← Back
AWS ML Blog

Adaptive agentic intelligence framework using Strands Agents

•10 min read•
#agents#amazon#llm
Level:Advanced
For:AI Engineers
✦TL;DR

The Adaptive Agentic Intelligence Framework (AAIF) uses a 1‑billion‑parameter llama3.2 SLM as a lightweight orchestrator to classify and route requests between local Ollama‑hosted models (including llama3.2‑vision) and cloud‑hosted frontier models accessed via Amazon Bedrock. By offloading routine telemetry and vision tasks to on‑device SLMs, AAIF cuts API costs and latency while still leveraging powerful Bedrock models for complex, multi‑step reasoning. Engineers can adopt this three‑layer architecture to achieve hybrid edge‑cloud AI deployments without redesigning existing infrastructure.

⚡ Key Takeaways

  • The orchestrator model is llama3.2:1b running on Ollama.
  • Architecture separates routing (SLM orchestrator) from processing tools (local telemetry, vision, and Bedrock expert insights).
  • Local processing runs on edge hardware such as a Raspberry Pi with 8 GB RAM, delivering low latency and no cloud dependency.
  • Cloud processing is invoked through the get_expert_insights tool using boto3 to call Amazon Bedrock.
  • The framework requires Ollama runtime for both the orchestrator and vision SLMs; without Ollama, local routing cannot function.
💡 Why It Matters

AAIF lets production teams dramatically reduce cloud inference spend and latency for high‑frequency, simple queries while preserving the ability to tap into state‑of‑the‑art LLMs for demanding analyses.

✅ Practical Steps

  1. Deploy Ollama on your edge device and load the llama3.2:1b orchestrator and llama3.2‑vision models.
  2. Implement the three Strands Agent tools (get_device_telemetry, analyze_camera_feed, get_expert_insights) and configure get_expert_insights to call Amazon Bedrock via boto3.
  3. Run the Strands agent, allowing the orchestrator to classify incoming requests and automatically dispatch them to either the local Ollama tools or the Bedrock endpoint.

Want the full story? Read the original article.

Read on AWS ML Blog ↗

More like this

Introducing workload recovery and expanded infrastructure checks on Amazon SageMaker HyperPod

AWS ML Blog•#deployment

Introducing Gemini 4 Neon

Google DeepMind•#llm

DaisyRose-27B: Pushing the Frontier of Post-Training Efficiency

Hugging Face Blog•#llm

Mathematical proof, powered by AI. And no hallucinations!

Amazon Science•#agents

EXPLORE AI NEWS

Daily hand-picked stories on LLMs, RAG, agents and production AI — curated for engineers who ship.

BROWSE NEWS

GET THE WEEKLY DIGEST

Join engineers getting the Monday signal-over-noise AI breakdown. No spam, unsubscribe anytime.

LEARN AI ENGINEERING

Curated courses, research papers, repos and tutorials built for engineers leveling up in AI.

START LEARNING