Frequently asked questions

ziptie.ai tracks how AI search engines talk about your brand and finds the gaps worth closing. We monitor ChatGPT, Google AI Overviews, Perplexity, Google AI Mode, Microsoft Copilot, Bing AI Overview, and Google Gemini - capturing mentions, sentiment, citations, full responses, and competitor recommendations - then turn that data into ranked next steps.

What you get:

• Answers captured from the live consumer products across all 7 major AI engines (not model API stubs)
• An AI-powered Project Wizard that suggests the right prompts in minutes
• AI Success Scores that prioritize what to optimize first
• A content generation module that drafts AI-search-ready pieces with a built-in fact-check loop (add-on)
• Public REST and MCP access, so AI agents can query your data directly (add-on)
• Coverage in nearly every country, with the wizard naming any engine your country cannot serve before you commit

One loop in three steps: discover what the engines say, prepare the change, verify what followed.

1. Discover. What do the engines say, and why? The Project Wizard interviews you, researches your business, and proposes prompts written the way your buyers ask. Each project runs its prompts in the country you pick, on the engines your workspace subscription includes, on its own cadence: daily, weekly or monthly. ziptie.ai keeps the full answer and the sources behind it, and reads whether you are named or cited, and how you are portrayed.
2. Prepare. What should we change? The Action Center turns those answers into one queue per project, ordered by severity, then type, then age, and each item links to the screen where you do the work. With the Content Generation add-on, a gap becomes a brief and then a draft, written from material you supply.
3. Verify. Did the answer move? Record the URL where a piece went live and see which engines cite it, mark the date on your AI Success Score trend, and read Search Console clicks and impressions alongside. You publish the change where you always do; ziptie.ai never touches your site.

Then the loop starts again: new prompts to track, fresh gaps in the queue, the next brief.

The core modules ship on every plan, each built to answer one question well:

• Overview Dashboard - your visibility at a glance
• Action Center - one ranked queue per project of what to look at next, each item deep-linked to the screen where you would act on it and ordered by how serious the signal is rather than by how far a number moved
• Competitors - who AI recommends instead of you, and why
• Prompts - the queries that move your business, monitored on schedule
• URLs - every page AI cites, every time it cites it
• LLM Trends - performance shifts across engines, side by side
• Data Explorer - slices of one project side by side, such as branded against unbranded prompts, or you against a competitor on one engine, each series with its own filters, read as trends or as a period snapshot, with every number opening the answers behind it
• Search Console - Google Search Console connects via OAuth today, bringing clicks, impressions, CTR, average position, and your top queries and pages alongside your AI search data. Connecting backfills the last 30 days and then keeps pulling new days forward, so it builds a running record from the day you connect rather than importing Google's full history
• Exports - CSV and Excel exports of Prompts list, Prompt detail, AI answers (all prompts), Competitors, URLs, UGC impact, Topic clusters, LLM trends, Data Explorer chart data, Data Explorer answers, Search Console and Overview, plus a PDF snapshot report, generated on demand and downloadable for 14 days. The UGC impact and Topic clusters exports come with the UGC Impact Analysis add-on. Action Center and Content Generation have no export
• Project Wizard - guided chat onboarding that proposes the right prompts from your site
• Team & roles - unlimited teammates with no per-seat fee, on Owner, Admin, User, or Guest workspace roles, with per-project Editor or Viewer access enforced on the server rather than hidden in the interface

Three optional add-ons extend the platform:

• UGC Impact Analysis - the community platforms shaping AI answers, ranked by citations and trend, with a deep read of every cited Reddit thread and YouTube video, topic clusters, a topic graph, and suggested prompts
• Content Generation - briefs and full pieces engineered for AI search, with a built-in fact-check and revision loop
• Public API & MCP - REST and Model Context Protocol surface for BI tools, automations, and AI agents

More detail in the ziptie.ai main features and what makes ziptie.ai unique.

SEO and content teams who need to understand and influence how AI search engines describe their brand and products. Agencies use ziptie.ai to report on AI visibility for their clients. In-house teams use it to brief content writers and prioritize site changes. Founders use it to see whether AI is recommending them - or quietly handing customers to a competitor.

ziptie.ai is built for AI search end to end: the data model, the metrics, the scoring and the workflow. It sits beside your SEO suite rather than replacing it.

• Specialized depth. Answers captured from the live products, an AI Success Score per engine, a deep read of the Reddit threads and YouTube videos the engines cite (UGC Impact Analysis), and content generation with a verification pass on every draft.
• Built for action. Optimization workflows, not just dashboards. You finish a session knowing what to change.
• Agent-ready. With the Public API & MCP add-on, a REST API and an MCP server let Claude Desktop, Cursor, and custom AI agents query your ziptie.ai data the same way your dashboards do.

Minutes for a single project via the Project Wizard. An hour for a careful multi-brand or multi-region rollout. The wizard reads your site, proposes prompts, configures your country and language, and you're collecting data the same day. Nothing on your site has to be wired up first.

No. No tracking tag, no DNS record, no code change. ziptie.ai observes what AI engines publicly say about your brand; nothing on your side has to be modified to start. The only optional integration is Google Search Console, read-only and connected through standard OAuth so you can see classic search performance alongside your AI search data.

Yes. Set up separate projects for each brand, product line, or business unit. Each project has its own prompts, competitors, country, and language - so you can compare performance side by side without mixing data. Agencies typically run one project per client. In-house teams often run one per product line.

Prompts and competitors are added through the UI - the Project Wizard suggests an initial set automatically from your site, and you can extend or replace the list any time. URLs you want measured are added in the URLs module and linked to the prompts they belong to. Pausing a prompt stops new measurements while keeping every past result; deleting one is permanent and takes its recorded results with it, and deleting a URL removes it from tracking and from the project's per-URL history.

Yes. Country, language, and check frequency live in project settings and are editable any time. Engine coverage sits on the workspace subscription, so adding or dropping an engine goes through Workspace settings → Subscription and adjusts your price. New measurements pick up the new scope; historical data stays intact under the previous scope, so you keep a clean record of what was measured when.

A workspace contains one or more projects and any number of members. There are no seat limits and no per-user charge on any plan. Each member holds one of four roles (Guest, User, Admin, Owner), and a User or Guest can be given Editor or Viewer access to individual projects. In-house teams usually keep every brand in one workspace, with a project per brand or product line.

Agencies have two workable setups, and the choice turns on how firmly client data has to be kept apart.

• One workspace for the whole roster. Each client is a project, every project draws on one shared prompt pool, and each client joins as a Guest with Viewer access to their own project. Access starts open: Owners and Admins reach every project by role, so a workspace narrows only as an Admin sets project access for each User and Guest.
• One workspace per client. Each client gets its own members, subscription and add-ons. It is the hard boundary when two clients must never see each other's numbers, and the trade-off is that capacity cannot move between clients.

How agencies use ziptie.ai sets the two side by side.

Mostly self-serve. The Project Wizard makes setup viable in minutes - most teams never need help. Larger teams can request a guided onboarding session by contacting us first; useful for multi-brand rollouts and large agency teams.

For every prompt, on every engine, ziptie.ai captures:

• Brand mentions - whether your brand appears in the AI answer
• Sentiment - how positively or negatively the answer describes you
• Citations - which of your URLs the engine links to
• Competitor share - which other brands the engine recommends instead
• AI Overview coverage - whether an AI answer is generated at all
• Full response text - the complete AI-generated answer, archived
• Source attribution - every domain the engine cited to build the answer

You can view performance at three levels:

• Project - the high-level picture across all prompts and engines
• Category - groups of prompts by tag or topic, manually or automatically tagged
• Prompt - the individual query, with the full AI response from each engine

Filters - date range, engine, tag - are respected across the dashboard, so a slice like "comparison prompts, last 30 days, ChatGPT only" stays consistent as you move between modules. To see several slices side by side rather than one at a time, use the Data Explorer, where each series carries its own filters. Country and language are set per project, so a locale comparison is a project-against-project view, and the Data Explorer, which works inside one project, does not compare countries either. Custom tags let you carve the data the way your team actually thinks.

Yes, in the Data Explorer, a core module on every plan. A view holds up to eight series. Each series is a slice of one project's answers with its own filters and its own subject: your brand, your domain, a competitor brand, or a domain the engines cite. Prompt filters choose which prompts count: tags, excluded tags, specific prompts, words in the prompt, tracked URLs, intent, funnel stage, when a prompt was added, and whether it is active or paused. Answer filters choose which answers count: engine, whether your brand or a competitor is named, sentiment, prominence, cited domains, the kind of source cited, and shopping results. "Branded against unbranded prompts, unbranded prompts on Google AI Mode alone, and the same three for a competitor" is one view.

Pick up to six metrics. Trend draws one chart per metric with every series on it. Snapshot sets the series against the metrics for one period, shows each value's change since the period before, and compares each series with a baseline series you choose: Likely a real difference, could be noise, or not enough data. A value resting on fewer than 30 answers is marked as a low sample and gets no verdict. When two series can share answers, such as all prompts against branded prompts, the verdict is marked indicative, because the test behind it assumes the two samples are independent. It tells you whether a gap is likely real, never what caused it. A series can also be shifted to the previous period, a set number of days back, or the same period last year, drawing on the checks the project has already run.

Click any point or cell to open the answers behind that number. Split a series by engine, tag, intent or funnel stage in one step. Save a series' filters as a filter preset, or the whole layout as a saved view, and everyone on the project can load it, from the menu or from a link. Guests and viewers can use saved views but cannot save their own. The numbers, and the answers behind every series, export from the same page.

Intent (informational, commercial, transactional or navigational) and funnel stage (awareness, consideration or decision) are labels ziptie.ai assigns to each prompt automatically with AI, from its wording and your project's context. Anyone who can edit the project can correct a label on the Prompts page, and a corrected label is never overwritten automatically. A prompt not yet labelled shows as Unclassified. The labels describe how a prompt is worded, not how often people ask it. The Data Explorer works inside one project, so it does not compare countries or projects, and it has no position-in-the-answer metric.

Prompt capacity depends on your plan; see pricing for current numbers. Prompts can be added, paused, resumed, retagged, or deleted at any point - adds start measuring on the next run, pausing stops new measurements while keeping every past result, and deleting removes the prompt together with its stored results. Refresh frequency (daily, weekly, or monthly) is configured per project.

Let the Project Wizard suggest them - it reads your site and proposes the right mix automatically. If you'd rather build the list yourself, four types cover most use cases:

• Category prompts - "best form builder", "top CRM for startups". Tests whether AI recommends you in your category.
• Sentiment prompts - "is acme good", "should I trust acme". Tests how AI describes you.
• Comparison prompts - "acme vs competitor-b", "alternatives to competitor-a". Tests where you sit against rivals.
• Persona prompts - layer audience or use-case onto the above: "best form builder for HR teams".

Each tracked prompt gets a single 0-100 score that summarizes how well you're performing on that prompt across AI engines. The score is built from three signals captured for every prompt:

• Mentions rate - how often your brand appears in the AI answer
• Mentions sentiment - how positively the answer describes you when you do appear
• Citations rate - how often the AI engine links to your URLs as a source

Sort by score and you have your optimization queue - work on what's underperforming first.

Two steps: decide what to change, then see what followed. The Content module (an add-on) helps you plan content updates for the prompts you care about - AI-search briefs and full drafts, with a fact-verification pass. You apply the changes where you publish, and read what the next cycle shows.

The Competitors module ranks every brand AI engines recommend in your space by mention share. Drill into any competitor to see exactly which prompts they win on, what sources back them, and which gaps you can close. You see where you're losing ground and where you can move fast.

ziptie.ai has an in-app notification center - a bell in the app with a live stream behind it - and it fires on a fixed list of events:

• your project's first check finishing, and each scheduled check finishing after that
• an engine that couldn't be checked this cycle, named engine by engine, with a retry on the next cycle
• a Google Search Console connection that has stopped working
• an export finishing and being ready to download
• content-module jobs finishing or failing, if you run that add-on
• workspace and billing events: invitations, role changes, ownership transfer, trial ending, payment succeeded or failed, cancellation

Some of that reaches you by email as well as in the bell. When a workspace's scheduled checks finish and are tallied for a day, ziptie.ai builds one digest for that workspace covering that day, then sends it to each member separately, with the projects that completed, the prompts checked, the citations and brand mentions found, and the change since that project's previous check. It is not a fixed daily send: every project runs on its own cadence - daily, weekly, or monthly - so a workspace whose projects run weekly gets a digest in the weeks its checks land, and each digest covers that single day rather than rolling a period up. It clears two gates before it sends: a cutoff hour it will not go out before, and a settling window that starts when the last check of that day lands, so it reaches you after that day's checks have settled, from early morning UTC onward, rather than at a promised minute. Your copy is filtered to the projects you have access to, it carries no attachment, and a workspace with more than fifty projects in one day gets them across consecutive emails. A project's very first check is not in it - that one gets its own "your first results are ready" note - and the first time a project appears there is nothing to compare against, so ziptie.ai shows a placeholder rather than a made-up change. Which ziptie.ai emails you get is yours to set: Settings > Notifications lists them as separate groups - check results, check problems, briefs and pieces, membership changes, payment receipts, integrations, exports, and support replies - each on by default and each switchable on its own, and every one of them carries a one-click unsubscribe. The same screen prints the list of messages we always send whatever you choose, such as an invitation, a failed payment, or a trial ending, so nothing is hidden behind a toggle you did not know about.

What ziptie.ai still will not do, so you can plan around it. There are no user-configurable alerts: you cannot set a rule like "email me if my mentions rate drops below 40%". The Good and Fair targets you set in project settings color the charts and drive the plain-English read-outs; they don't fire anything. And exports are produced on request rather than on a schedule - there is no recurring report, and no way to have a CSV, Excel, or PDF mailed out on a cadence.

Every plan supports CSV and Excel exports, taken module by module, plus a PDF snapshot report and public share links - non-guessable URLs that let stakeholders download a snapshot without logging in. Twelve module exports are on offer in CSV and Excel: Prompts list, Prompt detail, AI answers (all prompts), Competitors, URLs, UGC impact, Topic clusters, LLM trends, Data Explorer chart data, Data Explorer answers, Search Console and Overview, plus the PDF snapshot. The UGC impact and Topic clusters exports come with the UGC Impact Analysis add-on. The two Data Explorer exports start from the view you have open: Data Explorer chart data carries its numbers, with a sheet in the Excel file that spells out what each series measures and filters, and Data Explorer answers carries the answers behind every series. Generated files stay downloadable for 14 days, and a workspace can have five exports generating at once - enough for a reporting run, worth knowing if you script one. There is no JSON, XML, or Google Sheets export, and no way to schedule an export or have one emailed on a cadence. Invite collaborators directly to your workspace with a limited-access role to give them ongoing dashboard access without project-edit rights. The API & MCP add-on (covered under the Plans tab) exposes the same dataset programmatically for BI pipelines, custom dashboards, and AI agents.

ziptie.ai captures each prompt from the live consumer products rather than a model API, scoped to the country and language you select, with no logged-in account history influencing the answer. A model API does not return what people see in the consumer product, so the responses we capture are the responses your customers actually get.

A single AI answer is a sample, not a fact - the same prompt can vary cycle to cycle. ziptie.ai captures and archives every measurement, so trends are built across many measurements over time rather than depending on a single snapshot. A real change in visibility separates itself from random variance over a few cycles; one-off outliers stay visible as the underlying data point but don't move the longer trend. You can drill into any score and see the actual responses that produced it.

Because the order inside an AI answer does not hold still. Ask the same question twice and the brands come back in a different order, often as a different set. A position read from one answer describes that one run and nothing more, so a "rank in ChatGPT" would report noise as if it were a result.

ziptie.ai reports what the engines do support: how often you are named and cited across many prompts and checks, how you are portrayed when you are named, and how that moves over time. That is why every prompt runs on a schedule rather than once. The only positions ziptie.ai reads from a results page are the classic ones: Search Console's average position, and the organic results that come with Google AI Overviews and Bing AI Overview. The rank on the Competitors page is your place by share of citations or mentions across all your checks, never a position inside one answer.

We match the brand name(s) you configure for the project, with tolerance for capitalization and minor formatting differences. If AI engines describe you under a different name - a parent-company name, a legacy product, a regional brand - add that as an additional brand for the project so the mention is counted correctly.

Each AI answer is scored automatically against the context in which your brand is mentioned: positive, mixed, or negative. The scoring runs on every captured response, and we periodically review samples ourselves to verify quality over time. Each score is linked back to the source answer, so you can verify a disputed result by reading the actual paragraph that produced it.

Every URL an AI engine cites is parsed, normalized to its canonical form, and matched against your domains, your competitors' domains, and a general third-party-source classification. You see citation share at the answer, prompt, project, and source-domain level, so it's easy to spot which third-party sites are shaping the answer and where to focus outreach.

AI answers are heavily personalized. If you're logged into ChatGPT, your account history skews the response - ChatGPT may surface your brand because it knows you work there. Google may weight pages you've recently visited. Your location, device, and prior chat context all influence the answer you personally see.

ziptie.ai uses fresh, neutral sessions so the answer we capture is the answer a new customer would see. That's a much more reliable benchmark for the audience you're trying to win. If a result still looks off after that, tell us - every captured answer is archived and we'll investigate the specific sample.

Your history starts at your first check. ziptie.ai cannot show you how AI engines described your brand before you started tracking - nobody publishes an archive of past AI answers, so there is nothing to import. A new project's record begins the day it runs its first check, which is why it's worth adding the prompts you care about early even if you only review them monthly.

From there, every measurement is kept for as long as the workspace is active. The dashboard's date filter offers the last 7, 30, or 90 days, plus a custom range of up to 365 days, so a year is the widest window you can put on screen at once.

Google Search Console is the one place you get history from before you signed up, and it's a partial one: connecting a property pulls the last 30 days of Search Console data, not Google's full 16-month archive, and then keeps pulling new days forward from there.

Nearly every country, with two caveats we would rather state than have you discover.

Country first. Each engine is reached through a provider that serves its own list of territories, so which engines can actually run varies by country. ziptie.ai resolves that before you commit rather than after: the project wizard warns you at the country step which of your engines cannot be checked there, countries no engine can serve are not offered at all, changing a project's country warns you about the engines you would lose, and a pair that cannot be checked reads as "not available" in the grids instead of quietly as a zero. A small number of countries are withheld from the picker outright, China among them.

Language second. Google AI Overviews, Google AI Mode and Bing AI Overview accept a language parameter, so a project's language is sent to them with the query. The chat engines do not take one: they are asked in the language your prompt is written in, from a request configured to originate in the country you picked. Write the prompt in the language you want the answer in and the two line up. Run side-by-side projects to compare Germany vs. Austria or Brazil vs. Portugal, and US English vs. UK English on the three engines that take a language.

If an engine is unavailable for a cycle, the measurement for that engine on that cycle is recorded as missing rather than zero, and we retry on the next cycle. A real "no AI answer for this prompt" result is also captured - some prompts don't trigger an AI answer at all, and that's an honest signal you'll see reflected in your data. Engines that aren't included in your subscription aren't shown in the dashboard at all.

Every captured answer is archived with timestamp, engine, country, and language. Click any metric and you see the underlying answers - useful for verifying a surprising score, briefing a content writer, or sharing evidence with stakeholders.

Refresh cadence is your choice: daily, weekly, or monthly, set per project. Weekly is enough for steady monitoring. Daily is useful when you're actively optimizing and want the next cycle to pick up a change within a day of publishing.

Worth saying plainly: there is no "run a check now" button. Checks are driven entirely by the project's cadence - a project is picked up once its last check is a full cycle old - so after you publish a change, the soonest you can see its effect is that project's next cycle. Daily is the fastest cadence available; nothing runs continuously or in real time. The one shortcut: prompts you add on a day the project has already run join that same day's cycle instead of waiting for the next one.

The brand and competitor details you configure, the prompts you tell us to track, the AI responses we capture on your behalf, derived metrics (mentions, sentiment, citations), your account profile, and billing information. ziptie.ai does not ingest your site analytics, your customer lists, or any first-party data from your business. We observe what AI engines publicly say about your brand; nothing is pulled from your internal systems.

Customer data is stored on dedicated server infrastructure in the United States, with encrypted off-site backups in Amazon S3, also in the United States. Traffic between you and ziptie.ai is encrypted in transit over TLS, for the app, the public API and the MCP endpoint alike, with TLS 1.0 and 1.1 disabled.

We would rather be exact about encryption at rest than round it up, because a security reviewer will check. Encrypted at rest today: the off-site backups, your TOTP authenticator seeds, the Google Search Console access and refresh tokens a connected project holds, which are encrypted with a service key before they are written and decrypted only to make a call on your behalf, and any identity-provider client secret you give us for SSO, which is write-only and never returned by any API. Not encrypted at rest today: the primary database volume, and the Redis and configuration stores that sit beside it on the same dedicated hardware we control rather than on shared infrastructure. One consequence of that is worth stating rather than burying, since your reviewer will ask: the keys that decrypt those TOTP seeds and those Search Console tokens live in that configuration store, so that encryption protects the database column rather than the host. Encrypting all three volumes is scheduled work with a written procedure, not an open question. If any of this matters to your review, ask us and we will show you exactly where the line falls.

Access to production systems is restricted to individual engineer accounts using key-based authentication, with no shared logins and no password authentication, and requests are logged. Least privilege is written policy, and recurring access reviews are part of the SOC 2 program described below, whose evidence period opens in December 2026, so treat them as a control we are standing up rather than one with a year of records behind it.

No. Your prompts, the AI responses captured on your behalf, and your account data are used to operate the ziptie.ai monitoring service for you - not to train any model, and not shared with third parties for training purposes. The AI engines we monitor are public consumer products; the data we generate from them on your behalf belongs to you.

Email and password with strong-password enforcement, plus sign-in with Google. Multi-factor authentication is available on every plan and every workspace role - no upgrade required to turn it on - and supports TOTP authenticator apps, email one-time codes, trusted-device shortcuts, and printable recovery codes.

A workspace can also sign in through its own identity provider, over SAML 2.0 or OpenID Connect. Sign-in is routed by email domain, so someone types their work address on the sign-in page and goes straight to your provider with no password field. The workspace Owner sets the connection up from Workspace settings: describe the provider, add its signing certificate if you chose SAML, hand your provider the URLs the page generates, prove you control the email domain with a DNS TXT record, then activate. On SAML you can paste your provider's metadata rather than transcribing it, and certificates rotate without downtime, because every certificate that is not retired, not expired and already valid is trusted at once. The Owner can then require that everyone signing in to the workspace signs in through the provider, with named exemptions for the accounts that must not be locked out.

Be precise about what that requirement covers: interactive browser sign-in, to that one workspace. It does not stop a password sign-in from working in the other workspaces someone belongs to, and it does not reach API tokens, the public REST API or the MCP endpoint, which authenticates with those same tokens. A token proves which project it belongs to and nothing about who is using it, so there is nothing in one to check against your provider. The same screen lists every token that can reach the workspace, with the project it belongs to, so an Owner can revoke it there. One more thing worth knowing before you turn it on: arriving through your provider satisfies our multi-factor requirement on its own and raises no TOTP prompt, so for a member who was on a password plus TOTP it is one factor in place of two, and the strength of the sign-in becomes whatever your provider enforces.

Enterprise SSO is switched on per workspace by us rather than bought from the pricing page, and it is not part of the "Enterprise" preset, which is a starting configuration of engines, prompt volumes and source-impact reporting in the same configurator and nothing to do with authentication. Tell us you need it and we will enable it on yours.

Each workspace has its own members with role-based access: Guest (lowest), User, Admin, Owner (full control). Roles determine who can edit project configuration, manage members, and change billing. There is exactly one Owner per workspace - the role moves by an email-confirmed ownership transfer, not by promotion - and only the Owner can change or cancel the subscription or switch add-ons on or off.

You can also give an individual member per-project access, Editor or Viewer, either when you invite them or later from the Members screen; Owners and Admins reach every project by role. That access is enforced where the data is served rather than only hidden in the interface: a project a member holds no access to is refused when they open it, and the project list is filtered in the query behind it, so the picker and the page agree. Be precise about what that buys you today, though. Workspaces that already exist start open, because their members are Owners and Admins and those roles inherit every project, so the narrowing is something you set as you add members at User or Guest rather than something an existing workspace already has. API tokens sit outside this: a token is bound to a single project and refuses any other, but its reach is governed by issuing and revoking the token, so taking a member's project access away does not shrink the tokens they created. Agencies can work either way: one workspace with each client as a Guest viewer on their own project, which keeps one shared prompt pool, or one workspace per client, which is the cleaner boundary when two clients must never see each other's numbers.

One gap worth naming rather than letting you find it: there is no customer-visible audit log of your members. Nothing in the product shows your workspace which member did what and when, and there is no activity export.

Not as a log you can open, and we would rather say so than let a security questionnaire find it. ziptie.ai keeps no member-activity history inside a workspace: there is no screen showing which member changed a setting, added or removed a prompt, or was granted access to a project, and nothing of that kind to export. A workspace that signs in through its own identity provider gets one narrow exception, and it is not a member log: its SSO settings carry a read-only activity view of what that provider did in the workspace, what happened, whether it succeeded, which provider sent it and when, and there is nothing to export there either. Administrative actions taken by ziptie.ai staff are recorded on our side, and we can produce that record if you ask what we did to your workspace, but it is not a log of your own members.

Three narrower controls exist, and none of them is a substitute for a log. The workspace Owner is notified when a member is promoted to Admin or Owner, unless the Owner made the change or is the person being promoted, and that notice follows the Membership changes notification preference. Anyone can list their own active sessions in user settings and sign the other ones out. And each project's API Access screen lists that project's tokens with their access level, last use and expiry, so an unwanted token can be revoked there.

If a readable activity log is a requirement rather than a preference, tell us before you buy rather than after.

Capturing AI answers means sending your monitored query text out to the engines, so two data providers sit in that path. BrightData submits your queries to ChatGPT, Microsoft Copilot, Perplexity, and Google Gemini and returns the public answers and citations. DataForSEO does the same for Google AI Overviews, Google AI Mode, and Bing AI Overview, and also returns the organic results behind the Competitors tabs. The query text you configure is therefore shared with the provider and with the engine it is submitted to.

Analysis runs through OpenRouter, which routes text to the underlying model providers for sentiment scoring, dashboard insights, UGC Impact analysis, and the Content add-on. Around that: Stripe for payments, Postmark for transactional email, Google for optional sign-in, Cloudflare for DNS and TLS, and encrypted backups in Amazon S3. We publish the full table - what each one receives and what security attestation it holds - in the privacy policy, and we'll tell you plainly that OpenRouter does not currently hold a public attestation, because a security reviewer will find that anyway.

Not yet, and we'd rather say so than let a procurement questionnaire find out. ziptie.ai is running a SOC 2 Type II program covering the Security, Availability, and Confidentiality criteria: the policy set and control matrix are written and being rolled out, the evidence observation period opens in December 2026, and audit fieldwork is targeted for August 2027. There is no SOC 2 report to hand you before then, and we won't describe ourselves as certified, compliant, or audited until there is one.

There is also no contractual uptime commitment. Our Terms say it outright - the subscription does not include a service level, and any such commitment would live in a separate signed agreement - so treat any uptime number you hear as unofficial until it is in a contract. We do run a public status page, at status.ziptie.ai, and it is deliberately hosted away from the platform it reports on so that it stays up when we do not - the live badge in the footer of this page comes from it. Planned maintenance and incidents are posted there and announced as in-app system messages; the notices are written by hand, so the absence of one is not an automated all-clear. The fastest way to reach a human about a live problem is the contact form or, once you're signed in, a support thread.

The personal data within ziptie.ai is limited to the account information you provide (name, email, billing details); we don't process end-user personal data on your behalf, which keeps the data-protection surface small. You can correct your account details in user settings, and delete your account there once any workspace you own has been transferred or closed. Subject access requests and any other data-protection request are handled by contacting us directly.

You keep full access to your dashboard for the remainder of the paid period and can export your monitoring data any time: the twelve CSV and Excel module exports and the PDF snapshot. Action Center items and Content Generation briefs and pieces have no export. After cancellation, your data is retained for a limited grace period in case you reactivate. You can also delete your account yourself from user settings once any workspace you own has been transferred or closed, or contact us to request full deletion.

Yes. A 7-day free trial covering ChatGPT, Google AI Overviews, and Microsoft Copilot, with no card required at signup. The other engines (Perplexity, Google AI Mode, Bing AI Overview, and Google Gemini) unlock on a paid plan. You can downgrade or cancel from workspace settings any time. We don't refund paid subscriptions, so we recommend using the trial to validate ziptie.ai fits your workflow before upgrading.

What's in it: 25 monitored prompts on a daily refresh. The trial's capacity is all daily, so trial projects run daily checks - weekly and monthly cadences appear once you're on a plan that includes them. UGC Impact Analysis is not part of the trial; it is an add-on you switch on when you build a plan.

What isn't: the API and MCP surface stays off for the whole trial. A trial workspace is refused at the API regardless of the add-on, so schedule any integration testing for after you upgrade. Add-ons are chosen as part of the plan you subscribe to rather than bolted onto a running trial, so UGC Impact Analysis and Content Generation start when you add them to a paid subscription, not during the trial. And the trial is once per person - it's available on your first workspace and can't be started again on a second one.

Nothing is tiered: you build your own plan from prompt count, refresh frequency, and engine coverage, with optional add-ons on top. Teammates and projects are unlimited, with no per-user charge. The pricing page shows current numbers. Longer billing periods carry a discount vs. monthly: 3 months -5%, 6 months -10%, annual -20%. For unusual configurations or large prompt volumes, see custom plans below.

These never appear on an invoice:

• Seats, projects and competitors. Unlimited on every plan, with no per-user charge, no fee per brand or client project, and no cap on the competitors you track.
• Exports and share links. CSV, Excel and the PDF snapshot, shared through a link that needs no account.
• Google Search Console. Connect a property you already own, read-only, and see its clicks and impressions next to your AI data.
• Google and Bing organic results. They come with the Google AI Overviews and Bing AI Overview engines at no extra cost.
• The check summary email. It arrives when scheduled checks finish, with the change since the previous check.

What you pay for is the prompts you monitor, how often they run and how many engines run them, plus any add-on you switch on. The pricing page shows the numbers.

Prompts checked, not checks run. Your plan buys a number of prompt slots per refresh frequency. Every monitored prompt holds a slot, and a prompt checked in the current window keeps holding it until that window resets, even after you pause or delete it. A prompt on a daily project takes one daily slot and then runs every day at no further cost. That's why the price scales with cadence rather than with usage.

The slots sit in three separate pools, one per refresh frequency, and a prompt draws from the pool matching its project's cadence: a prompt in a weekly project uses a weekly slot, never a daily one. Each pool resets on its own clock - daily at midnight UTC, weekly on Monday at midnight UTC, monthly on the 1st of the month at midnight UTC. Practical consequences worth knowing before you buy:

• A prompt checked in the current window holds its slot until that window resets, even if you pause it, delete it, or move it to another cadence. Checking the same prompt again inside the window never costs a second slot.
• Pausing a prompt before it has been checked in the current window frees its slot right away, so you can swap it for a different prompt without buying more.
• Re-enabling a paused prompt is free if it already holds a slot this window; re-enabling one that would need a new slot fails when the pool is full - free one up or raise the plan.
• Deleting a whole project follows the same rule per prompt: any of its prompts already checked this window keep holding their slots until that window resets, not until your next renewal.
• When a window resets, paused and deleted prompts stop holding slots, and each pool starts again at your plan's allowance minus the prompts you still monitor. Nothing rolls over and nothing is lost.
• A project can only be set to a cadence your plan has capacity in. Buy no weekly capacity and Weekly won't be offered as an option.

There are no credit packs, top-ups, or overage charges to buy. If you need more room, you change the plan.

Both Google AI Overviews and Bing AI Overview are "two-for-one" data products. When you enable Google AI Overviews on a project, every check also captures Google's top-10 organic results for that query - these populate the Google Organic tab on the Competitors page. Same for Bing AI Overview: enabling it captures Bing's organic results for the Bing Organic tab. You only pay for the AI engine; the organic data comes with it.

Connecting Google Search Console adds your own first-party search analytics - clicks, impressions, CTR, average position, and your top queries and pages - on top of the public SERP capture, via a read-only OAuth integration.

Workspace settings → Subscription, as the workspace Owner. Every plan change offers two options: apply now, which issues a prorated invoice straight away, or apply at the next renewal. Cancellation always takes effect at the end of the period you have already paid for. No retention obstacle, no required call.

US dollars, and only US dollars. Every plan and every add-on is priced and charged in USD - there is no EUR or GBP billing and no local-currency pricing, so if your card is denominated in something else your bank does the conversion.

Listed prices exclude tax. Checkout asks for your billing address, and where VAT or sales tax applies it is calculated from that address and added on top of the price, so you see the full total before you pay.

Buying for a company? Signing up and the free trial don't ask for billing details; you add them when you subscribe. At checkout, tick "I'm purchasing as a business" to enter your company's legal name and VAT number or other business tax ID. We take a valid tax ID into account when calculating tax, and it appears on your invoices. To add or change it later, update your billing details in the billing portal linked from Workspace settings → Subscription; the change applies to invoices issued from then on. Your payment method, your next invoice, and every past invoice live there too.

And to restate it where a buyer will actually look for it: we don't issue refunds. Cancelling stops the next charge and you keep full access until the end of the period you have already paid for; switching off an add-on works the same way. That's the reason the free trial exists - use it to be sure.

Yes. If the standard configurator doesn't fit your prompt volume, check frequency, or engine mix, send us a note with your numbers and we'll quote a plan that does.

A quoted plan is also what carries single sign-on: sign-in through your own identity provider over SAML 2.0 or OpenID Connect. The self-serve configurator doesn't offer it, so tell us if that's part of what you need. Once it's on the plan, your workspace Owner sets the connection up in Settings.

Yes, and there are two gaps worth knowing before you put it in front of a client. ziptie.ai charges for the prompts you monitor and the engines you run them on. Seats, projects and competitors are unlimited, so a new client costs only its prompts. Every project in a workspace draws on one shared pool of daily, weekly and monthly prompts, so capacity moves with your roster.

• A project per client. Each client gets its own prompts, competitors, country and language.
• Client viewers. Invite a client as a Guest with Viewer access to their own project, or give each client a workspace of its own when their data must never meet.
• Share links and the PDF snapshot. Share the PDF snapshot or any export through a link that needs no account. Each link works until the export is 14 days old.
• One API token per project. With the API & MCP add-on, each token reaches one project, so ten clients through the API means ten tokens.

The two gaps, stated plainly. There are no white-label reports: the PDF snapshot and the share page carry the ziptie.ai mark. And there is no cross-client rollup: every client is its own project, and no view adds projects together. How agencies use ziptie.ai covers both setups and what a roster costs.

Yes. The ziptie.ai API & MCP add-on (+$10/mo) exposes every prompt, response, mention, citation, sentiment score, and competitor metric over two surfaces:

• REST API at api.ziptie.ai - for BI pipelines, internal dashboards, or scheduled jobs that poll for visibility changes. Tokens are created per project with a read, write, or admin scope, set to expire in 30, 90, or 365 days or never, and capped at 10 active tokens per project. Default rate limits are 60 requests per minute per token, 600 per minute per workspace, and 50,000 per day per workspace; every response carries your remaining budget in X-RateLimit-* headers, and going over returns a 429 with a Retry-After.
• MCP server at api.ziptie.ai/mcp - same data, same auth, exposed as Model Context Protocol tools. Drop the URL into Cursor, VS Code, Claude Code, or any client that speaks streamable-HTTP MCP; Claude Desktop connects through the standard mcp-remote bridge. Step-by-step guides for each client ship in the docs.

Full reference and OpenAPI spec at docs.ziptie.ai.

Native CSV, Excel, and PDF export plus public share links are available on every plan, no API required. Google Search Console connects via OAuth today.

For product support, use the contact form on our About page. Once you are signed in, every workspace also has a support center where you can search the knowledge base and open a thread with us.

For security reports, email security@ziptie.ai rather than support: a vulnerability report in the support queue is read late and by the wrong person. The same address is published machine-readably at /.well-known/security.txt. We follow responsible-disclosure practices and acknowledge reports within five business days. Tell us the URL or endpoint, the request, and what an attacker gets out of it; for a finding against the REST API, include the X-Request-ID from the response.