Introducing Klara v1.5: Better Answers, Better Safety, Same Vision

Klara launched as the AI layer of Polkassembly with a simple goal: increase adoption on Polkadot

Version 1 showed that people want governance information delivered in the same way they think about it: natural language, context-aware, and fast.

Today, we are releasing Klara v1.5. This update refines how Klara responds, improves clarity, and adds tighter guardrails so users get accurate, grounded outputs every time.

Klara still solves the same problem. The difference now is quality.

What’s New in Klara v1.5

1. More conversational, without losing accuracy

Klara now produces clearer, more direct answers. When users ask about governance mechanics, treasury flows, tracks, or identity, she delivers explanations that feel more natural without compromising correctness.

Examples:

• “Break down what happened in this referendum.”
• “Explain why conviction matters in one sentence.”
• “What is the difference between identity and judgement”

Responses are now structured, consistent, and easier to read.

2. Improved guardrails and safety behavior

We tightened Klara’s guardrails to ensure:

• stricter adherence to on-chain facts
• clearer disclaimers when data cannot be verified
• better differentiation between Polkadot, Kusama, and People Chain contexts

Klara v1.5 handles edge cases, missing data, and ambiguous questions more reliably.

3. Expanded knowledge coverage from real user questions

Klara v1.5 now handles more governance questions that users asked in the first release:

• Track parameters and deposit rules
• Preimage vs decision deposit
• Treasury thresholds by track
• Identity flows on People Chains
• Teleporting funds for identity
• Judgment types and registrar actions
• Bounty lifecycle
• Voting momentum and conviction distribution

This brings Klara closer to full OpenGov literacy.

Why We Built v1.5

Klara’s long-term direction is unchanged. She is not a chatbot. She is the AI layer for Polkadot governance.

To get there, each update focuses on:

• better precision
• clearer reasoning
• safer responses
• stronger knowledge boundaries
• more predictable behavior for developers, users, and contributors

v1.5 moves Klara closer to being a reliable, developer-ready component that parachains, tooling teams, and governance dashboards can build on.

What’s next?

The v2 version will extend beyond response quality into capability expansion.

They key focus for v2 will be data decentralisation and verifiability of responses.

Each release is structured in a way that it builds towards a shared architecture where Klara becomes the intelligence layer that any team in the Polkadot ecosystem can use.

Try Klara v1.5

Klara is live on Polkassembly: **klara.polkassembly.io**

If you used v1, you’ll notice the difference immediately. If you haven’t tried Klara yet, this is the best time to start.

Klara remains open source and community-driven.Developers can explore the repository, open issues, and contribute improvements. Klara isn’t perfect, even Chatgpt can make mistakes. Help us improve Klara by testing it and sharing feedback.

Stay Tuned for V2…

Author: Bhavya Batra, Head of Product, Polkassembly