Product engineering, design, and agent systems

Inside Anuvaya

Technical notes from the team building Vaya, Terra, Rune, and the systems behind agentic professional services.

Writing

note

Multi-Agent Conversational AI

What production conversational AI takes once you stop treating it as request-response: instant talk, slow thinking, one hard invariant, and patience for users who don't wait their turn.

Nitesh Kumar Niranjan

5 min read

note

Terra: An Agent Framework Extracted from Production

An Elixir agent framework built on gen_statem. Stateful processes, context aging, multi-agent coordination. Open source.

Nitesh Kumar Niranjan

5 min read

note

Rune - Long-Term Memory for AI Agents

An append-only knowledge graph that captures how a user's life evolves over time. Deterministic, causal, and fast enough for real-time conversation.

Nitesh Kumar Niranjan

6 min read

note

Your Eval Is Broken for the Same Reason Your LLM Is

Lessons learned from building LLM evaluation systems - the right prompts, the right context, and knowing what to measure.

Jashn Maloo

9 min read

note

RCE - Realtime Context Engine

A framework for managing dynamic knowledge bases, temporal data hierarchies, and externalized reasoning chains within conversational AI sessions.

Nitesh Kumar Niranjan

5 min read

note

Stateful Agent Orchestration - Human-Like Conversational AI

An orchestration architecture that enables human-like conversational behavior with natural delivery pacing, mid-response interruptibility, and graceful recovery.

Nitesh Kumar Niranjan

6 min read

note

The Boring Designer

A note on the importance of boring design.

Nitesh Kumar Niranjan

2 min read

Follow the work

Subscribe through RSS or use the language-model-friendly index when you need the full technical corpus.