Terms of service

Retorik is a service of RAAK.work, a sole proprietorship (eenmanszaak) of Wessel Janse van Rensburg, registered in the Netherlands. "We" in these terms means RAAK.work.

One address serves for contracts, invoices, privacy requests, security reports and complaints. A person reads it. We answer in English and Dutch. That address and the registered address above are also our point of contact for authorities, including for orders under the EU rules on electronic evidence.

Where the service runs. On servers of Scaleway in France and the Netherlands. The AI model runs at Anthropic in the United States under EU standard contractual clauses. Our data processing agreement lists every company involved and the safeguards against access by authorities outside the EU.

What Retorik does. Your comms team invites an executive, uploads what the executive has written and said, and Retorik builds a profile of their voice and ideas. From then on the team and the executive draft posts with that voice, check them, approve them per channel, and publish or schedule them to LinkedIn, X and Bluesky, or export long-form text. Every draft keeps its versions, its review notes, its sources and a record of who wrote, checked and approved what.

What we may change. We improve Retorik continuously. We may add, change or withdraw features; we do not withdraw the core of the service (drafting, review, publishing, export) during a paid term without 60 days’ notice. A change that reduces what you can do with your data is announced 30 days ahead (section 14).

Pilots. A pilot is a fixed period agreed in writing, for a pilot fee, to see whether Retorik works for your executive. During a pilot the service is provided as it is: we tell you about incidents, but we promise no uptime and no support hours. A pilot ends on its end date unless we both agree an order for paid use.

• Retorik is invite-only. You choose who gets an account: administrators, comms staff and executives. Each account is for one named person; do not share sign-ins.
• You are responsible for what your users do in Retorik and for keeping their access current. Remove people who leave. Tell us at info@retorik.pro the moment you think an account is compromised.
• Users sign in with a password or a passkey. Keep devices and inboxes safe: anyone who controls them controls the account.
• Solo executives. You can use Retorik on your own, without a comms team. Then you are the customer and the executive at once: you review your own drafts, and the sign-offs a team would give are yours to keep or to skip. These terms read with "your users" and "your administrators" meaning you. If you use Retorik for your own business or profession, that is all. If you use it as a private person, section 16 adds the consumer rules the law gives you.
• By using Retorik you confirm that you act for an organisation you may bind, for your own business, or for yourself.

• Yours. Everything you and your users put into Retorik, and everything Retorik makes from it for you, is yours: the executive’s writing, the voice profile built from it, the drafts, the review notes, the published posts and the record behind them. We claim no rights to it.
• Our permission. For as long as you use Retorik you allow us to store, process, back up and display that content, and to send it to the companies listed in our data processing agreement, in order to run the service. That permission ends when the content is erased (section 13), except for backups, which are erased on their own schedule.
• Your promise. You have the right to give us that content: the executive’s writing is theirs or your organisation’s, or you have permission for the parts that are not; and where it names other people, you may lawfully use it.
• What we do not do. We do not train any model on your content, ours or anyone else’s. We do not sell it, share it or use it for anything but running Retorik for you. We do not read it, except when needed to fix a fault or to answer your support request, and then only the person who runs Retorik, who is bound to confidentiality (section 8).
• What we keep for ourselves. Retorik itself: the software, the prompts, the design, the way it works. You may use it; you may not copy it, resell it, or build a competing service from it.

• The executive’s own accounts. An executive connects their own LinkedIn, X or Bluesky account to Retorik, and can disconnect at any time. We store the access tokens encrypted and use them for one thing: to publish what your people approved, and to show whether the connection is live. When an account is disconnected we delete the tokens and the profile details the platform gave us; the identifiers of posts already published stay in your record. A post published through Retorik appears on the platform as the executive’s own post, because it is, and the editor shows exactly what will be published before it is.
• Only on your instruction. Nothing is published unless a person in your organisation presses Publish or sets a schedule. Retorik does not post on its own. Where a channel needs approvals first, the approvals are yours to give.
• A link below the post. On channels where a link in the text hurts a post’s reach, Retorik places the reference link in a first reply under the post instead. The editor shows you where the link will go before you publish.
• The platforms’ rules apply. Each platform’s terms bind the executive as its account holder, and you agree that posts published through Retorik must comply with them. We are not responsible when a platform refuses, delays, changes or removes a post, changes its rules or its API, or is unavailable.
• Published is published. Once a post is on a platform, only the platform’s own tools can take it down. Retorik cannot unpublish. Published drafts are locked in Retorik so that the record stays true.

• What Retorik makes. Retorik drafts text from the executive’s own past writing and the ideas in it, using an AI model. A draft can be wrong: a fact, a figure, a quote, a name. Retorik’s checks (the figure check, the writing check, the references) help a reviewer see where to look; they are aids to your judgement, not a guarantee.
• A person decides. Retorik is built so that a person reads, edits and approves every post before it goes out, and the record of who did so is kept per draft. Do not publish anything from Retorik that a competent person in your organisation has not properly read; a glance is not a review. If a draft is changed by the AI after it was reviewed, review it again.
• Your editorial responsibility. Deciding to publish is your editorial decision, and the editorial responsibility for a published post is yours: the organisation’s, or the solo executive’s own. Retorik keeps the record that shows a person reviewed each post; it is yours to show if anyone asks.
• What the EU AI Act asks. Retorik is an AI system that generates text. Under the AI Act (article 50, in force since 2 August 2026) we are its provider and you are its deployer. A post drafted with Retorik and published to inform the public on a matter of public interest needs no "made with AI" label when it has been through real human review and someone in your organisation holds editorial responsibility for it. The review workflow and the record exist so that you can rely on that; whether your review meets the bar, and whether to label a post anyway, is your decision. Where the Act requires us, as provider, to mark generated text in a way machines can detect, we follow the European Commission’s code of practice on marking, never by adding visible words to your post.
• Third-party material. When you ground a draft on an article or a page, Retorik quotes and attributes it. You are responsible for using other people’s material within the law, and for the accuracy of what you publish.

Do not use Retorik, or let your users use it, to:

1. publish or prepare anything unlawful, defamatory, threatening, or that infringes another person’s rights, including their privacy;
2. impersonate a person or an organisation, or publish from an account that is not the executive’s own;
3. break a platform’s rules, including on automation, spam and manipulation;
4. attack, probe, overload or reverse-engineer the service, or use it to attack anything else;
5. resell Retorik or offer it to third parties as your own, unless we agreed that in writing.

If we learn of a breach we tell you in writing what we found and what we did, and we may suspend the account concerned until it is fixed. You can object by replying; a person looks again, never a machine. For anything urgent, such as a compromised account or a legal order, we may act first and tell you at once. Anyone can tell us about content they believe is unlawful at info@retorik.pro.

Each of us keeps the other’s confidential information confidential: your content and your business, our software and our commercial terms. We use it only for this agreement and share it only with people who need it and are bound to the same duty. This does not cover information that is public, already known, or that the law requires us to disclose; where the law allows, we tell you before disclosing. The duty outlasts the agreement.

Availability. We aim to keep Retorik up and we tell you about incidents. During pilots and month-to-month plans there is no uptime guarantee and no compensation for downtime. A written order may say otherwise; then the order wins.

Support. Email info@retorik.pro. We answer on working days, usually within one working day, faster for anything that blocks publishing or looks like a security problem.

Scheduled posts. A scheduled post is published by Retorik at the set time through the platform’s interface. If the platform or the connection fails, the post is not published and Retorik shows the failure; we do not retry on our own, so check the draft.

Security. Passwords are stored only as salted hashes; passkeys are supported; sessions can be revoked on every device; an administrator can remove a person’s access at once, which ends their sessions everywhere and stops the invitations, keys and connected accounts they held; sign-in and generation are rate-limited; platform tokens are encrypted with keys held outside the database; every organisation’s data is separated from every other’s on every request; uploads are checked by size and content before they are read; every URL you ask Retorik to fetch is checked before it is fetched; and every workflow action is written to an audit log you can see. The detail is annex B of our data processing agreement. We do not promise that security is absolute; we promise to tell you promptly if a breach affects your data.

Backups. The database and the executives’ voice data are backed up. Restoring a backup restores the whole service to that point; we cannot restore one draft on its own.

Our privacy notice says what personal data we hold about your users and why. For the content you put into Retorik you are the controller and we are your processor; our data processing agreement, which is part of these terms, governs that. You are responsible for telling your executives and users what Retorik is and how their data is used, and for the personal data of other people that appears in the executive’s writing.

• A pilot runs for the fee in the written pilot agreement.
• Paid use runs on a written order: a price per executive per month, or as the order says. A solo executive’s order can be an email, or, once self-serve sign-up exists, the price shown at sign-up. Prices exclude VAT; VAT is added where the law requires. For a consumer the price shown includes VAT.
• We invoice monthly in advance unless the order says otherwise. Invoices are due within 30 days. If an invoice is not paid within 14 days after a reminder, we may suspend the service until it is.
• We tell you at least 30 days before a price changes. If you do not want the new price, end the agreement before it applies.

• Your content, your posts. You are responsible for what you and your users put into Retorik and for what you publish with it.
• What you cover. If a third party brings a claim against us because of your content, a post you published, a breach of a platform’s rules by your account, or personal data you had no right to use, you defend us and cover the damages, costs and reasonable legal fees that follow. We tell you promptly, let you run the defence, and do not settle without your agreement.
• What we cover. If a third party claims that the Retorik software itself, as we supply it, infringes their intellectual property rights, we defend you and cover the damages and reasonable legal fees that follow, or we change or replace the infringing part, or, if neither is possible, end the agreement and refund what you paid for the period you could not use it. This does not cover what the AI wrote, your content, or a combination you made with something that is not ours. Our total under this indemnity counts towards the cap below.
• Nobody covers the AI’s words. A draft is reviewed by a person and published by your decision; neither of us indemnifies the other for what a published post says.
• Our liability. We are liable for damage we cause on purpose or through gross negligence, and for anything the law does not allow us to exclude. Beyond that, our total liability in any twelve months is limited to the fees you paid us in the twelve months before the event, or, if you paid nothing, to EUR 2,500. We are not liable for lost profit, lost revenue, reputational harm, the actions of a platform, or loss of data beyond what section 9 promises.
• Claims in time. Tell us about a claim within twelve months of the event it concerns.

• Term. A pilot runs for its agreed period. Paid use runs month to month, or for the term in the order, and renews for the same term unless one of us ends it.
• You end it with one month’s notice by email; a fixed term in an order runs to its end. No reason is needed and there is no fee for leaving.
• We end it if you break these terms and, where the breach can be fixed, do not fix it within the time we give you; or for any other reason with 60 days’ notice. A consumer can end it at any time, with no notice. If we close Retorik altogether, we give at least 60 days’ notice and keep export working until the end.
• Export. At any time, and for 30 days after the agreement ends, you can take everything: the executive’s writing, the profile, the drafts with their versions, notes and records, and the published posts, in ordinary machine-readable formats. Ask at info@retorik.pro if you need help moving to another tool; we help for up to 30 days at no charge.
• Erasure. After those 30 days we erase your data within a further 30 days, backups on their own schedule within 30 days after that, unless the law requires us to keep something, in which case we tell you what and for how long. Two things are kept for the seven years Dutch law requires of our administration: invoices, and the record of who accepted these terms and which version they accepted. The second is what shows the agreement existed; it holds the organisation, the person’s name and work email, the documents and the date, and nothing about what you did with the service.
• Retorik may change hands. We may transfer this agreement, and your data with it, to a company that continues Retorik under the same terms, for example when the business incorporates. We tell you at least 30 days before. You may not transfer this agreement without our written consent.

We change these terms when the service changes or the law does. We email your administrators at least 30 days before a change that affects you and say what changed. If you keep using Retorik after that date, the new terms apply; if you do not want them, end the agreement before then and the old terms apply until it ends. Changes required by law, or that only add to what you may do, can apply sooner.

Dutch law applies. Disputes go to the court in The Hague, unless you are a consumer, in which case you may also use the courts of your own country. Talk to us first at info@retorik.pro; most problems are misunderstandings and we would rather fix them.

This section applies only if you use Retorik as a private person, not for a business or a profession. It adds to the rest and wins where the two differ.

• You can withdraw within 14 days of subscribing and get your money back. If you asked us to start at once and used the service, we refund the part you did not use.
• The liability cap and the indemnities in section 12 apply to you only as far as the law of your country allows; the protection that law gives you stands.
• Nothing in these terms takes away a right the law gives consumers in your country of residence.