How to use AI in marketing: 11 examples by job

Using AI in marketing rarely means "the AI runs our campaign". In practice, a clearly scoped task goes out with a brief, a file comes back, and someone on the team checks it before it's used. This guide walks through 11 of those tasks, grouped by job: what you put in, what you get back and where a person has to look before anything ships.

The examples come from Sokosumi, a marketplace where marketing teams brief AI coworkers. We sell these tasks, so read accordingly. What each example says comes back is what the linked workflow page lists; a listing describes the workflow, it doesn't promise that every run is complete or correctly sourced.

AI adoption in businesses

In 2025, 26% of German companies with ten or more employees used AI, and 57% of those with 250 or more, according to the Federal Statistical Office (Destatis). Across the EU the share was 19.95%, and 34.7% of the companies that used AI used it for marketing or sales (Eurostat). These figures cover businesses in general, not marketing teams, and they don't say which tasks work. The examples below do.

What is an AI agent?

An AI agent is software that performs a defined task from a brief and returns a result, instead of waiting in a chat for your next question. An AI coworker covers a role, such as research or creative, and usually combines several agents. What's available: AI agents for marketing. How the coworker model works: AI employees.

11 examples, grouped by job

Research and market analysis

Example 1: the recurring market briefing. Leadership asks the same market questions every week, and someone starts the search from scratch every time. You set up a topic field once: markets, competitors, regulatory topics and the questions each briefing must answer. The workflow lists a short document that opens with what changed since the last run, a source for each claim and figures from the Statista database. Check the numbers at their source and the "why this matters to us" line before the briefing goes to a distribution list. The news research covers English-language coverage, so German trade press may be missing. Use case: market intelligence briefings.

Example 2: know a company before you call it. Agencies do this before a pitch, sales teams before a first meeting. The input can be as little as a company name; the brief and the URLs you want assessed make it sharper. The workflow lists a document on market, competition and positioning, plus the prospect's current Meta ads and named weak spots on its website. Verify each fact you plan to use in the conversation. A company that publishes little produces a thin document, and the file is internal preparation, not something to hand the prospect. Use case: new-business research; agent: Company Researcher.

Competitors

Example 3: weekly competitor monitoring. You list the competitors and the moves you care about (pricing, launches, paid social, job postings, messaging) and pick a cadence. The workflow lists what each run reports: changes since the last run, messaging changes quoted word for word, pricing changes linked to the page they came from, current creatives from the Meta Ads Library and traffic estimates across several domains in one table. Traffic figures are estimates, not measurements. Only public sources are read, so anything behind a login or from private sales calls is missing. The report ends with a short "what does this mean" section; your team decides what it means. Use case: competitor monitoring.

Social listening

Example 4: what customers say when you're not in the room. You name brands, topics and the communities to read. According to the workflow page, Reddit Research returns recurring themes, sentiment shifts with the posts that caused them, verbatim quotes and the threads that deserve a reply, written as a document rather than a dashboard. Sentiment is a judgment call, so open the linked threads before acting on a shift. Quotes are research material; clear the rights before a user's words appear in an ad. Private groups and deleted posts aren't covered. Use case: social listening on a schedule.

Example 5: vet a creator before the contract. According to the workflow page, TikTok Profile Analysis returns content strategy, engagement figures and brand-safety signals for a public profile; Instagram Page Analysis does the same for Instagram pages. The analysis only sees public data. Watch a handful of recent videos yourself, because whether a creator fits your brand is a decision, not a metric. Same use case: social listening.

Audiences

Example 6: audience profiles with sources. You give the category, the market and the decision you want to influence. The workflow lists segment profiles with media and behavior data, plus jobs, triggers and objections per segment, built from multi-source research and GWI's global survey data, with sources visible. The profiles describe a market, not your customers, so hold them against your CRM and what sales hears before you build a campaign on them. Use case: audience research sprint.

Example 7: test messages before the budget moves. Ask the Crowd puts the exact statements you want to test in front of five AI personas filtered by demographics; the workflow lists their qualitative reactions, labeled as a synthetic test. This isn't research with real people. Use it to narrow a long list of candidate messages to the few you then test with real customers. Same use case: audience research sprint.

Content and campaigns

Example 8: the launch content package. The brief says which product you're launching, who it's for and what changes for them; existing pages or a tone guide make the drafts sound like you. The workflow lists positioning (claim, proof, objection handling), landing page copy, fifteen headline candidates, channel-specific social variants and ad visuals, all built from the same brief. Read and fix the positioning first, since everything downstream comes from it. Review and edit the copy for tone and brand style. Product promises, prices, competitor comparisons and effect claims go through the same legal review as any other ad copy. Use case: launch content package.

Example 9: score a landing page before it goes live. Page Copy Assessment takes a URL, rates the copy against ten principles attributed to David Ogilvy and names the sections that score poorly. The loop is simple: score, fix what's named, score again. Treat the score as a reference point for the discussion, not a verdict; only a test with real traffic tells you whether the page converts. Use case: launch content package.

Reporting

Example 10: a channel report from public data. YouTube Channel Analysis reads the 30 most recent videos of a public channel; its listing names views, engagement rate, upload cadence, title patterns, Shorts versus long-form and the outliers above and below the channel median. Instagram Page Analysis covers content themes, benchmarked engagement, sentiment and brand voice. You give a channel or profile URL, optionally with a question. No login to the channel being analyzed is needed; running the task needs a Sokosumi account. Public data contains no internal reach or conversion data for your channel. For your own channel it's an outside view; for a competitor's it's the only one you get.

SEO and AI visibility (GEO)

Example 11: measure AI visibility next to your rankings. Visibility now has two parts: search rankings and mentions in AI answers. Optimizing for the second is called GEO, generative engine optimization. You provide your domain and the topics you care about. The workflow lists a topic map of your pages, rankings and AI visibility per topic, the competitors and sources AI models cite, and a ranking of which gaps you can work on with reasonable effort. AI answers shift, so rerun the same topics after you publish and compare runs instead of trusting a single measurement. The audit doesn't write the content; that's a separate task. Use case: SEO and AI visibility; agent: SEO & GEO Researcher; background: answer engine optimization.

What the 11 have in common

Each task has a clear brief, a repeating shape and a result the person who briefed it can check. That is why these tasks move first. What stays with the team is everything where writing the brief is the hard part: positioning, creative judgment, the client relationship and the sign-off. The checks above boil down to a short list:

Numbers and facts get checked at their source before they're quoted.
Copy gets reviewed and edited for tone and brand style.
Legal claims (prices, comparisons, effects) go through the same review as any other copy.
Synthetic panels and estimates stay labeled as what they are.

Rules that apply in the EU

The AI Act's transparency duties in Article 50 have applied since 2 August 2026, the Regulation's general date of application under Article 113. The Digital Omnibus on AI (Regulation (EU) 2026/1744) did not move that date. What it means for a marketing team:

Deepfakes must be disclosed. If you use AI to generate or manipulate image, audio or video content that counts as a deep fake, you must disclose that it's artificial (Art. 50(4)). For evidently artistic, satirical or fictional work, a disclosure that doesn't spoil the work is enough.
Some text needs a label, unless a person reviewed it. AI-generated text "published with the purpose of informing the public on matters of public interest" must be disclosed. The duty falls away where the text "has undergone a process of human review or editorial control" and someone holds editorial responsibility. Record who approved what. Whether a given piece falls under this clause is a question for your legal team.
The label must come early. Disclosure has to be clear and distinguishable "at the latest at the time of the first interaction or exposure" (Art. 50(5)).
Chatbots must identify themselves. Providers must design systems that talk to people so the people know they're dealing with AI, unless that's obvious (Art. 50(1)). If you run a chatbot on your site, ask the vendor how it meets this.
Machine-readable marking is the provider's job. Providers of generative systems must mark outputs as artificially generated (Art. 50(2)). Systems already on the market before 2 August 2026 have until 2 December 2026, under the Omnibus.
AI literacy is an obligation. Article 4 has applied since 2 February 2025. As amended by the Omnibus, providers and deployers must take measures to support the AI literacy of staff and others who operate or use AI systems on their behalf, taking into account their knowledge, experience and the context of use. The amended text no longer requires a guaranteed level for each individual; the measures themselves remain mandatory.

GDPR can apply to personal data supplied in a brief or collected during research, including public comments. Public availability does not remove the need for a lawful basis (Art. 6) and data minimization (Art. 5(1)(c)). A vendor processing personal data on your behalf needs sufficient guarantees and a contract (Art. 28). Leave customer lists and CRM exports out of a brief unless the task needs them, and ask what a research task collects about people. On Sokosumi, each coworker's profile shows the models and hosting where the vendor states them; check that before the first run that includes data. More on this: European AI. This section describes the legal texts; it isn't legal advice.

Where to start

Pick one task from the list that your team repeats every week or month. Run it twice, compare the result with what your team would have produced and keep the check step. An account is free, the free plan includes 250 credits a month, and every task shows its credit price before it runs (pricing). All workflows are on the use cases page.

Sources

Questions about AI in marketing

How can AI be used in marketing?

Mostly for clearly scoped, recurring tasks: market and competitor research, social listening, audience profiles, message tests, first drafts of campaign content, channel reports and SEO or AI-visibility audits. The team briefs the task, the AI returns a file, and a person checks sources, numbers, brand voice and legal claims before anything is used.

What are AI agents in marketing?

Software that performs a defined marketing task from a brief and returns a result, such as a competitor scan or an Instagram analysis. Unlike a chat assistant, you don't steer each step. An AI coworker goes one level up: it covers a role and usually combines several agents. The AI agents for marketing overview shows examples.

What are examples of AI in marketing?

A weekly competitor report with sources, a Reddit listening summary with verbatim quotes, audience profiles built on survey data, a launch package with fifteen headline options, a YouTube channel report and an audit of which brands AI answers cite for your topics. All 11 examples above include the input, the output and the check step.

Will AI replace the marketing team?

The examples show tasks AI can take on. They don't tell you how jobs and teams will change as a result. What can be checked today, and what is opinion, is in Will AI replace marketing jobs? What's changing in 2026.

Do I have to label AI-generated marketing content in the EU?

Deepfake images, audio and video must be disclosed under Article 50(4) of the AI Act, which has applied since 2 August 2026. AI-generated text published to inform the public on matters of public interest must be disclosed too, unless it went through human review and someone holds editorial responsibility. Ask your legal team which of your formats fall under this.

Can I put customer data into an AI tool?

Only with a legal basis under GDPR Article 6, only as much as the task needs, and only with a vendor that offers sufficient guarantees and a processing contract (Article 28). Public sources can contain personal data too, such as names in comments, and being public doesn't remove the need for a lawful basis. Leave personal data out of the brief unless the task needs it.