media Latent Space · 1h ago · agents

Glean uses model routing to cut costs and select optimal LLMs

from English

Driven by rising frontier model costs and the power of open-weight models, Glean employs model routing to dynamically select the most appropriate AI model for each enterprise task. This approach allows the platform to avoid using large language models for simple operations while optimizing performance and expense for complex queries.

Glean offers three levels of model selection: explicit user choice, administrative restrictions, and automatic dynamic mode.
Automatic mode is primarily adopted by customers for economic reasons, with co-founder Tony Gentilcore claiming the system is 4x more cost-effective than Claude Code.
The platform averages $0.45 per task compared to $1.84 for Claude Cowork, leveraging its harness and routing capabilities.
Glean’s widespread deployment across organizations like Zillow and Booking.com provides extensive data on how business users interact with AI.

By deeply understanding organizational knowledge and user behavior, Glean aims to serve as a comprehensive meta-harness that combines the power of multiple leading LLMs into a single, cost-efficient experience for enterprises.

Importance 2/3 Latent Space AI agents Retrieval & RAG

Related articles

media Hugging Face Forums · 1d ago · 1 view

AlphaAvatar v0.6.6 introduces event-driven memory, audio perception, and unified backend contracts

AlphaAvatar v0.6.6 shifts the framework's focus toward clearer runtime contracts and modular capabilities rather than introducing a single large feature. The release expands Environment Memory to include audio segments alongside visual observations and transitions memory updates from periodic polling to an event-driven model.

media Hugging Face Forums · 13d ago · 27 views

Aletheion releases ASM-CM, a compact memory architecture for persistent AI agents

Felipe Maya Muniz of AletheionAGI introduces ASM-CM (Aletheion Compact Memory Model), an experimental architecture that separates memory from language to allow agent history to grow without increasing active computational memory.

media Hugging Face Forums · 16d ago · 24 views

IntelShed combines hybrid RAG, GNN entity resolution, federated learning, and LLM multi-agent orchestration

The author presents IntelShed, an open-source system for open-source intelligence (OSINT) that aggregates 50 public data feeds to produce automated 24-hour security briefings. The architecture integrates hybrid retrieval, GNN-based entity resolution, federated learning with differential privacy, and LLM-compiled multi-agent orchestration.

media Hugging Face Forums · 32d ago · 27 views

Hierarchical Agentic Memory uses hyperbolic embeddings for structured retrieval

A new memory architecture employs hyperbolic embeddings to organize vector database entries by an "abstraction score," allowing agents to manage knowledge more effectively than flat RAG systems. This approach structures data so that abstract entries, like preferences, sit near the center while concrete events reside at the boundary, facilitating better clustering and retrieval.

arxiv arXiv cs.CL · 34d ago · 15 views

RAGthoven uses multi-stage RAG pipeline to tie Gemini 2.5 Flash in multilingual humor generation

The authors present RAGthoven, a system for SemEval-2026 Task 1 that decomposes creative text generation into a multi-stage large language model pipeline grounded in computational humor theory. The final configuration augments the Planner with retrieval-augmented generation from a curated joke corpus and evaluates agentic variants against a non-agentic baseline.