A week that shows how the cost of building software is changing, another leap in image generation and a tool to scale PostgreSQL databases.
One engineer and an AI model rebuilt Next.js in one week
Cloudflare says one engineer, with the help of an AI model, reimplemented the Next.js API on top of Vite in one week, for about $1,100 in tokens. The result, called vinext, replaces Next.js with no code changes and deploys to Cloudflare Workers with one command. In early tests, it built apps up to 4 times faster with bundles up to 57% smaller, and some customers already run it in production.
The case shows that projects once too expensive, such as rewriting a framework or replacing a system's foundation, are now within reach of small teams with AI. Companies locked into a platform by technical limits can revisit migrations that seemed out of reach.
How Dexpert helps: Web DevelopmentSources
- The Cloudflare Blog: How we rebuilt Next.js with AI in one week
Nano Banana 2 combines the Pro model's quality with Gemini Flash speed
Google DeepMind launched Nano Banana 2 (Gemini 3.1 Flash Image), an image generation and editing model that combines Nano Banana Pro's features with Gemini Flash speed. The model keeps people and objects consistent across images and follows instructions more precisely. It is rolling out to the Gemini app, Search and Google Ads, and Google strengthened the identification of AI-generated images with SynthID and C2PA credentials.
Creating and adjusting product images, campaigns and sales material gets faster and cheaper. Retail and marketing companies can integrate this kind of model into their own systems to produce variations at scale.
How Dexpert helps: Artificial Intelligence
PgDog scales PostgreSQL with connection pooling, load balancing and sharding
PgDog is an open source proxy written in Rust that sits between the application and PostgreSQL. It pools connections, like PgBouncer, but also spreads queries across replicas and splits whole databases into shards, with no changes to application code. The project can be deployed with Helm on Kubernetes, including with databases on AWS RDS.
When the database becomes the bottleneck, a proxy like this can delay or avoid an application rewrite. Test it before moving to a different database or a more complex architecture.
How Dexpert helps: Backend & APIsWant to put one of these to work in your company?
Dexpert builds custom systems, apps, integrations and AI solutions. Tell us what you need and get an estimate, or take the free digital maturity assessment.