Overview
This solution is an end-to-end AI pipeline that automates the creation, training, and evaluation of specialized Small Language Models (SLMs). It first extracts business rules from uploaded SOPs and processes user-provided tool descriptions for the backend agent, integrating real customer data to build grounded, realistic scenarios. Using these constraints, a synthetic user agent converses with the backend agent to achieve specific scenario objectives, generating a high-fidelity dataset of complex tool-use interactions and edge cases.
This rich data is used to train a lightweight SLM. Through this distillation process, custom behaviors and SOPs that previously required bulky text context for the backend LLM are now fully internalized by the small LM. Finally, an LLM-as-a-Judge evaluates post-training interactions, rigorously verifying tool retrieval accuracy, tool use accuracy, correctness, and faithfulness to ensure the model is production-ready.
Highlights
- AgentSLM automates a hands-off lifecycle that transforms static SOPs and tool descriptions into an optimized Small Language Model (SLM). By distilling complex business logic and tool-reasoning behaviors directly into the model's weights, it eliminates the need for manual dataset creation and removes the dependency on bulky, context-heavy prompts in production. This creates a lightweight, high-performance agent that has your specific operational "playbook" natively internalized.
- The solution intelligently extracts granular rules and variables to construct realistic, grounded scenarios. A synthetic user agent is then deployed to engage the backend agent, driven by specific objectives to uncover complex tool-use behaviors and combinatoric edge cases. This process is capped by a specialized LLM-as-a-Judge framework that rigorously evaluates tool retrieval accuracy, tool use, and faithfulness, ensuring the distilled model is both highly capable and production-ready.
Details
Introducing multi-product solutions
You can now purchase comprehensive solutions tailored to use cases and industries.
Features and programs
Financing for AWS Marketplace purchases
Pricing
Dimension | Description | Cost/unit |
|---|---|---|
Per Pipeline Run | Billed each time an SOP document is processed and the generative AI pipeline executes. | $3.00 |
Vendor refund policy
NIL
How can we make this page better?
Legal
Vendor terms and conditions
Content disclaimer
Delivery details
Deploy AgentSLM on Amazon ECS (Fargate) using CloudFormation
- Amazon ECS
Container image
Containers are lightweight, portable execution environments that wrap server application software in a filesystem that includes everything it needs to run. Container applications run on supported container runtimes and orchestration services, such as Amazon Elastic Container Service (Amazon ECS) or Amazon Elastic Kubernetes Service (Amazon EKS). Both eliminate the need for you to install and operate your own container orchestration software by managing and scheduling containers on a scalable cluster of virtual machines.
Version release notes
This is the first version.
Additional details
Usage instructions
Support
Vendor support
For any assistance reach out to us at:
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.