What changed

AWS announced the Adaptive Agentic Intelligence Framework (AAIF), which uses small language models (SLMs) for routine tasks and routes complex requests to cloud-hosted models via Amazon Bedrock. This framework is built with Strands Agents and is designed to reduce API costs, latency, and dependency on cloud infrastructure for simple queries.

What this means in practice

The framework organizes AI processing into three layers: smart orchestration, local processing, and cloud processing. Smart orchestration uses a lightweight SLM (llama3.2:1b) to classify incoming requests and route them to the appropriate processing layer. Local processing handles routine tasks like device telemetry retrieval and image analysis using SLMs running on local hardware through Ollama. Cloud processing handles complex pattern analysis and strategic recommendations through frontier models in Amazon Bedrock.

Limitations and availability

The framework is available for use with an AWS account, Amazon Bedrock model access, Python 3.10+, Ollama, and Strands Agents. It runs on edge devices like a Raspberry Pi with 8 GB of memory, in full cloud deployments, or in hybrid configurations. The framework adapts to deployment constraints without requiring infrastructure restructuring. The solution uses an IoT monitoring scenario as a reference implementation, with tools like get_device_telemetry, analyze_camera_feed, and get_expert_insights for local and cloud processing.

Source facts for AWS Introduces Framework to Route AI Requests to Appropriate Models: This one-size-fits-all approach can drive up API costs, introduce unpredictable latency, and remove the possibility of offline operation. Even routine queries that could resolve locally end up requiring a full network round-trip to a cloud-hosted large language model (LLM). The Adaptive Agentic Intelligence Framework (AAIF), built with Strands Agents, takes a different approach.
Original AINewsTrend explainer based on the cited source. This graphic summarizes source statements; it is not an independent benchmark.

Sources and further reading

  1. awsml: Adaptive agentic intelligence framework using Strands Agents