- Популярные видео
- Авто
- Видео-блоги
- ДТП, аварии
- Для маленьких
- Еда, напитки
- Животные
- Закон и право
- Знаменитости
- Игры
- Искусство
- Комедии
- Красота, мода
- Кулинария, рецепты
- Люди
- Мото
- Музыка
- Мультфильмы
- Наука, технологии
- Новости
- Образование
- Политика
- Праздники
- Приколы
- Природа
- Происшествия
- Путешествия
- Развлечения
- Ржач
- Семья
- Сериалы
- Спорт
- Стиль жизни
- ТВ передачи
- Танцы
- Технологии
- Товары
- Ужасы
- Фильмы
- Шоу-бизнес
- Юмор
How AI Agents Remember (AgentCore Memory + Security Explained)
🚀 How do AI agents remember users, conversations, and preferences — safely?
In this video, we break down **Amazon Bedrock AgentCore Memory** — including how to create memory, enable long-term strategies, and secure it against real-world risks like prompt injection and memory poisoning.
---
💡 What you’ll learn:
• How to create AgentCore Memory using CLI and SDK :contentReference[oaicite:0]{index=0}
• Short-term vs long-term memory in AI agents
• Memory strategies (summarization, user preference extraction)
• Event retention and lifecycle configuration
• How memory is stored and retrieved
• Encryption using AWS KMS (customer-managed vs AWS-managed keys)
• Prompt injection risks and how to mitigate them
• Memory poisoning attacks and prevention strategies
• Shared responsibility model for AI security
---
🧠 Key Insight:
AI agents don’t just respond — they remember.
But memory introduces risks:
• Wrong data persistence
• Malicious prompt injection
• Long-term incorrect behavior
👉 Memory must be designed securely.
---
📌 Core Concepts:
🔹 Short-Term Memory
Stores raw events (conversations, interactions)
🔹 Long-Term Memory
Extracts meaningful insights (preferences, summaries)
🔹 Memory Strategies
Define how raw data becomes structured knowledge
🔹 Encryption
Secure memory using KMS keys
---
⚡ Why this matters:
Without memory:
→ Agents feel generic ❌
Without security:
→ Agents become dangerous ❌
With proper design:
→ Personalized + safe AI systems ✅
---
🏗️ Real-world use cases:
• Personalized AI assistants
• Customer support agents
• AI copilots with user preferences
• Enterprise AI platforms
• Multi-session conversational systems
---
🔗 Topics covered:
AgentCore Memory, AI memory systems, prompt injection, memory poisoning, AI security, GenAI architecture
---
#AWS #AmazonBedrock #AgentCore #AIMemory #GenAI #LLM #AIEngineering #AgenticAI #AISecurity
Видео How AI Agents Remember (AgentCore Memory + Security Explained) канала Pushkar Mishra
In this video, we break down **Amazon Bedrock AgentCore Memory** — including how to create memory, enable long-term strategies, and secure it against real-world risks like prompt injection and memory poisoning.
---
💡 What you’ll learn:
• How to create AgentCore Memory using CLI and SDK :contentReference[oaicite:0]{index=0}
• Short-term vs long-term memory in AI agents
• Memory strategies (summarization, user preference extraction)
• Event retention and lifecycle configuration
• How memory is stored and retrieved
• Encryption using AWS KMS (customer-managed vs AWS-managed keys)
• Prompt injection risks and how to mitigate them
• Memory poisoning attacks and prevention strategies
• Shared responsibility model for AI security
---
🧠 Key Insight:
AI agents don’t just respond — they remember.
But memory introduces risks:
• Wrong data persistence
• Malicious prompt injection
• Long-term incorrect behavior
👉 Memory must be designed securely.
---
📌 Core Concepts:
🔹 Short-Term Memory
Stores raw events (conversations, interactions)
🔹 Long-Term Memory
Extracts meaningful insights (preferences, summaries)
🔹 Memory Strategies
Define how raw data becomes structured knowledge
🔹 Encryption
Secure memory using KMS keys
---
⚡ Why this matters:
Without memory:
→ Agents feel generic ❌
Without security:
→ Agents become dangerous ❌
With proper design:
→ Personalized + safe AI systems ✅
---
🏗️ Real-world use cases:
• Personalized AI assistants
• Customer support agents
• AI copilots with user preferences
• Enterprise AI platforms
• Multi-session conversational systems
---
🔗 Topics covered:
AgentCore Memory, AI memory systems, prompt injection, memory poisoning, AI security, GenAI architecture
---
#AWS #AmazonBedrock #AgentCore #AIMemory #GenAI #LLM #AIEngineering #AgenticAI #AISecurity
Видео How AI Agents Remember (AgentCore Memory + Security Explained) канала Pushkar Mishra
Комментарии отсутствуют
Информация о видео
26 апреля 2026 г. 11:45:06
00:05:15
Другие видео канала





















