Intellisea is a marketing platform run by AI agents. Those agents can research, write, and — where you allow it — publish to your own website, your Facebook Page, your Instagram professional account, your LinkedIn company Page and your Google Business Profile, and manage your advertising campaigns on Google Ads, Meta and LinkedIn. This page explains, in plain terms, what that means in practice: what the software does on its own, what it will not do without you, how AI-generated material is identified, and what happens to your data.
This page sits alongside our Terms & Conditions, which remain the governing contract. Where this page summarises something the Terms state more precisely, the Terms prevail.
Every agent in Intellisea is software. The CEO Agent, the Director agents and the Specialist agents are not people, and no human reviews your conversations with them in the ordinary course of using the platform. When you ask a question in chat, an AI model produces the answer.
We say this plainly because you are entitled to know it, and because EU law now requires it: Article 50(1) of the EU AI Act obliges providers of systems that interact directly with people to make the AI nature of the system apparent.
When you connect an integration, you are granting an automated system standing permission to act in your name. The scope of that permission depends on which integrations you connect and how you configure them. Broadly:
Scheduled work runs whether or not you are logged in. A scheduler checks for due work every few minutes and executes it. This is the point of the product, and it is also the thing to understand clearly before connecting an account.
Generalities are not much use when you are deciding whether to hand over a live account, so here is the concrete list of what an agent can do on each connected platform without a person present at the moment it happens. Everything here is an action you have to connect an account to enable, and everything here can be put behind an approval step (section 3) or revoked (section 7).
Two things follow from this that are worth saying out loud. Content published this way appears as your Page, your company, your website — not as Intellisea, and not marked as written by software unless you mark it. And replies to comments and to Messenger conversations may be composed by an AI agent and sent to a member of the public who is talking to your business. If you would not be comfortable with that, put those channels behind approval or do not connect them.
You decide, and the default is that approval is off. Intellisea includes an approval workflow for content that would be published outside the platform. It is opt-in: when you create a campaign there is a single setting — "require approval for external publishing" — which starts switched off. Left off, external publishing happens automatically when scheduled. Switched on, you nominate who approves, and you can do that for everything at once or per type of action (blog posts, Facebook and Instagram posts, LinkedIn posts, Google Business Profile posts, advertising changes, email), and override it on an individual item.
With approval on, the nominated person is sent the actual draft — over Slack, Microsoft Teams or email — before it goes anywhere, and the item is held until they answer. Approving releases it; holding it returns it to drafts and it never publishes. If nobody answers, nothing publishes: an unanswered item stays held indefinitely rather than going out by default.
This is the most consequential setting in the product, so we want to be direct about the consequence. With approval on, a person on your team sees each item before it goes live, and that person's review is the human editorial control over what is published in your name. With approval off, content researched, written and scheduled by AI is published to your website or social accounts without human editorial review.
That distinction is not only ours. Under Article 50(4) of the EU AI Act, someone publishing AI-generated text to inform the public on matters of public interest must disclose that it is AI-generated unless the content has undergone human review and a person or organisation holds editorial responsibility for it. Turning approval off is what removes that review. If you operate in or publish into the EU, keep this in mind when configuring it.
Sections 4.1, 4.4 and 9.2 of our Terms cover the same ground contractually.
Material produced by Intellisea is AI-generated: written content, images, voiceover audio, and AI spokesperson video. Inside the platform, generated items are identifiable as such.
Once you publish that material to your own website or social accounts, you are the publisher, and any disclosure or labelling obligation attaching to the published content is yours. We provide the tools; we cannot make the disclosure on your behalf on a platform we do not control.
Two cases deserve particular care:
A further EU AI Act requirement — Article 50(2), machine-readable marking of synthetic output — has a transition period that runs to 2 December 2026 for systems already on the market. We will update this page as our implementation of it lands.
AI output can be wrong. It can state a fact that is not true, cite something that does not exist, misread data, or produce a claim about your business that you cannot substantiate. This is a property of the technology, not a defect we expect to eliminate, and it is why the approval workflow exists.
Review AI output before relying on it, and review it particularly carefully where it makes a factual or comparative claim, quotes a statistic, or touches a regulated subject — health, legal, financial, or anything where a claim carries legal weight.
Responsibility for what is published in your name sits with you. If you have disabled approval, it still sits with you. Sections 4.1 and 9.2 of our Terms set this out.
Your business data, your conversations with the agents, and the content you create in Intellisea are never used to train or fine-tune AI models — not ours, and not any model provider's. This applies to everything you put into the platform and everything the platform reads from your connected accounts.
This is contractual, not merely a preference: every model provider we use is engaged under terms that bar training on data we send them. It is stated once here, and section 3 of our Privacy Policy and our Subprocessors page point back to it rather than restating it.
We do analyse aggregated, anonymised usage patterns — which features are used, where the platform is slow — to improve the product. That analysis does not involve your business content.
Everything you grant, you can withdraw:
Inference runs on models operated by third parties. Rather than keep a second list here that will drift out of date, we maintain one: see the Subprocessors page for who they are, what each one processes, and where.
Anything on this page: privacy@intellisea.com.