Beatly, a Nordic influencer-marketing platform, from agency to AI-first Saas

AI decreases the value of content generation and shifts focus towards attention capture

Beatly’s CTO Erwan Lemonnier is pretty straight to the point with us in this interview: “When brand managers can generate campaign briefs in minutes and creators can spin out on-brand UGC fast, the agency that still edits briefs via email loses.” Beatly’s founders have seen this cliff coming in real time: In the agency business, monthly contracts are common, it’s problematic with the wait between pitch and publish that stretches the sales cycles and creates uncertainty. It is from here, the company wide priority to transform from agency to subscription platform, comes from. In a world that commoditizes everything, where image, text and complete statements can be prompted, how do you turn an agency into a trusted go-to subscription business that customers are willing to pay for?

Some context

CTO Erwan Lemonnier joined Beatly and took on the mission to create an infrastructure on which Beatly can move fast. He has already rebuilt Beatly’s campaign portal twice to keep pace with a chaotic social landscape.

Of course, AI is an obvious tool for transformation - it’s just figuring out how, the angle, what is production ready influencer subscription AI? The definition of done for v1 is difficult to articulate.

Large language models are ready to automate some of the slowest steps—drafting brief , finding creators, tagging assets, performance reporting—but only if the team could move quicker than the market. Erwan has, from the start, wanted to build a robust AI system around this - something that aligns well with the work he’s already done with the foundational tech infrastructure they now have.

From Campaign Tool to Always-On Platform

Erwan Lemonnier sees a clear shift in how users can engage with the platform. “We’re not just a tool for one-off influencer campaigns anymore,” he explains. “Users can return regularly—between campaigns—to plan, analyze, and iterate.” That stickiness signals a successful evolution from a single-purpose utility into a full-fledged SaaS platform. Initially, most customers come in with influencer marketing pain points. But as conversations deepen, Beatly uncovers a broader challenge: marketers—especially in smaller teams—are juggling disconnected tools and scattered workflows. “The real pain isn’t just campaign management,” says Lemonnier. “It’s the chaos around it.” To solve that, Beatly is integrating more of the marketing and campaign management workflow directly into the platform. A standout example is their use of large language models (LLMs) to streamline campaign planning. A newly launched feature allows campaign managers to auto-generate campaign briefs—matching the quality of traditional marketing consultancies, but at a fraction of the time and cost. “It’s about bringing the power of a full marketing team into one interface,” Lemonnier says.

The AI powered UGC platform in more detail

Beatly uses Opper.ai’s Python SDK, powering a fully self-service user-generated content (UGC) engine tailored for scale. Campaigns are dynamically generated using large language models—customized to each request, each user, and each brand.

Two LLMs work in tandem through this campaign builder:

• GPT-4o handles the strategy, narrative, and text content.
• Imagen 3.0 generates on-brand visuals.

The campaign builder guides users through six smart steps:

• Brand – Analyse the brand and what the marketing campaign will promoteIden
• Objectives – Define specific, measurable targets.
• Market Analysis – Contextual insights about the brand and its consumer landscape
• Audience – LLM-generated buyer personas and tone.
• Concept – Creative advertisement suggestions and hooks.
• Media Plan – Generate a fitting mediamix for the campaign, mixing content creation with social media distribution, both organic and paid, and amplification via third party channels

Each step triggers LLM calls to propose draft content, but everything is presented in editable modules—giving teams full control to tweak and optimize before going live.

Interview snapshot

Q: “Why not stay with a single LLM provider API?”
A: “Strategically, I never wanted to build myself into a lock-in position with one provider. I want flexibility to change models according to need and to easily swap in newer better models. JSON schemas and eval scores gave us the confidence to switch between models. We’re also using fallback models for reliability.”

Q: “Biggest surprise?”
A: “UI/UX and tracing together matters so much! Real-time status updates hide LLM latency; without tracing we couldn’t tune that experience.”

Impact, from prototype to profitability

Beatly quietly launched UGC Only about six months ago. The results are interesting. “On a company level, we’re more profitable than ever,” Lemonnier shared. AI has helped streamline development and increase efficiency and the UGC platform now actively supports campaign managers that are becoming increasingly efficient. What started as a prototype now has an impact on both product and efficiency.

A strong foundation for Agentic AI

With a stronger data foundation now in place, Erwan Lemonnier is focused on what’s next: embedding more intelligence directly into Beatly’s product. “The new data pipelines unlock a whole new layer of features,” he says. Central to this evolution is deeper LLM integration—pushing beyond content generation into decision-making and automation.

One area Beatly is actively exploring is agentic frameworks: AI systems that don’t just assist but take action. Lemonnier envisions a future where campaign workflows are executed by agents that can handle tasks like setting up briefs, reaching out to creators and optimizing spend. “We’re evaluating whether to define a clear list of allowed actions or let the agent interact with our full internal API,” he explains.

This isn’t just a tech experiment—it’s a vision for the future of marketing agencies. Lemonnier believes that AI-powered agents will eventually take over much of the manual work done today, enabling leaner teams to do more with less. Beatly intends to be at the leading edge of that shift.

The real edge comes from combining agentic AI with Beatly’s deep domain knowledge.