Privacy
Dispatched arranges home services through the businesses’ own systems. This notice explains what we collect, why, who we share it with, how long we keep it, and how to have it corrected or deleted. Account sections apply only when account continuity is enabled for you.
What we collect
When you use the conversation or an AI assistant
Dispatched runs a first party conversation at dispatchedhome.com and the same tools inside AI assistants such as ChatGPT, Claude, and Gemini through our Model Context Protocol server. In both places we receive the messages you send and the facts a search or a request actually needs:
- the service you want and a plain description of the job;
- a postal code, and a city or region;
- a preferred date and time window, and a backup;
- property facts you choose to give, such as the number of bedrooms and bathrooms;
- an email address, so we can send you the status of a request you create;
- a name, only if you give one. It is optional.
We do not ask for a full street address, and we do not ask for card or bank details in the conversation. Please do not send them. If you do, they are treated as request content and are subject to the deletion route below.
If you use account continuity
You can use Dispatched and submit a service request without signing in. When account continuity is enabled and a request is successfully saved, Dispatched may create a provisional account for that browser so you can return to the request. A provisional account is not a verified identity and cannot authorize an AI client to access your saved context.
You may choose to save only these home-service facts:
- a property label and postal code;
- bedroom, bathroom, and square-footage counts;
- supplies and scheduling preferences;
- whether pets are on site; and
- a structured parking preference.
Dispatched does not save current-turn hints or model guesses unless you explicitly confirm an allowed fact. Account context cannot store a raw transcript, street address, access instructions, contact details, authentication data, payment data, or arbitrary notes.
If you continue with Google
Google is used only to authenticate you. Dispatched requests exactly the openid, email, and profile scopes and does not request offline access. Supabase Auth processes the Google identity. The Dispatched application database does not store a Google access token, refresh token, authorization code, profile photo, or Google profile payload.
Automatically
- Standard server logs from our hosting provider, including IP address, user agent, and timestamps.
- A one way keyed hash of the network address, used only to enforce rate limits on anonymous use. The hash cannot be reversed to an address.
- Product analytics events that record which step happened, never the content of your messages.
If you represent a home service business
Applications to join the network are covered by the separate provider intake notice, which governs the business name, the authorized representative’s contact details, and the systems the business uses.
Signing in to the provider portal to describe and manage your business is covered by the separate provider portal notice. It governs your sign-in identity, the private draft description of your business, its connected systems, how long an unused business is kept, and the difference between removing your own access and deleting the business.
Why we use it
- To understand your request and ask for anything still missing.
- To search the directory and, where a business is connected, to retrieve prices and availability.
- To record an appointment request, show you its status, and let you cancel it.
- To pass the request to a business only after you explicitly authorize that contact.
- To keep the service working and safe, including abuse and rate limiting.
We do not sell personal information, and we do not use your messages to train models.
Who it is shared with
- AI assistants
- When you use Dispatched inside ChatGPT, Claude, Gemini, or another assistant, that platform processes your conversation under its own privacy policy before it reaches us. We receive only the tool call. We do not send your data back to a platform beyond the tool result it asked for.
- Connected AI clients
- A connected client may read or change only the Dispatched context you explicitly saved, and only while its scoped grant remains active. Revoking the client ends that access immediately. We do not import the client’s chat history.
- Home service businesses
- Only after you explicitly authorize contact, and only the minimum a business needs to respond: the service, the postal code, your preferred and backup times, the job description, and a way to reach you. We do not share a full street address, because we do not collect one.
- Model provider
- The first party conversation on dispatchedhome.com uses OpenAI to generate replies. We may use Anthropic when we change our configured AI provider. The selected provider processes your messages and the tool results needed to answer them. AI-assisted website import and booking workflows send the relevant website content to the selected provider. OpenAI requests are configured not to store Responses API results; the provider may still retain data under its API data policies.
- Infrastructure
- Hosting is Vercel; the database is Supabase (Postgres). Both process data on our instructions.
- Legal
- If the law requires it, or to investigate abuse, fraud, or a threat to safety.
Identifiers
A request gets an opaque identifier such as dreq_ followed by random characters, plus a secret status token that only you receive. The link that carries both is the only copy: anyone holding it can view or cancel that request, and without it the request cannot be read, even by us through the public interface. Transactions and offers use similar opaque identifiers. None of these identifiers encode who you are.
You do not need to sign in to use Dispatched. When account continuity is enabled, a successfully saved request may create a provisional browser account. It does not identify you as a person or permit access from another device or AI client.
How long we keep it
- Conversations
- The first party conversation is not stored as a durable transcript tied to you. It lives in your browser for the session and is replayed to the model to continue the thread.
- Appointment requests
- Kept while the request is open, which is 72 hours unless you cancel earlier, and then retained as a closed record so that status stays truthful and the same request cannot be created twice. Thirty days after a request closes, the personal details in it are automatically erased: the email address, name, phone number, and job description are redacted and the status token is rotated, leaving only a non-identifying record that the request existed.
- Provisional accounts
- An unclaimed provisional account is anonymized 30 days after it is created. Its saved context and identity link are removed while opaque request references remain so request and transaction history stays internally consistent.
- Claimed accounts and saved context
- Kept until you ask us to delete the account. A deletion request immediately ends account access and revokes connected clients. After we verify the request, we delete saved context and unresolved conflicts, remove the authentication identity, anonymize the account record, and confirm completion within 30 calendar days. Opaque references may remain where needed to preserve request and transaction integrity or prevent abuse; they do not contain saved context values.
- Provider applications
- 180 days, then purged automatically. See the provider intake notice.
- Provider portal businesses
- Kept while the business is in use. A business nobody has signed in to for 180 days that was never filled in and has no connected system is removed automatically; a deletion an owner requests is completed within 30 calendar days. See the provider portal notice.
- Server logs
- Kept for a short operational period by our hosting provider and then rotated out.
Security
- Everything is served over HTTPS.
- Secrets stay server side. No key, token, or database credential is exposed to the browser or to an AI assistant.
- Consequential actions require explicit authorization material and an idempotency key, so a repeated call cannot act twice.
- Access to the production database is limited to Dispatched server side code.
- Status tokens are random, are shown once, and are the only way to read a request.
No service can promise perfect security. If we discover a breach affecting your information, we will act on it and notify affected people where the law requires.
Your choices
You can cancel any request yourself from its status link. With an account, you can turn personalization off, edit or forget saved facts, revoke a connected AI client, sign out, or request account deletion. To ask what we hold about you, to correct it, or to have it deleted, write to privacy@dispatchedhome.com. The founder-operated support queue reads that mailbox. We acknowledge a verified request within five business days and complete it within 30 calendar days. Include the request or account identifier if you have one, because without it we may not be able to find the record.
Depending on where you live you may have rights to access, correct, delete, or port your information, and to object to certain processing. We honour these requests through the same address and do not charge for them.
Children
Dispatched is for adults arranging work on a home. It is not directed at children under 13, and we do not knowingly collect their information.
Changes
If this notice changes materially we publish a new version number and effective date here. The version in force is shown at the top of this page.
Contact
Dispatched, privacy@dispatchedhome.com. For anything else, see support.