Privacy & data handling

Know what you share.

This notice covers Mottobits, Dotnetguides, and our ChatGPT Sites inquiry service. It explains what the website collects, where it goes, and how to ask about your information.

Updated

Project inquiries

The form asks for your name, email, selected service, project brief, desired outcome, and monthly team budget or total initial-engagement budget, or acknowledgement of the applicable $8,000 USD minimum. Company, timing, and current-system details are optional. Mottobits uses this information to assess fit, understand scope, and discuss your request. Keep initial details high level. Do not include passwords, access tokens, production records, health information, or confidential customer data.

Submitting requires your agreement to store the request and use your details for that conversation. The saved record includes your answers, a request reference, submission time, campaign category, consent-notice version, handling status, and any associated AI review reference. Inquiries are stored in the ChatGPT Sites database backed by Cloudflare D1. They are accessible through private site administration, with no public inquiry-read endpoint. The form confirms receipt only after saving succeeds; it does not send a confirmation email, contact a CRM, or subscribe you to a mailing list.

Optional AI brief review

When this feature is enabled, an unchecked option lets you choose AI review. Only if you select it does continuing send your service, brief, budget choice, and desired outcome to our inquiry service for review, with a reference and spam-check field. Budget and outcome are checked there. Only the service and brief are passed as project content to TypeSafe AI’s Jev model through Vercel AI Gateway; the separate name, email, company, timing, outcome, system-details, and budget fields are not passed to the model. Anything you put in the brief itself is included.

If you opt in, review happens before the final inquiry-storage consent step. A review record can remain even if you never submit an inquiry. The model classifies the brief as ready, needing clarification, or likely noise; it does not rewrite your words. A clarification can be submitted unchanged. A likely-noise result asks you to revise the brief before proceeding. All saved inquiries still need Mottobits’ review, and no automated result agrees to an engagement. If review is unavailable, ordinary submission remains possible. Leave AI review unselected to continue with ordinary intake without a model call.

Review records retain a reference, time, service, one-way fingerprint of the service and brief, review state, assessment, predefined follow-up question, and available model usage and cost metadata. They do not duplicate your raw brief or separate contact fields. The integration requires a no-model-training route, but does not request zero retention. Vercel may route Jev requests through DigitalOcean to TypeSafe AI. Provider operation, security, and legal retention can still apply. See Vercel’s data-retention controls, DigitalOcean’s data-privacy information, and TypeSafe AI’s data-processing terms.

Website requests and service providers

Vercel hosts the main websites. OpenAI’s ChatGPT Sites and Cloudflare provide the inquiry service, database, and optional measurement counters. These services receive request metadata, such as IP addresses and request headers, to deliver and protect the website. Their operational logs are separate from our application database. Our application error logs use operational identifiers and error categories rather than raw briefs or contact details. Hosting and model services may process information outside your country; this website does not promise a particular storage region.

Optional website measurement and cookies

With no saved choice, optional measurement starts only after a fresh, uncached request to our main website confirms that Vercel reports a US request. Other or unknown countries, unavailable policy checks, and previews remain opt-in. IP-based country information is approximate request information, not proof of residence or nationality. The policy response does not include your IP address or geographic fields, and our application does not store those fields.

You can turn measurement off, or explicitly allow it, using the sitewide measurement controls. We keep that choice in session storage for this browser tab, separately on each website origin. Existing refusals remain in force. Before any optional collection, a fixed temporary session-storage check confirms that the tab can save a refusal; successful checks remove that value immediately. A failed preference write keeps measurement off for the page. Global Privacy Control, Do Not Track, declared browser automation, unavailable session storage, and our QA/test setting keep measurement off. Those browser checks run before requesting the measurement policy. The QA opt-out also stays in this tab’s session storage. Explicit preview mode shows a notice and carries this exclusion through links between our websites, including links opened in a new tab.

When permitted, the browser sends fixed categories for pages, services, published use cases, internal campaign context, calls to action, downloads, form steps, validation failures, submission attempts, and visible receipts. The event identifies its basis as explicit opt-in or default-permitted; a default is not recorded as your consent. Our same-origin Vercel endpoint checks policy again before forwarding default-permitted events to the existing Sites service. Only sanitized category data is forwarded; we do not copy your IP, geographic headers, cookies, full URL, or raw referrer into that application request. The public collector’s origin, site, and basis assertions are not authenticated proof, so fabricated requests can pollute counts. Reports are not proof of genuine people or demand.

After measurement permission, the first measured view in a loaded document uses a coarse referrer hint: internal, another owned website, external, search-provider hint, or direct-or-unknown. The browser briefly classifies its document referrer without sending or retaining the URL, hostname, path, query, or fragment. A search-provider hint uses a small explicit domain list and can include non-search navigation; it does not prove organic search or a person. Missing or suppressed referrers remain direct-or-unknown. Later measured views within the same loaded document are internal. Vercel adds only a Mottobits or Dotnetguides website label from the validated request origin.

Event payloads contain no form answers, names, email addresses, full page URLs, raw referrer URLs or hostnames, IP addresses, or browser profiles. Each event receives a fresh random UUID for 24-hour deduplication; it is not a visitor, session, or journey identifier. New v4 browser reports retain independent daily category counts and coarse UTC dates, separated by website and measurement basis. Earlier v3 observations and accepted-inquiry counts remain in a separate cohort, without added website or referrer claims. They do not retain joined hourly dimensions or precise observation timestamps. Separate operational deduplication and rate-limit tables retain expiry times. Low-volume counts can still be distinctive; we do not promise absolute anonymity. Reports cover today and the preceding 89 UTC days and do not establish unique visitors, linked journeys, conversion rates, or statistical significance.

Accepted-inquiry measurement requires a separate explicit optional measurement choice. A default-permitted browser receipt count does not claim server acceptance. Inquiry storage, its operational saved-request count, and optional AI-review consent remain separate and do not depend on default measurement. Campaign categories describe internal link or page context, not where a visitor came from; the default category is unattributed and does not identify organic search. These categories can travel in links and be saved with an inquiry. Measurement requests omit browser credentials and referrers. Our application does not set analytics cookies or load Google Analytics or advertising scripts. The separate short-lived tab token described below is the only optional reusable measurement identifier; category events and inquiries never contain it.

We separately estimate short-lived tab visits after the same measurement permission and exclusions. Your tab stores a cryptographically random token in session storage on that website origin. It is valid for at most 30 minutes and only within the same UTC date; the next eligible page-view attempt replaces an expired token. There is no background timer or idle-tab deletion. Turning measurement off removes it when storage permits, and the website does not deliberately share it across origins. Browsers can copy session storage into a duplicated or opener-created tab, so those tabs may initially count together. Independently opened tabs and activity after expiry can count again.

The token goes only to a separate visit-count endpoint with the measurement basis. The main website adds its validated site label. The collector uses it only to deduplicate that site’s UTC-day estimate across both bases; the first admitted basis owns the count. It never joins this token to page categories, CTA clicks, form events, inquiries, contact details, referrers, or campaigns. Aggregate visit rows contain only UTC date, website, measurement basis, and count. No fingerprint, IP-derived identity, browser profile, cross-site identity, or person-verification claim is added. These estimates are not unique visitors, humans, buyer counts, or conversion rates. Privacy exclusions, copied tabs, clock changes, traffic caps, and delivery failures affect coverage. Old category counts are not backfilled into this new estimate.

Abuse prevention

The inquiry service uses an hourly changing hash derived from the request IP address to limit repeated inquiry submissions. Raw IP addresses are not stored in that application table. The counters expire at the end of the hour and expired rows are deleted when further requests trigger cleanup; expiration is not an immediate deletion guarantee. The optional measurement counters do not use IP-derived identifiers. Hosting providers can maintain separate operational and security logs containing IP addresses and request headers.

Browser tools

The readiness assessment and agent kit process your choices locally, without uploading those answers, saving them to browser storage, or calling a model. Copying and downloading are your choice; a download increments an optional measurement counter only when your explicit choice or the verified-US default policy permits it.

Where the experimental browser WebMCP capability is supported and enabled, read-only tools expose published services and Fieldnotes. Search terms and note IDs stay in the browser and are not added to campaign events. Loading the public catalog makes an ordinary website request. The tools cannot read inquiries or submit forms. A browser agent or any service you copy results into has its own data handling.

How long information remains

There is currently no automatic deletion schedule for saved inquiries or AI review records; they can remain until Mottobits removes them through site administration. Optional measurement uses 24-hour event-UUID deduplication and a 90 UTC-day reporting window for v3 and v4 daily totals. Prior v2 hourly and daily tables remain separate from new reports; no new observations are added to them. Expired v4 rows are removed in bounded batches on later admitted v4 events. V3 expiry cleanup also runs on explicitly opted-in inquiry submissions. Reports exclude expired totals even before cleanup. Idle rows can remain until later cleanup; provider backups can remain separately. Visit admission state expires at the next UTC midnight, no more than 24 hours after admission. Visit reports cover today and the preceding 89 UTC dates. Later valid visit requests remove at most 100 expired rows from each visit table per attempt; Requests recognized as duplicates before the transaction skip cleanup; racing duplicate requests can still run a bounded cleanup batch. This is not a scheduled deletion job: idle rows can remain, and reports exclude expired totals before physical cleanup. This is not a guarantee of immediate physical deletion at the expiry time. Legacy daily campaign totals collected before this opt-in system have no automatic deletion schedule. Provider logs and backups follow the providers’ separate retention arrangements. Requesting deletion from Mottobits does not itself erase a provider’s retained records. Ask about the information you shared using the route below.

Your choices and privacy requests

Contact Adnan on X about privacy to ask about access, correction, or deletion of information you have shared, or to withdraw your agreement to further inquiry follow-up. You do not need to submit a new project inquiry or meet the project budget minimum to make a privacy request. Ask for a private way to discuss your request; do not post your email, project details, or request reference publicly.

Depending on the law that applies, you may also have rights to restrict or object to processing, obtain a portable copy, or complain to a data-protection authority. Some information may need to be retained where the law requires or permits it. Withdrawing consent does not undo processing that already took place. Identity may need to be verified before information is disclosed or changed.

Email and public-profile providers process anything you send under their own policies. The website cannot tell whether you sent a message through those links. Project-specific confidentiality, access, and permitted AI use are agreed separately before client work begins.