{
  "id": 12449778,
  "title": "Building a context-aware AI assistant on AgentCore and OpenClaw",
  "url": "https://urgent.news/2026/10/06/building-a-context-aware-ai-assistant-on-agentcore-and-openclaw",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-06T19:19:15.000Z",
  "source": {
    "name": "AWS Machine Learning",
    "slug": "aws-machine-learning",
    "url": "https://aws.amazon.com/blogs/machine-learning/building-a-context-aware-ai-assistant-on-agentcore-and-openclaw/"
  },
  "original_language": "en",
  "account": "Building a context-aware AI assistant is challenging, as off-the-shelf assistants struggle with continuity. They provide accurate answers individually but fail to maintain any understanding of previous conversations. The user must constantly re-explain context, leading to a poor user experience.\n\nThe solution lies in building a personal assistant that accumulates context over time. This is achieved using OpenClaw, an open source agentic system, running on AgentCore runtime, a capability provided by Amazon Bedrock AgentCore. AgentCore memory allows the assistant to transform one-off chats into long-lasting knowledge.\n\nThe example presented is Sprout, a gardening assistant, but the architecture is completely domain-agnostic. By swapping the persona and skills, the same pipeline can serve a support bot, fitness coach, or internal help desk. All of this runs within a single AWS CloudFormation template, deployable with a single command. The system operates on a consumption-based model, costing just a few dollars per month for light personal use.\n\nThe end-to-end request flow begins with inbound Telegram webhook or Amazon EventBridge schedules, both invoking the same AgentCore runtime agent. The agent coordinates OpenClaw gateway, AgentCore memory, and Amazon Bedrock Converse API. Telegram messages come through Amazon API Gateway and a webhook AWS Lambda function, while scheduled jobs like watering reminders come through Amazon EventBridge Scheduler and a cronjob Lambda function. Both trigger the same InvokeAgentRuntime API on the AgentCore runtime. A thin server.py process handles coordination between OpenClaw gateway, AgentCore memory, and Bedrock's Converse API.\n\nAmazon S3 provides workspace storage, AWS Key Management Service handles encryption, AWS Secrets Manager stores the bot token, and Amazon CloudWatch captures logs and metrics. The deployment requires Amazon Bedrock AgentCore access, including AgentCore runtime and memory. Models like Claude Haiku 4.5 for text and Claude Sonnet 4.5 for vision (or equivalent models) are needed. Docker with linux/arm64 build support and AWS CLI configuration are also prerequisites. Familiarity with agent orchestration concepts and CloudFormation is helpful but not mandatory.",
  "summary": "Off-the-shelf AI assistants forget you between conversations. This post shows how to build a personal assistant that accumulates context using OpenClaw on Amazon Bedrock AgentCore runtime, with AgentCore memory turning disposable chats into durable, structured knowledge you can retrieve with metadata filters.",
  "key_points": [
    "Sprout is a gardening assistant built on OpenClaw and AgentCore",
    "AgentCore memory enables long-lasting context accumulation",
    "System runs on AWS CloudFormation template with single deploy command"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}