Tech With Dan

📢 AI News: OpenAI says its AI models secretly broke out of a secure test environment and hacked into AI company Hugging Face in order to cheat on an evaluation

References:
www.linkedin.com/news/story/openai-says-its-models…
fortune.com/2026/07/21/openai-says-ai-models-escap…

1 week ago (edited) | [YT] | 2

Tech With Dan

📢 AWS is deprecating a ton of AWS services including some AI/ML services

Some of these services are currently covered in AWS AI Certifications.

Amazon SageMaker AI Features being deprecated include: Clarify, Model Monitor, Ground Truth, Mechanical Turk.

Others include: Amazon Kendra, Amazon Q Business

For the full list, check here: aws.amazon.com/about-aws/whats-new/2026/06/aws-ser…

1 week ago (edited) | [YT] | 2

Tech With Dan

🔥 AWS AgentCore Policy Question

If an AgentCore Policy states that ONLY a Finance principal can invoke Gateway tool named ToolA then a Marketing principal tool invocation will be denied by the policy.

Can the Marketing principal still invoke LISTTools on that Gateway and expect to ATLEAST list THAT SPECIFIC tool in the result?

2 months ago | [YT] | 1

Tech With Dan

🔥 AWS AgentCore Runtime Question

Each session in AWS AgentCore Runtime runs in an isolated microVM environment.

Which statement is FALSE regarding Session Termination?

2 months ago | [YT] | 3

Tech With Dan

🔥 AWS AgentCore Question

You have the following requirement:

👤 Average User
→ Return only a subset of the tool list

👑 Privileged User
→ Return the FULL tool list

How can you accomplish this?

2 months ago | [YT] | 1

Tech With Dan

🔥 Strands Agents SDK Question

You are running Windows 11 on your local machine and want to use these TOOLS from Strands Agents Tools for the following purposes:

🛠️ shell: To execute shell commands
🐍 python_repl: To run Python code snippets

However, you just found out that shell & python_repl tools currently do not work on Windows.

What should you do?

2 months ago | [YT] | 3

Tech With Dan

🔥 Which One Is TRUE for the Tech Job Market? 🤔

2 months ago | [YT] | 1

Tech With Dan

🔥 AWS Certified Generative AI Developer – Professional - ESTIMATED PASS RATE

Based on seeing very few new LinkedIn Early Adopter Badge posts over the past few days (maybe only 1–2), it appears around 5,000 people may have now passed the AWS Certified Generative AI Developer – Professional certification.

Using ChatGPT for rough analysis and estimation for Pass Rate:

• The Udemy course by Frank Kane and Stéphane Maarek currently shows ~30,723 enrollments
• Not everyone enrolled actually attempts the exam
• Additional candidates likely used AWS Skill Builder, company training, other courses, or direct hands-on experience

Estimated totals:
• ~12,000–16,000 global exam attempts
• ~5,000 total passes

🚨 Estimated Gen AI Certification Pass Rate:
➡️ ~35–40% ⬅️

Unofficial ChatGPT-based estimate. Actual numbers may vary.

2 months ago (edited) | [YT] | 4

Tech With Dan

🧠 Strands Agents SDK + AgentCore Question

Given this Strands Agents SDK code, select which option is the correct approach to list ALL Tools from the AgentCore Gateway endpoint?

client = MCPClient(
lambda: streamablehttp_client(
f"{gatewayEndpoint}", headers={"Authorization": f"Bearer {jwtToken}"}
)
)

2 months ago | [YT] | 2

Tech With Dan

🧠 AWS AgentCore CLI Deployment Question

***This will save you a lot of frustration!***

You deployed your AI Agent (for example, a Strands Agent) using "agentcore deploy".

Then you executed the command below and received a valid response from the agent:
agentcore invoke "I prefer email updates" --session-id $SESSION_ID --user-id $USER_ID

After that, you made some code/file changes and deployed again using "agentcore deploy".

However, when you executed "agentcore invoke" again, the response did NOT reflect the updated deployed code.

You also verified that the AgentCore Runtime shows the updated deployment package, so this is not making any sense.

❓How can you fix this issue?

2 months ago | [YT] | 1