Accel is doubling down on its efforts on Fibr AI as agents turn static websites into one-on-one experiences

While advertising and targeting have become increasingly personalized, the website — the final destination of that traffic — has remained largely constant. Artificial Intelligence Fibers It aims to bridge this gap by using AI agents to transform generic web pages into individual experiences tailored to each visitor, a thesis that has prompted Accel to redouble its efforts at the company.

Accel led Fibr AI’s $5.7 million seed round after a previous $1.8 million investment in 2024. The new funding also included participation from WillowTree Ventures and MVP Ventures, along with Fortune 100 operators who joined as angel investors and advisors, bringing the startup’s total funding to $7.5 million.

For large companies, the gap between increasingly personalized ads and generic website experiences has traditionally been bridged largely through a combination of personalization software, engineering teams, and marketing agencies — a model that is slow, expensive, and difficult to scale. While ads can be tailored on the spot to suit different audiences, changing what happens once a visitor lands on the site often requires weeks of coordination and limits teams’ ability to run only a few experiments each year. Fibr AI argues that this human-heavy operating model no longer works. Instead, the startup uses autonomous AI agents to infer intent, generate variations, and continuously optimize pages in real-time.

Fibr AI replaces the agency and engineering-heavy model with autonomous systems that operate continuously, Ankur Goyal, co-founder and CEO, said in an interview.

“We are the software, and the agency is the workforce for the agents that we deploy,” Goyal told TechCrunch, adding that this allows Fibr AI to run thousands of experiments in parallel instead of a few dozen each year.

Adoption was slow at first. Founded in early 2023 by Goyal and Pratam Roy (pictured above, left), Fibr AI had just one or two customers for most of its first two years as companies took time to evaluate the approach. That began to change last year, Goyal said, as adoption spread among large U.S. companies, including banks and health care providers, bringing the total number of customers to 12.

“We are an infrared layer,” Goyal told TechCrunch. “Once you set it up, no one wants to think about it again.” This dynamic has led Fibr AI to sign three- to five-year contracts with large enterprises, which tend to treat website infrastructure as something to be standardized rather than constantly revisited, he added.

On a technical level, Fibr AI acts as a layer on top of an existing website, connecting to a company’s advertising, analytics, and customer data systems to understand how visitors arrive and what they are likely searching for. AI agents then collect and adjust page content, such as copy, images, and layout, treating each URL as a system that is continuously learning and improving rather than a static page. Instead of relying on manually configured rules or sequential A/B tests, the platform runs a large number of small experiments in parallel and systematically updates the experiments as traffic flows from different channels.

This shift has direct cost implications for large companies. Traditional website customization typically combines software licenses, agency agents, and engineering time, tying costs to people rather than results. Goyal said companies are increasingly evaluating the Fibr AI platform based on the cost of the experience and conversion impact, rather than the number of tools or people involved.

For Accel, this operating model – not the AI ​​hype – was a key element in the decision to invest again. “Advertising today is one-to-one, but when users land on a website, it becomes one-to-many,” said Priyank Swaroop, partner at Accel. “You can create hundreds of ads for different audiences, but they all still land on the same page.” Fibr’s ability to transform this dynamic into individual customization stood out because it removed agency and engineering bottlenecks that typically limit the extent to which companies can drive experimentation, he said.

Swarup added that early institutional adoption, especially among banks and healthcare companies, helped validate the thesis. “These are regulated and conservative industries,” he said. “When they start saying, ‘We need this, and we’re willing to pay for it,’ then we feel confident to redouble our efforts.”

Future-proofing the era of agent trading

While most of Fibr AI’s work today is based on personalizing experiences for human visitors, Accel and Fibr AI also see potential in how AI agents can begin to mediate online discovery. As users increasingly research, compare and shortlist products using large language models and AI-based chatbots, including OpenAI’s ChatGPT, before visiting a website, Swarup said, the ability of sites to adapt based on what the visitor knows — or the AI ​​system working on their behalf — could become more important over time.

“This part is still early, but companies that are working to meet the needs of today while preparing for this transformation tomorrow are the companies we want to support,” Swarup said.

With the new funding, Fibr AI plans to focus on expanding its sales and customer-facing teams in the US, while continuing to build its technology base in India. The San Francisco-headquartered startup maintains an office in Bengaluru, with 17 of its 23 employees based in India and the remaining six in the US.

The startup is targeting about $5 million in annual recurring revenue by the end of this year and about 50 enterprise customers, Goyal said.

Fibr AI enters a field long dominated by established companies like Adobe and Optimizely, which provide experimentation and customization tools for large enterprises. But Goyal and Swarup argued that these platforms are limited by how they are built and sold, and typically rely on marketing agencies and engineering teams to configure and operate them. This model makes it difficult to move quickly or scale experimentation, even as customer acquisition and messaging become more dynamic, they said.

“Incumbent companies have been slow to roll out products,” Swarup said, adding that even when new features arrive, they often come years after demand has shifted.