It depends on whose name the AI system is placed on the market or put into service under. An agency that offers an AI workflow or chatbot under its own name or brand is the provider, and its client is the deployer. If the agency builds a system to order and the client puts it into service under the client's own name, the client is in principle the provider and the agency acts as a contractor. For high-risk systems, Article 25 can also turn a deployer or distributor into the provider.
What do the Article 3 definitions say?
Article 3 of Regulation (EU) 2024/1689 sets out the key roles:
- Provider (point 3): whoever develops an AI system or a general-purpose AI model, or has one developed, and places it on the market or puts it into service under its own name or trademark, whether for payment or free of charge.
- Deployer (point 4): whoever uses an AI system under its authority, except for personal non-professional activity.
- Distributor (point 7): a supply-chain actor, other than the provider or importer, that makes an AI system available on the EU market.
- Putting into service (point 11): supplying an AI system for first use directly to the deployer, or for own use, in the EU.
Two consequences for agencies. First, the deciding factor is the name or brand the system is offered under, not how much code you wrote. Second, n8n and Make are only the orchestration layer: the AI system being assessed is what you assemble (trigger, prompt, model call, tools, output), as explained in is an AI agent an AI system?.
When does a deployer or distributor become the provider?
Article 25(1) lists three situations in which a distributor, importer, deployer or other third party is treated as the provider of a high-risk AI system:
- it puts its name or trademark on a high-risk system already on the market or in service, without prejudice to contracts that allocate the obligations differently (point (a));
- it makes a substantial modification to a high-risk system that remains high-risk (point (b)). Article 3(23) defines a substantial modification as a change not foreseen in the initial conformity assessment that affects compliance with the requirements or changes the intended purpose;
- it changes the intended purpose of an AI system, including a general-purpose AI system, that was not high-risk, so that it becomes high-risk (point (c)).
Point (c) is the one agencies hit most often. Plugging a general-purpose model into a workflow that screens job applications gives it an Annex III, point 4 purpose: you become the provider of a high-risk system, even though you did not build the model.
The initial provider then stops being the provider of that system but must cooperate with the new one (documentation, technical access), unless it clearly ruled out conversion into a high-risk system (Article 25(2)). Article 25(4) also requires a written agreement between the provider of a high-risk system and third parties supplying tools, services or components. Since the Omnibus, breaching paragraphs 2 and 4 is expressly fineable (Article 99(4)(da)).
Chapter III, including Article 25, applies to Annex III systems from 2 December 2027 (see the EU AI Act timeline). For systems that are not high-risk, roles follow the Article 3 definitions, notably for the Article 50 disclosure duty that has applied since 2 August 2026.
What are the typical agency scenarios?
| Scenario | Agency | Client |
|---|---|---|
| Custom workflow delivered to a client who runs it under its own name | Contractor; set out in the contract | Provider and deployer, in principle |
| Workflow or chatbot sold under the agency's brand to several clients (SaaS, subscription) | Provider | Deployer |
| White-label chatbot shown under the client's name | Developer; provider role follows the displayed name | Provider in principle; Article 25(1)(a) if high-risk |
| Internal agency tool built on a third-party model | Provider (put into service for own use) and deployer | Not involved |
| Agency using an off-the-shelf AI assistant as is | Deployer | Not involved |
| Workflow that repurposes a general-purpose model for an Annex III use | Provider of a high-risk system (Article 25(1)(c)) if put into service under its name | Deployer, or provider if put into service under its name |
The first scenario is the most common and the least settled. The text covers whoever "has an AI system developed" and puts it into service under its own name, so a client who commissions a workflow and runs it under its brand can be the provider without writing a line of code. The Commission has not published a specific position on bespoke development, so put the allocation in writing.
What should agencies do in practice?
- For each project, record the AI system, its purpose, the name it is put into service under and each party's role.
- Write that allocation into the quote or contract, including who builds and maintains the Article 50(1) disclosure and who marks generated content.
- Always check whether the purpose falls under Annex III; if it does, plan for provider-of-high-risk status by December 2027.
- For white-label work, make sure the "you are talking to an AI" notice survives any rebranding.
- Keep what you hand over to the client (instructions, known limitations): it is how you show the allocation of roles, as covered in how to prove compliance.
VerifactAI plugs into n8n and Make workflows: for each system declared with the agency's and the client's roles, it identifies the obligations, can block or warn before the model call, and records the controls applied and the evidence.
Sources
- Regulation (EU) 2024/1689, EUR-Lex
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), EUR-Lex
- Article 3 (definitions), AI Act Service Desk
- Article 25, AI Act Service Desk
- Annex III, AI Act Service Desk
This article is general information, not legal advice.