%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23e60012;%20}%20.st1%20{%20fill:%20%23666;%20}%20.st2%20{%20fill:%20%23111;%20}%20%3c/style%3e%3c/defs%3e%3cpath%20class='st2'%20d='M208.79.2c10.12-1.03,18.93,1.81,26.29,8.72l77.13,96.19c2.89,2.23,6.06-.63,6.05-3.83l-.02-87.19c1.16-18.51,31.42-18.04,32.08.55v119.76c-1.34,25.11-32.65,27.51-48.09,13.24l-75.42-95.14c-2.82-2.58-7.03-2.45-7.42,1.87l.03,86.1c.01,21.09-31.94,21.37-32.63.55l.03-118.08c1.06-12.97,8.93-21.42,21.97-22.74Z'/%3e%3cpath%20class='st2'%20d='M12.13,1.28c28.23,1.02,58-1.81,86.06-.16,58.19,3.42,87.38,69.72,59.24,118.26-30.33,52.32-91.41,33.17-141.59,36.82-7.03-.55-12.94-3.9-15.01-10.94C-.84,102.42.61,59.25.1,16.3.7,8.64,4.37,3,12.13,1.28ZM39.75,31.63c-4.09.5-6.61,3.23-7.05,7.3,1.69,25.45-2.25,53.81-.02,78.92.43,4.85,2.61,7.02,7.46,7.45,31.53-2.7,72.07,10.71,90.43-23.37,16.64-30.88-3.09-68.26-38.47-70.47-17.01-1.06-35.22.72-52.35.17Z'/%3e%3cpath%20class='st0'%20d='M451.82.76c14.02-1.45,22.07,5.56,28.57,16.93,7.85,13.74,14.8,30.14,21.65,44.58,11.76,24.81,22.94,49.91,34.42,74.86,6.22,17.37-16.52,25.98-26.77,12.95l-49.9-105.74c-2.92-4.12-8.88-2.9-10.92,1.47l-48.61,103.72c-5.67,9.12-21.27,9.25-26.5-.49-3.21-5.96-1.09-10.33,1.15-16.01,8.28-20.95,19.9-43.62,29.61-64.21,7.23-15.34,14.36-31.83,22.28-46.71,5.84-10.97,11.36-19.93,25.03-21.34Z'/%3e%3cpath%20class='st1'%20d='M354.87,260.86c-1.34-1.43-.43-6.14-.7-8.14-8.6,12.04-28.47,12.98-39.48,3.85-12.17-10.1-12.56-35.25-1.89-46.55,10.21-10.81,31.8-10.33,40.81,1.87-.23-1.61.59-3.13.61-4.69.06-7.12-1.35-14.7-.53-21.99.44-3.92,1.31-5.93,5.76-5.34,5.36.72,2.99,8.02,3.07,11.28.47,19.6,1.49,41.18.49,60.74-.07,1.41-.64,8.36-1.05,8.88-1.68,1.61-5.54,1.72-7.09.09ZM329.73,210.39c-19.97,2.69-20.94,38.81-1.64,43.83,12.31,3.2,23.12-3.1,25.38-15.7,3-16.68-5.49-30.59-23.74-28.13Z'/%3e%3cpath%20class='st1'%20d='M414.37,235.07c-.12,17.69,19.96,25.23,33.88,16.46,2.61-1.64,7.32-7.85,10.24-2.89s-5.56,9.99-9.34,11.52c-29.14,11.79-53.82-15.18-40.78-42.94,10.64-22.66,48.71-19.04,51.78,7.15.27,2.28.92,10.7-2.43,10.7h-43.36ZM451.38,227.9c.55-21.16-29.25-24.48-35.92-5.26-.31.89-1.81,5.26-.25,5.26h36.18Z'/%3e%3cpath%20class='st1'%20d='M176.19,202.67c7.46-.84,18.53-.08,24.43,5.08,2.3,2.01,5.28,6.34,2.25,8.88-4.25,3.56-7.89-3.3-11.65-5.01-6.03-2.75-18.03-2.84-22.38,3.02-4.03,5.44.25,9.84,5.57,11.56,9.51,3.1,27.8,2.02,30.8,14.48,5.86,24.38-32.03,27.02-45.18,15.34-2.55-2.26-5.29-7.78-1.25-9.5,4.64-1.98,6.61,3.18,10.34,5.38,6.78,3.98,23.63,5.29,27.18-3.21,5.03-12.04-14.18-12.31-21.11-13.88-9.73-2.21-17.6-6.26-16.54-17.67.82-8.84,9.49-13.57,17.54-14.47Z'/%3e%3cpath%20class='st1'%20d='M63.23,260.86c-.65-.69-.64-1.96-.72-2.87-1.39-16.22,1.11-34.37,0-50.81-.05-2.53,1.07-3.91,3.62-4.14,6.44-.56,5.15,4.71,5.25,9.4l4.41-4.69c10.88-8.45,30.96-7.68,36.56,6.53,3,7.61,2.95,33.74,2.13,42.6-.64,6.91-7.92,6.29-8.74,2.62-.84-10.96.7-22.69-.1-33.56s-5.96-16.52-17.44-15.73c-24.4,1.7-14.9,32.77-16.94,48.73-.64,3.45-5.97,4.09-8.02,1.91Z'/%3e%3cpath%20class='st1'%20d='M514.35,213c1.29.33.93-.3,1.27-.7,1.69-1.97,2.63-3.88,4.86-5.6,3.68-2.83,17.08-7.73,17.13-.05.04,5.45-5.56,3.82-9.56,4.86-19.56,5.06-12.3,32.91-13.71,47.55-.72,3.82-7.91,4.17-8.3.04l.54-54.68c1.57-2.62,7.77-1.21,7.77.59v8Z'/%3e%3cpath%20class='st1'%20d='M6.59,203.19c2.04-.48,4.95-.53,5.82,1.68l.45,53.68c-.6,4.61-8.21,5.19-8.88-.57-1.84-15.83,1.41-35.13,0-51.33.06-1.73.89-3.05,2.61-3.46Z'/%3e%3cpath%20class='st1'%20d='M253.48,203.19c2.93-.68,6.16.05,6.25,3.46-1.4,16.21,1.84,35.49,0,51.33-.56,4.81-5.89,5.51-8.56,1.91l-.29-53.24c.06-1.73.89-3.05,2.61-3.46Z'/%3e%3cpath%20class='st1'%20d='M6.05,180.04c11.37-2.9,10.88,13.61,1.22,11.77-6.66-1.27-6.07-10.53-1.22-11.77Z'/%3e%3cpath%20class='st1'%20d='M252.93,180.02c11.46-2.59,10.5,13.27,1.14,11.91-6.22-.9-6.41-10.72-1.14-11.91Z'/%3e%3c/svg%3e)
%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23e60012;%20}%20.st1%20{%20fill:%20%23a1a1a1;%20}%20.st2%20{%20fill:%20%23f2f2f2;%20}%20%3c/style%3e%3c/defs%3e%3cpath%20class='st2'%20d='M208.79.2c10.12-1.03,18.93,1.81,26.29,8.72l77.13,96.19c2.89,2.23,6.06-.63,6.05-3.83l-.02-87.19c1.16-18.51,31.42-18.04,32.08.55v119.76c-1.34,25.11-32.65,27.51-48.09,13.24l-75.42-95.14c-2.82-2.58-7.03-2.45-7.42,1.87l.03,86.1c.01,21.09-31.94,21.37-32.63.55l.03-118.08c1.06-12.97,8.93-21.42,21.97-22.74Z'/%3e%3cpath%20class='st2'%20d='M12.13,1.28c28.23,1.02,58-1.81,86.06-.16,58.19,3.42,87.38,69.72,59.24,118.26-30.33,52.32-91.41,33.17-141.59,36.82-7.03-.55-12.94-3.9-15.01-10.94C-.84,102.42.61,59.25.1,16.3.7,8.64,4.37,3,12.13,1.28ZM39.75,31.63c-4.09.5-6.61,3.23-7.05,7.3,1.69,25.45-2.25,53.81-.02,78.92.43,4.85,2.61,7.02,7.46,7.45,31.53-2.7,72.07,10.71,90.43-23.37,16.64-30.88-3.09-68.26-38.47-70.47-17.01-1.06-35.22.72-52.35.17Z'/%3e%3cpath%20class='st0'%20d='M451.82.76c14.02-1.45,22.07,5.56,28.57,16.93,7.85,13.74,14.8,30.14,21.65,44.58,11.76,24.81,22.94,49.91,34.42,74.86,6.22,17.37-16.52,25.98-26.77,12.95l-49.9-105.74c-2.92-4.12-8.88-2.9-10.92,1.47l-48.61,103.72c-5.67,9.12-21.27,9.25-26.5-.49-3.21-5.96-1.09-10.33,1.15-16.01,8.28-20.95,19.9-43.62,29.61-64.21,7.23-15.34,14.36-31.83,22.28-46.71,5.84-10.97,11.36-19.93,25.03-21.34Z'/%3e%3cpath%20class='st1'%20d='M354.87,260.86c-1.34-1.43-.43-6.14-.7-8.14-8.6,12.04-28.47,12.98-39.48,3.85-12.17-10.1-12.56-35.25-1.89-46.55,10.21-10.81,31.8-10.33,40.81,1.87-.23-1.61.59-3.13.61-4.69.06-7.12-1.35-14.7-.53-21.99.44-3.92,1.31-5.93,5.76-5.34,5.36.72,2.99,8.02,3.07,11.28.47,19.6,1.49,41.18.49,60.74-.07,1.41-.64,8.36-1.05,8.88-1.68,1.61-5.54,1.72-7.09.09ZM329.73,210.39c-19.97,2.69-20.94,38.81-1.64,43.83,12.31,3.2,23.12-3.1,25.38-15.7,3-16.68-5.49-30.59-23.74-28.13Z'/%3e%3cpath%20class='st1'%20d='M414.37,235.07c-.12,17.69,19.96,25.23,33.88,16.46,2.61-1.64,7.32-7.85,10.24-2.89s-5.56,9.99-9.34,11.52c-29.14,11.79-53.82-15.18-40.78-42.94,10.64-22.66,48.71-19.04,51.78,7.15.27,2.28.92,10.7-2.43,10.7h-43.36ZM451.38,227.9c.55-21.16-29.25-24.48-35.92-5.26-.31.89-1.81,5.26-.25,5.26h36.18Z'/%3e%3cpath%20class='st1'%20d='M176.19,202.67c7.46-.84,18.53-.08,24.43,5.08,2.3,2.01,5.28,6.34,2.25,8.88-4.25,3.56-7.89-3.3-11.65-5.01-6.03-2.75-18.03-2.84-22.38,3.02-4.03,5.44.25,9.84,5.57,11.56,9.51,3.1,27.8,2.02,30.8,14.48,5.86,24.38-32.03,27.02-45.18,15.34-2.55-2.26-5.29-7.78-1.25-9.5,4.64-1.98,6.61,3.18,10.34,5.38,6.78,3.98,23.63,5.29,27.18-3.21,5.03-12.04-14.18-12.31-21.11-13.88-9.73-2.21-17.6-6.26-16.54-17.67.82-8.84,9.49-13.57,17.54-14.47Z'/%3e%3cpath%20class='st1'%20d='M63.23,260.86c-.65-.69-.64-1.96-.72-2.87-1.39-16.22,1.11-34.37,0-50.81-.05-2.53,1.07-3.91,3.62-4.14,6.44-.56,5.15,4.71,5.25,9.4l4.41-4.69c10.88-8.45,30.96-7.68,36.56,6.53,3,7.61,2.95,33.74,2.13,42.6-.64,6.91-7.92,6.29-8.74,2.62-.84-10.96.7-22.69-.1-33.56s-5.96-16.52-17.44-15.73c-24.4,1.7-14.9,32.77-16.94,48.73-.64,3.45-5.97,4.09-8.02,1.91Z'/%3e%3cpath%20class='st1'%20d='M514.35,213c1.29.33.93-.3,1.27-.7,1.69-1.97,2.63-3.88,4.86-5.6,3.68-2.83,17.08-7.73,17.13-.05.04,5.45-5.56,3.82-9.56,4.86-19.56,5.06-12.3,32.91-13.71,47.55-.72,3.82-7.91,4.17-8.3.04l.54-54.68c1.57-2.62,7.77-1.21,7.77.59v8Z'/%3e%3cpath%20class='st1'%20d='M6.59,203.19c2.04-.48,4.95-.53,5.82,1.68l.45,53.68c-.6,4.61-8.21,5.19-8.88-.57-1.84-15.83,1.41-35.13,0-51.33.06-1.73.89-3.05,2.61-3.46Z'/%3e%3cpath%20class='st1'%20d='M253.48,203.19c2.93-.68,6.16.05,6.25,3.46-1.4,16.21,1.84,35.49,0,51.33-.56,4.81-5.89,5.51-8.56,1.91l-.29-53.24c.06-1.73.89-3.05,2.61-3.46Z'/%3e%3cpath%20class='st1'%20d='M6.05,180.04c11.37-2.9,10.88,13.61,1.22,11.77-6.66-1.27-6.07-10.53-1.22-11.77Z'/%3e%3cpath%20class='st1'%20d='M252.93,180.02c11.46-2.59,10.5,13.27,1.14,11.91-6.22-.9-6.41-10.72-1.14-11.91Z'/%3e%3c/svg%3e)
AWS 示範以 OpenClaw 與 AgentCore 打造可累積對話記憶的助理
AWS 於 2026 年 10 月 6 日發布技術指南，示範如何以 OpenClaw 和 Amazon Bedrock AgentCore 建立能保留結構化對話資訊的助理。範例以園藝助理 Sprout 展示記憶設計、訊息入口與模型分流。
完整報導
AWS 於 2026 年 10 月 6 日發布一份技術指南，說明如何將 OpenClaw 部署在 Amazon Bedrock AgentCore runtime，並以 AgentCore memory 將對話整理成可透過 metadata filters 擷取的結構化知識。這套設計的示範助理名為 Sprout，聚焦園藝；AWS 表示，架構也可藉由更換 persona 與 skills manifest，改作支援機器人、健身教練或內部服務台。
指南中的系統以單一 AWS CloudFormation 範本部署。使用者可透過 Telegram 傳送訊息，或由排程工作觸發；兩種入口都會呼叫同一個 AgentCore runtime agent。Telegram 訊息經由 Amazon API Gateway 和 webhook Lambda 傳入，排程則由 Amazon EventBridge Scheduler 與 cronjob Lambda 處理。進入 runtime 後，server.py 負責協調 OpenClaw gateway、AgentCore memory 和 Amazon Bedrock Converse API。
範例也將文字與影像請求分開處理：文字對話使用 Claude Haiku 4.5，影像理解使用 Claude Sonnet 4.5。文字回合經 OpenClaw gateway 傳送；影像回合則由 server.py 直接呼叫 Bedrock，並傳入影像內容。AWS 在 2025 年已介紹 AgentCore Memory 作為代理記憶管理服務；這篇指南進一步展示它如何融入一套可部署的助理架構。
為什麼值得注意
這篇指南把持續性記憶放進一個具體部署範例，並說明訊息、排程、記憶與文字／影像模型如何連接。對評估個人化助理的開發者而言，重點不只是記住對話，也包括如何沿用同一套架構切換助理用途。
來源證據
先看主要證據，再查看佐證來源與來源脈絡。
主要證據
1Building a context-aware AI assistant on AgentCore and OpenClaw
Thiago Verney
“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.”
推薦消息
根據品牌、主題與近期相關消息推薦