitive prose is encrypted by your agent before it reaches our server, with a key derived on your side that is never sent to us. We store ciphertext we cannot read — and neither can anyone who compels us to hand over your data.

What sealing does not hide: the decision's structure — labels, step layout, numeric scores — stays readable so the document still works as a document. And your agent's own conversation with you happens under your model provider's terms, outside our reach either way. One real warning: lose the passphrase and the sealed content is gone — that's the promise working as designed.

Who can see what when I share?

• A share link shows exactly the scope you minted it for — the whole document or a single section — read-only. Revoke it any time and it dies immediately.
• Collaborators you invite by email can read the full document and leave comments and approvals under their own name.
• Public surfaces never show full email addresses — visitors see display names only.
• Sealed content stays sealed on every shared surface unless the key travels in the link fragment you chose to send (#k= never reaches our server).

How do approvals work?

Anyone you've shared a decision with — collaborator or share-link visitor — can cast an approval on the recommendation. The document shows the running tally wherever it's rendered, including print. It's a lightweight commitment record: who has seen it, who stands behind it, captured in the artifact itself.

How do I cancel or change my plan?

Account → Billing — cancel takes effect at the end of the period you've already paid for, and you can resume before then with one click. No retention flows, no email required. Card processing is handled by Helcim (Canada); we never see your full card number.

What about refunds?

If something went wrong with a charge — a duplicate, a mistake, a renewal you meant to cancel — write to privacy@peopleandplanet.consulting and we'll sort it out. A human (the founder) reads every message.

Where does my data live?

On our server at Hetzner in Germany, with rotating encrypted backups capped at roughly two weeks. No third-party analytics, no trackers. The full picture — sub-processors, retention windows, your rights — is in the privacy notice, written for humans. Deleting your account hard-deletes your decision content within minutes.

Can I get my decisions out?

Always — every decision exports as Markdown or print-grade PDF, and your profile has a full JSON export. Your agent can also read everything back out through the same tools it used to write it. No lock-in is a design constraint, not a promise.

My agent can't connect — help?

The troubleshooting section covers the common snags: expired tokens, stale tool lists, ChatGPT's Developer-mode gate for write tools. Still stuck? privacy@peopleandplanet.consulting.