Hello from Surya Digital!

The world of AI went through a shake up due to Cursor’s pricing changes—taking effect recently. Yes, waves were created due to a lack of transparency in pricing but the truth of the matter remains: subsidised pricing for AI-backed products isn’t sustainable in the long run.

For Lidana, we have maintained a consistent policy of transparency in pricing with our customers. Our pricing is rooted in the reality of LLM pricing offered by vendors, so our customers can Generate with AI without unexpected bills.

As for our usage of tools like Cursor, internally, our team continues to be bullish on tools like Claude Code, and Gemini CLI which are able to produce better results when anecdotally compared to Cursor’s Agent mode.

We suppose pricing, as with general software development, shouldn’t be so…cursory. On with the updates of the month:

Under the Hood

Transaction Isolation Level In PostgreSQL

READ COMMITTED (default), REPEATABLE READ, and SERIALIZABLE — know how they differ, what anomalies they allow or prevent (e.g. non-repeatable reads, phantom reads, write skew), and what performance trade-offs come with each. 

Building a Scalable Theme System in React

Discover how to build a scalable theming system in React by using context and design tokens. This guide walks through creating flexible, reusable themes that keep your UI consistent as your app grows.

Breakroom Buzz

Hermes V3: Building Swiggy’s Conversational AI Analyst
Swiggy’s Hermes v3 shows how AI is evolving from assistants to true teammates—making data accessible to everyone, not just analysts. A smart step toward democratizing insights.

UUIDv7 Comes to PostgreSQL 18

One of those small-but-significant changes: UUIDs have always been great for uniqueness and distributed systems, but their randomness has often come at a cost in database performance. With UUIDv7, Postgres is bridging that gap—making UUIDs more “friendly” to index ordering without sacrificing their benefits. For teams building large or distributed systems (or anyone who’s struggled with UUID index bloat), this might be the moment to reconsider using UUIDs more broadly.

Off The Assembly Line

One of the questions we often hear from teams is: “Can we use Lidana within Azure DevOps?” The answer is now a resounding yes. We’ve made Lidana compatible with ADO, so your team can generate AI-powered test cases right where your projects live. No switching tools, no extra steps—just faster, smarter QA inside your existing workflows.

You can book a call with us here to learn more.