Case StudiesResearchBlogOpen SourceSolutionsAbout
Production-Grade AI Systems — Est. 2022

We build AIsystems thatactually work.

Drut AI is a research-driven AI engineering company. We build agents, fine-tuned models, memory architectures and the infrastructure that makes them production-ready.

0+
Papers Published
0+
OSS Packages
0%
Retention %
DRUT
Agent SystemsModel Fine-TuningMemory ArchitecturesAI InfrastructureInference OptimizationRAG SystemsAgent SystemsModel Fine-TuningMemory ArchitecturesAI InfrastructureInference OptimizationRAG Systems
Core Capabilities
01
Agent Systems
Autonomous agents with tool use, reasoning loops, and self-evaluation.
02
Model Fine-Tuning
Domain-specific LLMs with instruction tuning and cost reduction.
03
Memory Architectures
Long-term agent memory, user context, conversation persistence.
04
AI Infrastructure
Evaluation pipelines, inference optimization, deployment at scale.
05
RAG Systems
Knowledge base agents, enterprise search, long-context retrieval.
Featured Research

Latest from our lab

View all research →
Tool Use
Tool-Calling with Open LLMs: LLaMA, Gemma, and GPT-OSS
Open large language models (LLMs) like Meta's LLaMA, Google's Gemma, and OpenAI's GPT-OSS now support tool-calling (aka function-calling) to let the model invoke external APIs or tools. We survey how each model supports tool-calling, methods for impr...
Memory
Adaptive Long-Term & Short-Term Memory Architecture for AI Systems
We propose a dynamic memory architecture for AI agents that enables persistent cognitive behavior across sessions. By combining importance-based memory creation, vector-based short-term storage, decay dynamics, reinforcement mechanisms, and hierarchi...
Memory
RAG Memory Architectures for Long-Term Agents
We introduce a hybrid vector-symbolic memory architecture that enables persistent episodic memory in production AI agents. Our approach combines dense retrieval with symbolic consolidation, achieving 3.2× better recall over 30-day horizons while redu...
Case Studies

Results that speak

All case studies →
Research Automation
AI Research Agent for Market Intelligence
Autonomous research agent with multi-source retrieval, synthesis, and daily briefing delivery.
62%
Reduction in manual research time
Inference Cost
Customer Support AI with Long-Term Memory
Fine-tuned support model with persistent user context and conversation memory at scale.
40%
Reduction in inference cost
Agents
Sales Agent with Multi-Tool Reasoning
End-to-end sales agent with CRM integration, outreach generation, and pipeline management.
10×
Better retrieval accuracy
Open Source

Built for developers

GitHub →
Python
agent-memory
Long-term episodic memory system for production AI agents.
★ 2.1k stars
Python
rag-eval
Evaluation framework for retrieval-augmented generation pipelines.
★ 1.4k stars
Python
prompt-router
Intelligent prompt routing between models based on cost and capability.
★ 890 stars
NPM
agent-sdk
TypeScript SDK for building and orchestrating AI agents.
★ 743 stars
NPM
llm-eval
Lightweight LLM evaluation utilities for Node.js.
★ 512 stars
Rust
fast-vector-search
High-performance HNSW vector search with SIMD acceleration.
★ 1.8k stars
Contact

Let's build something exceptional together.

Whether you're launching an AI product, need a production-grade agent, or want to fine-tune a model for your domain — we respond within 24 hours.

Emailfounders@drut.ai
OfficeHyderabad, IN
HoursMon–Fri, 9AM–6PM IST