Your information.
Before you sign in, your starting inputs stay in this browser tab for up to 24 hours. Closing the tab can remove them. Signing in to save or download your first brief attaches those inputs to your verified account. This private application then saves the website, product or business details, customer goal and optional public integration references you provide. Drafts are linked to your signed-in account and stored privately.
When you save a plan, we also create a product record, customer task and a snapshot of the facts you confirmed, so the same scope can be used for delivery work. A basic check identifies missing fields in that snapshot. The starting form alone does not crawl your website. If you choose to check your live website, we read up to three public pages through Jina Reader and ask OpenAI to prepare and review your business context. After you confirm a buying question, we sample three searched answers, from OpenAI and, when enabled, Anthropic (Claude), Google (Gemini) or Perplexity, and prepare a reviewed first decision. We retain those page snapshots, answers and provider receipts privately. We do not execute integrations, publish changes or collect payment through the free check.
In your workspace, we also store supplied specifications, documentation, file versions, approvals, publication references and test observations. Downloads are under your control. Recording a publication does not publish anything automatically. If payment is enabled, card details are collected by the payment provider; this app stores order and payment status.
On your first Agent Insights workspace visit, we read up to four relevant public pages through Jina Reader. OpenAI reviews those excerpts to suggest your business description and customer questions. Suggestions stay separate from your saved report settings until you confirm them. We retain the source quotes and provider receipts.
When you run Agent Insights, your selected public-page URLs are sent to Jina Reader. Your customer questions are sent to OpenAI for search-enabled answers and, when enabled, to Anthropic (Claude), Google (Gemini) and Perplexity. The business name and website are kept out of those search prompts. OpenAI then receives the retrieved public excerpts, questions and observed answers to draft and review recommendations. Private uploaded API specifications and documents are checked inside ByAgents and are not sent to the AI provider.
We retain report settings, page snapshots, original answers, citations and provider receipts privately so you can review and compare reports. OpenAI background responses are stored by that provider for retrieval. A report sends no requests to your business API, submits no forms and publishes no changes.
If you join the Autopilot early-access list, we save your email, website and consent date so we can contact you about that service. Joining the list does not create a subscription or start agent work.
If you turn on the agent front door, people and AI agents can send you requests through it. We store each request, including the sender's name, email address and message, and a keyed hash of the sender's network address used only to limit repeated requests; we never store the address itself. We email requests and your check alerts to you through Resend. Sign-in is handled by Clerk, and subscription payments by Stripe, which collects your card and billing details; we store the resulting customer, subscription and payment status, not card numbers. The app runs on Cloudflare.
Please do not enter passwords, API keys or other secrets. To ask about access or deletion, contact team@byagents.dev from your account email. We may need to verify ownership.