Pull up ChatGPT and ask it to recommend project management tools. Now try Gemini. The lists won't match—and if your brand doesn't appear on either, you've got a problem traditional analytics can't diagnose, much less fix.
This is the wedge sitefire is driving into the market. The Munich-based startup, which emerged from Y Combinator earlier this year, isn't content to simply tell companies where they're missing from AI-generated recommendations. Instead, the platform attempts something considerably more ambitious: it generates the content designed to capture those citations, then publishes it directly into customers' content management systems.
Whether that crosses the line from useful automation to brand risk depends partly on your tolerance for machine-written marketing copy. But it represents a notable evolution in what founders Jochen Madler and Vincent Jeltsch call "Generative Engine Optimization"—optimizing not for Google's algorithm, but for the citation logic of ChatGPT, Gemini, Perplexity, and their algorithmic kin.
"Citation rate differs wildly across models," Madler said in a March exchange on Product Hunt, describing the team's daily work probing hundreds of prompts to map how different AI systems surface sources. ChatGPT leans on something approximating Bing's index, he noted, while Gemini hews closer to Google's. Those aren't trivial differences when you're trying to ensure your SaaS product or financial service appears in the answer.
Beyond the Dashboard
Most tools in this nascent category—Otterly.ai, BrndIQ, LLMClicks among them—focus on monitoring. They'll track whether your brand gets mentioned, how often, and in response to which queries. Useful intelligence, perhaps. But then the work begins: figuring out what content to create, writing it, optimizing it, publishing it.
sitefire collapses that workflow. The platform ingests a company's sitemap, identifies which pages currently earn citations, then reverse-engineers the patterns AI models seem to favor. From there it maps what the founders call "citation pathways"—not just your own pages, but third-party coverage, user forums, news mentions that influence how AI models construct their responses.
The "Earn Media" module goes further still, identifying journalists by coverage patterns and suggesting data-driven outreach angles. Integration with Google Search Console and CDN logs rounds out the intelligence layer. Then comes the automation: sitefire generates articles based on what's working for competitors, pushes drafts into Webflow or Framer, and waits for human approval before anything goes live.
"We went from 'here's what you should do' to 'we did it for you,'" Madler said around early March, calling it their "hello world moment."
The human review gate matters. "We don't auto-publish quite yet," co-founder Jeltsch clarified days later on the same thread. Customers vet drafts before publication, and the system logs those edits to refine future outputs. It's a concession to reality: few marketing directors are ready to let algorithms write and publish brand-critical content unsupervised, no matter how sophisticated the model.
The Enterprise Question
Launch materials name BMW, Xtrackers, and DWS as enterprise customers—notable brands, if accurate. No independent confirmation from those companies has surfaced, and sitefire's founders didn't respond to requests for case study details or customer references. It's early days, and perhaps more formal partnerships are still being finalized.
Madler's background blends academic rigor with practical finance. He studied deep reinforcement learning at Stanford before pursuing doctoral work at the Technical University of Munich. Jeltsch came from software engineering at robotics startup RobCo and launched a healthcare initiative in Germany. Both passed through CDTM, Munich's interdisciplinary technology program known for blending design thinking with technical depth.
The company operates through pulse Energy GmbH, registered between Munich and Delaware—a structure common among European startups eyeing U.S. expansion.
What It Costs
Three pricing tiers frame the offering, each bundled with automated content generation rather than treating it as an add-on. Lite runs $249 monthly for 150 tracked prompts across ChatGPT, Gemini, and Perplexity, eight action briefs, and three generated articles. Pro doubles to $499 with 300 prompts, 18 briefs, CMS integration, and a personal onboarding call. Enterprise pricing is custom, adding human-in-the-loop workflows, custom brand positioning, AI crawl analytics, and dedicated Slack support. Seven-day trials come standard.
The pricing structure telegraphs ambition. This isn't analytics software with a content suggestion feature tacked on. It's infrastructure betting that brands will pay for execution—that the real value isn't knowing you're absent from AI citations, but having a system that actively works to change that.
Competitors take different approaches. Wix rolled out native AI optimization features last August, though confined to its own platform. Yext published research in October analyzing 6.8 million citations across AI models, documenting how source preferences vary—but offered strategy frameworks, not automated implementation. The gap between analysis and action remains wide across the category.
Agents All the Way Down
sitefire's launch coincides with a broader enterprise fascination with "agentic" marketing systems—software that doesn't just recommend actions but executes them. Adobe unveiled its Agent Orchestrator in March of last year. Salesforce shipped Agentforce the previous September. AppsFlyer, Siteimprove, and Netcore all released products through late last year with "agentic" in the marketing copy, though most focused on customer experience orchestration rather than AI discovery.
The semantic shift reflects genuine changes in how marketing technology operates, or at least how vendors want it perceived. Madler and Jeltsch have hinted at deeper agent integration ahead—Slack bots, Model Context Protocol management, workflow orchestration from chat interfaces. For now the product addresses a more immediate pressure point: ensuring your brand appears when potential customers ask AI for recommendations, and doing it faster than competitors can manually optimize their way there.
Academic papers on Generative Engine Optimization proliferated through the latter half of last year and into this one, formalizing what had been ad hoc experimentation. But formalizing best practices and automating their implementation are different propositions entirely.
The Trust Problem
Whether enterprises will let automated systems generate and publish content tied to brand positioning and customer acquisition remains genuinely uncertain. The human review requirement suggests sitefire's founders recognize the stakes—and the discomfort many marketing leaders feel about ceding creative control to algorithms, even for something as tactical as SEO-adjacent blog posts.
Then again, reluctance might be a luxury brands can't afford much longer. AI answer engines are already replacing traditional search as the first touchpoint for product discovery and vendor research across categories. Waiting for perfect, fully trustworthy solutions could mean losing the market position before the tools mature.
That's the bet anyway. Whether citation visibility in AI responses translates to actual customer acquisition at scale—and whether automated content generation proves more efficient than human writers who understand brand nuance—are questions that won't resolve quickly. But in the meantime, someone has to build the infrastructure for a world where Google search is just one discovery channel among many, and perhaps no longer the dominant one.
sitefire is wagering that someone might as well be them.
