WeWeb goes big and puts AI into its no-code app builder

WeWeb joined the Y Combinator accelerator programme in 2021. This was a major turning point for the start-up, which was then focused on creating static websites front-ended with headless CMS. “We were applying as a straightforward website builder. But user feedback was clear: they wanted to build web apps, not just static websites,” recalls Raphaël Goldsztejn, CEO and co-founder.

Attentive to market demands, WeWeb decided to shift to a web app builder model. This change required more complex infrastructure: integration with existing backends, authentication management, publication of project-specific code, customised storage, peak traffic management, etc. None of this could be easily achieved with the existing Heroku environment.

“When we wanted to build a CDN per customer, register domains, or automate SSL certificate generation, we ran into Heroku's limitations. We needed more flexibility and granularity,” explains Marc Fabre, CTO and co-founder. WeWeb then began its migration to AWS, driven by a need for quick scalability without compromising reliability.

This infrastructure change allows the team to lay the foundations for the new product. The platform has entered a new phase of development, focused on performance, automation and scalability. This change will be supported by AWS services to take WeWeb to the next level.

When WeWeb transitioned to AWS, none of the co-founders were infrastructure experts. However, the need was urgent: to provide more powerful, scalable, and flexible hosting, especially for critical projects. “We started with Elastic Beanstalk because it was similar to Heroku. But we quickly realised that we needed something more to grow,” says Marc Fabre.

WeWeb rolled out Amazon S3 to store files generated when apps are published, and CloudFront to distribute them. This combo is a game changer. Before, each page was generated on the fly. It took up to 1.5 seconds to display a site. Now it takes 18 to 20 milliseconds. This new model also allows WeWeb to absorb traffic spikes, as it can now capitalise on AWS's uptime with its prospects and customers.

WeWeb relies on Amazon ECS to streamline container execution and automatically manage peak loads. When traffic spikes, new tasks are instantly launched. Once demand stabilises, unused resources are freed up, without any manual intervention. This elasticity becomes a pillar of the reliability of the service.

By migrating its bases to Amazon RDS, WeWeb gained greater control. With RDS, we can precisely size the bases, use read replicas, and change the storage type to increase IOPS. WeWeb has gradually expanded its architecture with other AWS services: ElastiCache Redis for the most demanding cache systems; Parameter Store to secure environment variables; Route 53, for centralised management of domain names and certificates.

In line with WeWeb's growing number of large clients in highly regulated industries, security expectations are becoming more stringent. To meet these requirements, the team is gradually deploying: AWS WAF to protect CDNs and load balancers from DDoS attacks or malware injections. CloudTrail and AWS Config, currently being implemented as part of the SOC 2 certification process.

“Safety and compliance are a priority for WeWeb in 2025. These are important requirements for our customers, and they also enable us to consolidate our editor,” summarises Raphaël Goldsztejn.

With AI-based code generators on the rise, WeWeb decided to turn this threat into an opportunity. The team made a bold move to natively integrate AI into the platform while keeping the original promise of no-code: empowering people who can't code.

And so WeWeb 3.0 was born, a new version of WeWeb that allows users to automatically generate an app from a prompt and then take control using the visual editor. “What the AI generates can be modified piece by piece. This is what allows users to stay in control — and therefore trust the editor,” he explains.

This is where WeWeb turns to Amazon Bedrock, and more specifically Anthropic's models. “We use the Bedrock API to perform inferences. It's simple, effective, and allows us to scale,” says Marc Fabre, CTO. The switch to Bedrock has made it possible to absorb the increased load associated with surging usage: since the launch of WeWeb 3.0, weekly registrations have increased fivefold.

By simplifying interface generation, AI significantly reduces the no-code learning curve while making the tool more accessible. “For us, AI is a key lever to streamline app development in SMEs and digital factories within large companies,” concludes Raphaël Goldsztejn.

WeWeb is a no-code platform that allows you to create web apps without coding, while retaining the flexibility to add code if needed.  Thanks to its decoupled architecture, it can be integrated with any backend and deployed on AWS or internally.