Social Rocket
The Canadian marketing agency BaD Mktg works with just two full-time partners and uses AI for idea development and internal processes. Freelance staff are added depending on the project. The case that Digiday presents on 13 October 2026 shows one possible agency model for smaller teams – but no proof that artificial intelligence can replace a complete twelve-person agency.
The most productive use of AI is not necessarily the finished advertising
Ian Buck and Yotam Dor use two types of AI support. In creative work, variants, storyboards and preliminary campaign drafts are developed. In addition, an internal tool called Goodie coordinates information from workflows, priorities and administrative processes, among other things. In the Digiday interview, Buck stresses that experienced people select from many suggestions and evaluate the result.
This reflects a basic problem of generative campaign work: 100 ideas are not yet 100 usable ideas. The benefit comes when drafts become visible earlier and an experienced team consistently discards the weak options.
A further report on the company, which BaD Mktg documents on its website, describes the use of animated AI storyboards and the continued use of a pool of freelance specialists. This makes clear that "two people" refers to the number of full-time staff, not the total number of people involved in a client project.
What the revenue story says – and what it doesn't
According to the founders in the Digiday conversation, the agency, 2023 launched, generated more than two million US dollars in revenue over three years. The concept phase is also said to have shrunk from around six weeks to at most six days. Independently audited annual accounts, project costs or reliable before-and-after control groups have not been published.
No industry-wide productivity gain or actual margins can be derived from this. In particular, information is missing on freelancer fees, software, revision rounds, client acquisition and founders' working hours for the economics.
The consequences for German agency contracts
With smaller AI-supported teams, three questions become especially important.
1. Who does the work? The contract model should disclose which services the core agency handles and where external specialists or AI service providers are involved. Size alone is not a quality benchmark.
2. What is billed? Reduced time spent does not necessarily mean lower project prices. A fixed price can be plannable for clients if quality, the number of correction loops and usage rights are clearly defined. Our analysis of AI costs, token audits and agency remuneration fits here.
3. Who protects client data and content? Internal prompts, briefings and creative drafts can contain sensitive business details. What matters are the providers used, access rights, confidentiality and the agreed use of outputs. Which AI clauses in agency contracts make sense therefore depends more on the workflow than on the advertising promise.
BaD Mktg is not alone in understanding AI as an operating system and not merely as an image generator: Liquid Death and Power Digital are experimenting with AI-supported ongoing optimization. In agency pitches, data access and transparent performance criteria remain central.
Conclusion: The relevant signal is not "two people replacing twelve", but a different cost structure: experienced leads, variable specialists and automated routine work. Whether a client receives better advertising will have to be shown by the quality and impact of their specific campaign – not by the number of AI tools used.