AI Proof of Concept on AWS — 4-Week Build by Horizon Digital

What we deliver: • Week 1 — Discovery: We agree the use case, success criteria, and data sources. We provision the right AWS AI services for your use case (Amazon Bedrock and/or Amazon SageMaker), set up Arize Phoenix for LLM observability, and define the evaluation rubric. • Week 2 — Build: We implement the AI agent on AWS using the foundation models best suited to your use case. We wire in retrieval-augmented generation (RAG) over your chosen data, add guardrails, and instrument every prompt with Phoenix. • Week 3 — Evaluate: We run the agent against your test cases, capture quality, latency, and cost metrics, and iterate on prompts and retrieval. You get full visibility into what the model is doing and why. • Week 4 — Handover: We deliver the working POC, a written production-readiness plan covering security, scale, cost, and governance, and a roadmap for moving to production.

Who this is for: • SMB tier: A single use case, one data source, one user group. Best for teams validating a focused GenAI idea. • Enterprise tier: Multiple use cases or data sources, agent orchestration, SSO integration, and a deeper production-readiness assessment.

Why Horizon Digital: • AWS Partner with delivery experience across AWS AI services (Amazon Bedrock and Amazon SageMaker) and leading foundation models. • Arize Phoenix observability built into every engagement — tracing, evaluation, drift detection, and cost telemetry. • Australian-based delivery team with experience across regulated industries. • Fixed fee, fixed scope, fixed timeline.

Related AWS services: Amazon Bedrock, Amazon SageMaker, AWS IAM, Amazon S3, AWS Lambda, Amazon OpenSearch (for RAG), CloudWatch.