Frequently Asked Questions
Find answers to common questions about AtBice, our services, and how we handle your data.
AtBice is a Swedish technology company focused on digital sovereignty. We provide IT consulting, custom development, DevOps services, and a suite of free and premium self-hosted tools for individuals and businesses.
Digital sovereignty means having full control over your data and digital infrastructure. At AtBice, this means we offer self-hosted solutions with no third-party cloud dependencies, so your data stays under your control at all times.
No. All media conversion tools (image, video, audio converters) work directly in your browser without an account. Your files are never uploaded to our servers.
You can reach us through the Contact page, by email at [email protected], or through our community chat. Paid subscribers receive priority support with faster response times.
We accept card payments, Swish (Sweden), SEPA Direct Debit (EU), Klarna, and Stripe Link. All payments are processed securely through Stripe.
Yes. You can cancel your subscription at any time from your profile page. Your access continues until the end of your current billing period. No cancellation fees.
We offer a 14-day money-back guarantee for all new subscriptions. If you're not satisfied within the first 14 days, contact us for a full refund.
Yes. Our Enterprise plan includes dedicated support, custom integrations, SLA guarantees, and can be tailored to your organization's needs. Contact our sales team for details.
We follow GDPR and Swedish data protection laws. We collect only what's necessary, store data securely in our own infrastructure (no third-party clouds), and give you full control to export or delete your data at any time.
Yes. Under GDPR Article 20, you have the right to data portability. Go to your Profile page and use the 'Export Data' feature to download all your personal data as a JSON file.
You can delete your account from the Profile page. This permanently removes all your personal data from our systems. Note that some financial records may be retained for 7 years as required by Swedish accounting law (Bokforingslagen).
No. AtBice has a strict no-cookie policy. We do not use cookies, tracking pixels, or any other tracking technologies. Authentication tokens and preferences are stored in your browser's localStorage, which is never sent to third parties. We believe your browsing activity is yours alone.
Yes. All image, video, and audio conversions happen locally using WebAssembly (FFmpeg.wasm). Your files never leave your device, ensuring complete privacy.
We support a wide range of formats including PNG, JPEG, WebP, GIF, BMP for images; MP4, WebM, AVI, MOV for video; and MP3, WAV, OGG, FLAC for audio. New formats are added regularly.
Since processing happens in your browser, limits depend on your device's memory. Most devices can handle files up to 500MB-1GB comfortably. For larger files, we recommend using a desktop browser.
Premium subscribers get access to our self-hosted suite: collaborative docs, file sharing, Git hosting, and more. All tools are hosted on our own infrastructure for maximum privacy.
It means the model runs on machines you control — a workstation, a server in your office, or your own rack in a data centre — not on someone else's cloud GPUs. The model weights live on your storage, inference happens on your silicon, and the prompts and documents you feed it never cross your network boundary. You own the hardware, the weights, and the infrastructure. There is no per-token bill and no third party reading your inputs.
No. With a local deployment, every prompt, document, and generated response stays inside your perimeter. The model is loaded from your own disk and runs against your own CPU or GPU, so there is no outbound API call to a hosted service and nothing to log on a vendor's servers. For our browser-based AI tools, inference runs client-side in WebAssembly/WebGPU — the data never even leaves the tab. You can verify this by watching your network traffic: there is no model endpoint to phone home to.
On-premises means the AI runs on infrastructure you own and operate, but that infrastructure may still have a controlled internet connection — for updates, monitoring, or other services. Air-gapped goes further: the machines running the model have no network path to the public internet at all, so data physically cannot leave. We deploy both. Most organisations are well served by on-prem; regulated, classified, or high-sensitivity environments choose air-gapped, where we ship models, weights, and updates on physical media and the whole stack runs in isolation.
Open-weight models you can self-host — families such as Llama, Mistral/Mixtral, Qwen, Gemma, and Phi, plus specialised models for coding, embeddings, speech-to-text, and OCR. We match the model to your hardware: small quantised models run on a single consumer GPU or even a modern laptop, while 70B-class models want a server-grade GPU or a multi-GPU box. Because the weights are open and stored locally, you are never tied to a single vendor and you can pin a version so behaviour does not change underneath you.
For most real business tasks, yes. Open-weight models now handle summarisation, drafting, classification, extraction, retrieval-augmented Q&A over your own documents, and code assistance at a quality that is more than sufficient — and you can fine-tune them on your data, which closes most of the remaining gap for domain-specific work. The largest frontier hosted models still lead on the hardest open-ended reasoning, so we are honest about the trade-off: if a task genuinely needs that frontier capability, we will tell you. For the everyday workload that fills most teams' days, local models win on privacy, cost predictability, and control without a meaningful quality penalty.
When inference happens on hardware you control, your data residency story is simple: the data stays exactly where you put it. There is no transfer to a US cloud, no sub-processor to add to your records of processing, and no exposure under foreign surveillance regimes — which directly addresses GDPR, Schrems II, and sector rules like Sweden's accounting and healthcare requirements. You decide the physical location, the retention, and the access controls. For audits, this is far easier to defend than a hosted AI API, because there is no third party in the data path to account for.