GENERAITR is an AI Media Production platform for teams, built on control, collaboration and compliance. An AI assistant, grounded in a 400+ parameter production taxonomy, turns your intent into structured, reproducible GENERAITR Templates that anyone on your team can run. Every generation is fully documented for compliance. Available as a managed web app (Online, running on GENERAITR Engine – our own production engine) and as a fully on-premise installation (Local, with ComfyUI available if you want it).

Online is hosted and managed by us on EU servers (Hetzner), running on GENERAITR Engine – our own production engine, purpose-built MCP-native for agentic AI. You sign up, open a browser, and start generating. Generation credits for API models are included with your plan. Local runs on your own hardware within your own network. You install GENERAITR yourself – Docker and the GENERAITR configuration; ComfyUI is optional if you want to bring your own workflows. Open source models run entirely on your own GPU at zero cost per generation. Both versions include the same compliance features: metadata, watermarking, trace graph and audit trail.

ComfyUI is an open source node-based interface for building AI generation workflows. GENERAITR Online doesn't use it at all – it runs on GENERAITR Engine, our own production engine. GENERAITR Local supports ComfyUI as an option: if you already have workflows built in ComfyUI, or want to build your own, you can bring them in and publish them as GENERAITR templates that the rest of your team runs through a clean interface with parametric controls, no ComfyUI knowledge required. ComfyUI knowledge is an advantage for Local power users – never a requirement to get started.

For GENERAITR Online: no. Generation credits are included with your plan (700 GENERAITR Credits (GC)/month on Core, 2 000 GC on Pro, 5 000 GC on Scale) and cover API model generation. No GPU, no separate account needed.

For GENERAITR Local: yes, if you want to run open source models on your own hardware. For API models on Local, you connect your own Comfy account – no GPU required for those.

GENERAITR comes pre-configured with the world's leading AI models for image and video generation – available from day one, no manual installation needed. For Online, models are accessed via ComfyUI Partner Nodes (API). For Local, both API models and open source models running on your own hardware are supported. Audio and 3D generation are on the roadmap and will be added in a later release. We document each model we offer: training data status, license type, and EU AI Act Article 50 compliance status.

GENERAITR Credits (GC) are the generation currency for API model generation in GENERAITR Online. Credits are consumed per generation depending on the model and output type. Every Online plan includes a monthly GC allocation: see the next question for how far each plan's credits stretch.

Top-up packs are available in the platform (500–10 000 GC) if you need more.

For Local customers using open source models: no credits needed: generation runs free on your own hardware. Local users generating with API models use their own Comfy Credits directly.

Models are grouped by cost per generation:

Fast image models (< 5 GC/run): e.g. Flux.2 Pro (1K), Seedream (1K), WAN 2.5 T2I (1K) · Fast upscaling: Stability Upscale (→ 2K), SeedVR2 (→ 4K)
Core: 140–175 images · Pro: 400–500 · Scale: 1 000–1 250

Standard image models (5–15 GC/run): e.g. Reve Image Create (1K), GPT Image 2 (1K), Recraft V4 (1K), Nano Banana 2 (1K) · Standard upscaling: Topaz Reimagine (→ HD)
Core: 50–140 images · Pro: 135–400 · Scale: 335–1 000

Short video & premium image (15–50 GC/run): e.g. Nano Banana 2 (2K), Nano Banana Pro (1K / 2K), Seedance (5s 1K), LTX-2 Fast (5s), Vidu Q3 Turbo (5s 1K), Runway Gen4 Turbo (5s) · Premium upscaling: Topaz Reimagine (→ 1080p / 4K)
Core: 14–47 generations · Pro: 40–134 · Scale: 100–335

Standard video (50–150 GC/run): e.g. Kling3 (4s 1K), WAN 2.7 T2V (5s 1K), Vidu Q3 Turbo (5s HD) · Heavy upscaling: Stability AI Conservative (→ 4K)
Core: 5–14 generations · Pro: 13–40 · Scale: 33–100

Exact costs per model are listed in the platform's model library.

Yes, from day one, across all eight requirements of the regulation.

EU AI Act Article 50 – mandatory from 2 August 2026 (read the article)

Every generated file carries three automatic layers of AI labelling: an invisible steganographic watermark embedded in the pixels, a C2PA Content Credentials manifest signed into the file itself, and a full lineage record (model, template, parameters, source media, operator, timestamp). C2PA is the open standard developed by Adobe, Microsoft, Google, the BBC and others. Any client, regulator or platform can inspect the file and read its AI origin without needing access to your GENERAITR workspace.

For visible labelling at export: the Export panel includes configurable text and image overlays (your organisation's watermark, a custom label, or the GENERAITR mark) applied on top of the image before download.

For audit and disclosure: the Export panel includes a Metadata button that downloads a machine-readable .meta.json disclosure record for each asset, covering model, pipeline, parameters, assembled prompt and timestamp. Portable and attachable to any deliverable.

Note on Early Access: manifests are signed with a self-issued certificate. External tools (contentcredentials.org/verify) show "Content Credentials found: not verified by trusted authority." The manifests are valid and compliant. CA-signed certificates, giving a fully verified green check, are introduced with the Beta release in August 2026.

EU AI Act Article 4 – AI competence (in force since February 2025)
Every template card displays the AI modality and a usage notice at the point of selection, so users understand they are operating a generative AI system before they run it. The onboarding flow and compliance page provide further context on EU AI Act obligations.

Human oversight
Every generated output has a human review toggle. A named operator must mark each image as reviewed before it can be bulk-exported. The reviewer identity and timestamp are stored, giving an auditable trail.

Data residency
For Online: all processing and storage on EU servers (Hetzner, Germany). For Local: nothing leaves your own infrastructure.

Model transparency
We document training data status and licence type for every model we offer. Where information is incomplete or publicly unavailable, we say so.

What happens if organisations don't comply?

Fines for breaching Article 50, the transparency and labelling requirements, reach up to €15 million or 3% of global annual turnover, whichever is higher. Enforcement is decentralised: each EU member state designates its own supervisory authority to handle complaints, conduct market surveillance, and levy penalties. In Sweden, PTS (Post and Telecom Authority) is the proposed authority, with IMY retaining jurisdiction where AI use intersects with personal data.

Monitoring is not an automated system scanning every AI-generated image. In practice, enforcement is triggered by complaints from competitors, consumer groups, journalists, or regulators conducting proactive reviews. The risk is real even if it isn't guaranteed to be immediate. Once a complaint is filed, organisations either have an audit trail or they don't. GENERAITR ensures you always have one.

Aligned with the EU's official Code of Practice

GENERAITR adopts the European Commission's official EU icons (AI generated, AI modified, and a basic interactive AI mark) as the default label on every export, in line with the Commission's Code of Practice on marking and labelling AI-generated content. Tomlin Studio AB adheres to the Code as both a provider and a deployer.

EU organisations using other tools on 2 August will need to build or bolt on compliance after the fact. GENERAITR customers are already covered.

Responsible means your data and your choices stay yours. For Online: media and metadata are hosted on EU servers. For Local: everything stays on your own hardware – nothing leaves your network. On model ethics: we document training data status and license type for every model we offer. Where that information is incomplete or publicly unavailable, we say so. We do not control how model providers train their models – but we believe you have the right to know what we know before you choose.

When you generate with API-based models via ComfyUI Partner Nodes, your input data passes through the model provider's servers – typically outside the EU. The data should not be stored there, but transit occurs. We communicate this openly and document it per model. Full data sovereignty is only achievable with GENERAITR Local and open source models. If this is a hard requirement, Local is the right choice.

The Founding Partner rate is the Early Access price: Core €79/month, Pro €149/month, Scale €249/month. For Online, the EA window is open from 25 May to 31 July 2026. For Local, the invite-only EA window runs from 25 May to 18 June 2026. Everyone who joins during EA has their rate locked at signup – it does not change when Beta pricing starts on 3 August.

EA customers keep their Founding Partner rate as long as they stay subscribed. Between 27 July and 30 August, all EA customers (Online and Local) are offered an Annual Founding Partner plan at a further reduced effective rate. New customers from 3 August pay Beta pricing: Core €99/month, Pro €199/month, Scale €349/month.

Local is a platform license sold via Stripe as a monthly or annual subscription. Once payment is confirmed, GENERAITR issues a license key. The Local installation validates this key against the GENERAITR license API every 24 hours. Internet access is required for this periodic check – not for generation itself, which runs entirely on your own hardware. A grace period of 3–5 days applies for payment issues, so your installation is not interrupted without warning.

A server capable of running Docker and ComfyUI, on Windows, Mac or Linux. For open source model generation, a GPU is required. For API model generation via Comfy Partner Nodes, no GPU is needed. Storage and hosting are your own responsibility. Additional compute nodes beyond your tier are available as an add-on.

Your team works from an AI assistant and Greyboxing guardrails, not ad-hoc prompts. Instead of writing free-text prompts, the assistant reasons through your intent and sets organised input fields for you: subject type, environment, lighting, style, format and more, depending on the template. Greyboxing references act as parametric guardrails for geometry and camera, so composition is set, not guessed. Every output is tied to a specific, reproducible set of parameters.

You can save your preferred parameter selections as a reusable GENERAITR Template. Once saved, your team selects it and runs it without reconfiguring from scratch. A power user (admins) or the GENERAITR/TOMLIN STUDIO team builds the template; everyone else runs it through a clean interface. The output is reproducible, consistent and traceable every time.

Every generation lands in a shared media library where you can build comparison boards to view variants and versions side by side. Leave threaded comments or drawing annotation directly on a specific generation, not in a separate chat thread disconnected from the output. Invite teammates or external clients into a project with the right permission level, no sharing the whole account or exporting files for review outside the system. Approval decisions stay linked to the exact version and generation metadata, searchable afterwards.

For Local customers: yes. You connect your own ComfyUI account and can build, test and import workflows directly into GENERAITR as templates. For Online customers: Online runs on GENERAITR Engine, not ComfyUI, so ComfyUI workflow imports aren't part of the Online product – you work from GENERAITR's pre-built templates (GENERAITRs) instead.

Most tools – from individual generators like Midjourney and Adobe Firefly to team-oriented platforms like Krea, Magnific and Flora – share the same paradigm: give everyone credits and let them generate individually. The output depends on who is prompting and when.

GENERAITR is built differently, on control, collaboration and compliance: an AI assistant and Greyboxing guardrails standardise how your whole team generates, so the result is reproducible regardless of who runs it; review and approval happen inside the system, not over screenshots; and EU AI Act Article 50 compliance with C2PA provenance metadata is embedded in every exported file. That combination, plus the choice between managed cloud (Online) and full on-premise (Local), does not exist elsewhere in the category today.

ComfyUI is an open source, node-based generation interface. GENERAITR Online doesn't run on it – Online is powered by GENERAITR Engine, our own production engine, built MCP-native for agentic AI from the start. GENERAITR Local is different: it's designed to work alongside ComfyUI if you want it. You connect your own ComfyUI account, build and test workflows there, and deploy them as GENERAITR templates.

Either way, GENERAITR adds the control, collaboration and compliance layer the raw engine doesn't have: an AI assistant and GENERAITR Templates your whole team can run without touching a node, a shared media library with comparison boards and threaded review, EU AI Act compliance with C2PA metadata embedded in every export, and team access management. Power users on Local can still build directly in ComfyUI – everyone else gets a clean interface and reproducible results.

Our team generates on your behalf – live via Zoom in a Start or Pro Session, or as a standalone package delivery. All generation happens inside GENERAITR using our studio account. You guide the direction, you receive the files, and every output carries the same EU AI Act metadata and watermarking as any GENERAITR generation.

Yes, for GENERAITR Local. You get one week free, connected to your own ComfyUI account and credits. Open source models run on your own hardware at zero cost per generation; API models run via your own Comfy Credits.

For GENERAITR Online, there is no open self-serve trial. Registering for Early Access is free and commitment-free – you pay when you sign up and pick your tier plan. The Founding Partner rate, from €79/month, is the lowest price GENERAITR will be offered at. Sign up for Early Access.