AI Agents Now Recruit Influencers, Negotiate Terms and Write Reports

Six to eight hours of campaign analysis cut to under two — that is the number Featuring, a Korean influencer marketing startup, says its new AI agent delivers. It is one of three companies that pushed influencer automation past the discovery stage this month, and the pattern is the same everywhere: the agent recruits, negotiates, schedules and reports, while a human approves the key steps.

The tools arrive as the global influencer marketing market, pulled along by K-beauty, runs ever larger campaigns with creators worldwide. Someone has to recruit them, haggle over rates and chase product deliveries. Increasingly, that someone is software.

What do these agents actually do?

The pitch is simple. Type a marketing goal into a chat window — say, give sunscreen to American surfer creators and have them make content — and within seconds the agent proposes 50 micro-influencers and 50 nano-influencers, reviews the schedule and requirements, and produces a quote with estimated costs.

Storika, Indent Corporation and Featuring each rolled out AI agents, industry sources said on the 4th. Solutions that once stopped at finding influencers now pitch collaborations, negotiate terms and manage content, which is why the companies call them "AI marketers" rather than search engines with better manners.

Three technologies widened the scope of automation: vision AI that analyzes video and images, graph databases that structure relationships between creators and brands, and agent technology that chains multiple tasks in sequence. The buzzwords, for once, correspond to shipped products.

What did Indent Corporation ship?

Indent M, launched last month, builds on roughly two years of campaign plans and execution data gathered through the company's existing service, Spray IO. The agent analyzes a brand's marketing objective, suggests suitable influencers and estimated collaboration costs, then handles everything from recruiting through campaign operations to results reporting.

"We designed it so that global viral seeding marketing can be carried out in 72 countries," said Yoon Tae-seok, chief executive of Indent Corporation. "The goal is to lower the burden on small and mid-sized Korean brands of building overseas marketing and sales infrastructure." Seventy-two countries is a specific claim; the agent, presumably, does not need a visa.

What is Storika offering?

Storika unveiled an open beta service in August this year and applied its existing recommendation data to full campaign automation. Its agent analyzes data from roughly 8 million creator profiles and more than 100 million pieces of content to narrow down influencers, with AI agents assigned to specific roles handling negotiations within budget, product delivery tracking and post reviews.

"We analyze the relationships among creators, brands and content, reflect staff feedback in the next round of recommendations, and are improving accuracy through web searches and integration with outside services," said Bryce Lee, chief executive of Storika. The human stays in the loop, in other words, but the loop is getting quieter.

How does Featuring's agent work?

Featuring draws on information covering about 18 million influencer channels in Korea and abroad, plus some 150,000 conversation records accumulated while running campaigns. The agent analyzes how creators responded to past collaboration offers and the history of cost and schedule adjustments, then proposes terms for new campaigns. Past behavior as a negotiating script, in effect.

"By automating report writing for each advertiser and analysis of each influencer's advertising history, we have cut the time needed to analyze campaign performance from an average of six to eight hours to under two," said Choi Hwa-seop, chief technology officer at Featuring. "Marketers can now focus on planning instead of repetitive search work." A 75 percent reduction, achieved by deleting the repetitive parts of the job.

What comes next?

All three companies plan to extend the agents' remit into planning and performance analysis, and to improve decision accuracy. They intend to feed performance data gathered from campaigns into the planning and execution of follow-up advertising, raising efficiency so marketers can concentrate on creative work.

The trend line is clear enough: recruiting, negotiating, scheduling and reporting are moving to agents, and the marketers' job shrinks toward strategy, approvals and whatever "creative work" means once the machines have taken the paperwork. The numbers — 8 million profiles, 18 million channels, 72 countries, two hours instead of eight — suggest the vendors have data to stand on. What they do not yet have is a track record for the judgment calls that campaigns are made of. That part, for now, still comes with a human attached.