Privacy policy

Fortell is operated by Barko Studio, based in the Republic of Ghana. For the purposes of the Data Protection Act, 2012 (Act 843), we are the data controller for the personal data described here. Barko Studio is the name the business trades under, so to be precise about who that controller is: Richard Acheampong and Deon Takyi, trading as Barko Studio. They can be reached at the address below.

You can reach us at support@fortell.site. That address is monitored for privacy requests as well as support, and it is the fastest route to a deletion or a copy of your data.

Everything below is either given to us by you or produced by your use of the product.

• Your email address. Fortell has no passwords. Signing in happens through Google or through a one-time link sent to your inbox, so your email address is your identity here. We store a normalised copy of it as well, which collapses +tag and Gmail dot variants so one inbox cannot quietly become several accounts.
• Your name and Google account identifier, if you sign in with Google. We do not receive or store your Google password.
• The files you upload and everything the pipeline derives from them. Section three covers this on its own.
• Models you train, their scorecards, and the registry entries recording who promoted a model to production and when.
• Research reports you generate, including the generated PDF.
• Usage records. Which AI-touching action you ran, when, and how many compute tokens it cost. This is what your plan allowance is measured against, so it is not optional.
• Payment records. The amount, currency, plan or credit pack, status and the reference our payment provider issued. We never see or store card numbers — card entry happens on Paystack's own checkout page, not ours.
• Runner device records, if you connect one: a device name, the agent version it reports, and a hashed form of its access token. The token itself is shown once, at creation, and is not recoverable from our database afterwards.
• Workspace membership, if you invite people or are invited: which organisation you belong to and your role in it.
• Server logs. Every request is stamped with a request identifier and logged with its path, status and timing. Logs are for debugging and abuse prevention, and are not used to build a profile of you.

When you upload a spreadsheet, the file itself is stored — compressed — so that your work survives a server restart. Each step of the pipeline (filling gaps, handling outliers, scaling, selecting features, splitting) produces a new derived version, and the current one is checkpointed so you can close the tab and come back to it.

In plain terms — Your file is not read by a person, is not used to train any model that other customers use, and is not sold or shared for advertising. It is held so the product can do the thing you asked it to do, and it is deleted when you delete it.

Two things about this are worth being explicit, because they are the parts people are usually surprised by:

• Your data is processed on our servers, not in your browser. Profiling, cleaning and training all happen server-side. There is no version of Fortell where the file stays on your machine.
• If you use an AI feature, some of your data goes to an AI provider. This is covered in full in section five. If that is unacceptable for a particular dataset, the entire machine-learning pipeline — profiling, cleaning, training, tuning, evaluation and prediction — works without any AI feature and sends nothing to any AI provider.

Please do not upload data you are not entitled to process. If your file contains other people's personal data, you are the controller of it and we are your processor; you are responsible for having a lawful basis to upload it. Fortell is not designed or certified for special-category data — health records, biometrics, and the like — and you should not upload it.

You can delete any dataset from the workbench at any time, which removes the stored file and its derived versions.

We do not sell personal data, we do not share it with advertisers, and we do not make automated decisions about you that produce legal or similarly significant effects.

This is the section most worth reading. Fortell's AI features are built on models we do not host, which means using them sends data to a third party.

• Data insights and explanations. When you ask for an AI reading of a dataset, we send a summary of it — column names, data types, summary statistics and a small sample of rows — to Google's Gemini API. Column names and sample rows can contain personal data if your file does.
• Research papers. The research feature sends your research question, and passages from the published papers it retrieved, to an AI provider. It does not send your uploaded dataset unless you explicitly run the data-driven paper mode, in which case it sends the summarised results of your analysis.
• Literature search. Your research question is sent as a search query to scholarly databases — OpenAlex, Crossref, Semantic Scholar, Europe PMC, CORE, DOAJ and others. These receive the query, not your data.

Fortell's AI requests are made through our own account with each provider, under their API terms rather than their consumer terms. At the time of writing, Google states that data submitted through the paid Gemini API is not used to train its models. We cannot give you a warranty about a third party's conduct, which is exactly why the sentence above about running the whole pipeline without any AI feature is there.

In plain terms — Train models, forecast, cluster, visualise and download predictions, and nothing leaves our servers. Press an AI button and a summary of your data goes to Google.

Fortell can hand a training job to a machine you control — your laptop, a workstation, or a Colab session — by installing the Fortell agent. When you do this, a snapshot of the dataset for that job is sent from our servers to your machine, over an authenticated connection, so it can be trained on there.

This moves data outward, not inward: the snapshot is deleted from our side when the job reports its result, and swept up if the job ends any other way. The trained model comes back as a portable graph file. Anything the runner writes on your own machine is on your machine and outside this policy — including the copy of the snapshot it downloaded, which you can remove yourself.

These are the companies that handle some part of your data on our behalf. Where a service is optional, it is marked, and if this deployment is not configured for it, it receives nothing at all.

Fortell uses very few, and none of them are for advertising or cross-site tracking.

The session cookie is strictly necessary: it exists only to run a service you asked for, so it does not require consent, and there is no version of a signed-in app without it. There are no third-party cookies, no advertising pixels, and no cross-site trackers on any Fortell page.

You can block or clear cookies in your browser. If you block the session cookie you will not be able to sign in.

We use Vercel Analytics and Vercel Speed Insights to count page views and measure how fast pages load. Both are cookieless: they set no cookie, store no identifier in your browser, and do not follow you to other websites. What we get back is aggregate — how many people opened the pricing page, how slow the landing page was on a mid-range phone — with no way to single you out.

We also count page views ourselves, on our own server, so that we can see them next to how far people get through the product. This one is cookieless too, and it keeps less than the sentence above might suggest: what is stored is a count — the date, the page, and how many views it had. No row records a person. To avoid counting the same reader ten times in one day, the server works out a scrambled, one-way value from your network address and browser, holds it in memory only, and throws it away at midnight; it is never written to disk, never leaves our server, and a new one is generated each day, so it cannot be used to recognise you tomorrow or to link today to yesterday. We do not record the part of the address after the page name, so search terms, campaign tags and payment references are never collected.

This is why the notice at the bottom of the page is a notice rather than an accept-or- reject gate: there is no tracking cookie to consent to. If we ever add one, it will be off until you turn it on, and this section will say so.

Deleting your account removes your datasets, models and reports. Payment records are the exception: we keep the minimum needed to satisfy tax and accounting obligations.

Fortell's servers, database and storage are hosted outside Ghana, and the AI and payment providers listed above operate internationally. Using Fortell therefore involves transferring your data across borders.

Where a provider offers standard contractual clauses or an equivalent transfer mechanism, we rely on it. Section 18 of Act 843 permits such transfers where the receiving jurisdiction provides an adequate level of protection or where you consent to the transfer; by uploading data to Fortell you are asking us to process it on those servers.

• Traffic to Fortell is served over HTTPS.
• There are no passwords to steal — sign-in is a Google account or a single-use link that expires and cannot be replayed.
• Runner device tokens are stored hashed, so a copy of our database does not yield a working token.
• Error reports are stripped of personal data and of the local variables in a stack frame before leaving our servers, because a stack frame inside a training route holds your dataset.
• Models returned by a runner are accepted only as a portable graph format, never as a format that can execute code on our servers.
• Access inside a shared workspace is scoped by membership, so a workspace member sees that workspace's data and not another's.

No system is perfectly secure, and we will not pretend otherwise. If you find a vulnerability, please write to support@fortell.site rather than disclosing it publicly, and we will work with you.

Under the Data Protection Act, 2012 (Act 843), you have the right to:

• be told whether we hold personal data about you, and to be given a copy of it;
• have inaccurate data corrected;
• have your data deleted, or its processing blocked;
• object to processing that is causing or likely to cause you damage or distress;
• prevent processing for the purposes of direct marketing.

Write to support@fortell.site and we will respond within 30 days. Most of these you can also exercise yourself: datasets, models and reports each have a delete control in the app.

If you are not satisfied with our response, you can complain to the Data Protection Commission of Ghana, the supervisory authority established under Act 843.

If you are in the UK or the European Economic Area, we will honour the equivalent rights under the UK GDPR and the GDPR — access, rectification, erasure, restriction, portability and objection — and you may complain to your national supervisory authority.

Fortell is not intended for anyone under 18 and we do not knowingly collect data from children. If you believe a child has created an account, tell us and we will remove it.

We will update this page when the product changes in a way that affects it — a new sub-processor, a new category of data. The date at the top always reflects the current version. If a change materially reduces your protections, we will email account holders rather than relying on you noticing the date.

Barko Studio, based in the Republic of Ghana.
Email: support@fortell.site

See also the Terms of Service, which govern your use of Fortell.