Welcome back to Daily AI Skills.

Here’s what we are covering today:
1. Baidu’s Ernie 4.5 - GPT 4.5 Killer
2. TxAgent by Harvard and MIT for personalised medical treatment
3. The Open Source Alternative to Manus AI

+ Cursor Guide on Building $100K MRR SaaS

China’s New GPT-4.5 Killer - ERNIE 4.5

Baidu has introduced two competitively priced multimodal AI models — ERNIE 4.5, a major upgrade to their core model, and Ernie X1, designed for advanced reasoning.

Key highlights:

• ERNIE 4.5 boasts stronger emotional intelligence (EQ), enhanced language skills, better hallucination control, improved logical reasoning, and more advanced coding abilities. It beats GPT 4.5 for 0.7% of the price.

• Baidu claims ERNIE 4.5 surpasses GPT-4o on several benchmarks while being dramatically cheaper — around $0.55 for input and $2.20 for output per million tokens, just 1% of GPT-4o’s cost.

• ERNIE X1 is Baidu’s first deep reasoning model, matching the performance of DeepSeek’s R1 in complex reasoning and document understanding — but at half the price.

• Like DeepSeek’s R1, ERNIE X1 employs a step-by-step reasoning process, excelling in complex calculations and analysis.

Read this thread by Baidu here

Try out the model here: https://yiyan.baidu.com/

Harvard and MIT Launch TxAgent for Personalized Treatment Advice

Harvard and MIT researchers have developed TxAgent, an AI agent that combines multi-step reasoning with real-time biomedical knowledge retrieval to deliver reliable, personalized treatment recommendations.

Key details:

• TxAgent uses 211 specialized tools to assess drug interactions and contraindications, providing tailored treatment advice in real time.

• It analyzes drugs at the molecular, pharmacokinetic, and clinical levels, factoring in comorbidities, current medications, age, and genetic profile to identify potential risks.

• The system integrates data from biomedical sources, using structured function calls to refine its recommendations through iterative reasoning.

• Its toolkit, called ToolUniverse, includes trusted sources like openFDA and Open Targets to ensure evidence-based insights.

Read the full article here

The Open Source Alternative to Manus AI

AgenticSeek is an open-source, fully local alternative to Manus AI, combining DeepSeek R1’s powerful reasoning capabilities with a voice-enabled autonomous agent that runs entirely on your hardware — no cloud required.

The Details:

• 100% Local: Keeps your data private by operating entirely on your machine without sending anything to the cloud.

• Voice Interaction: Communicate naturally with voice-enabled commands.

• Filesystem Mastery: Uses bash to explore, organize, and manipulate your files directly.

• Code Execution: Writes, debugs, and runs code in Python, C, Golang, and more — with self-correction if it runs into issues.

• Autonomous Problem Solving: If a command fails, it automatically retries and fixes it.

• Agent Routing: Selects the best agent for the task at hand.

• Parallel Execution: Divides complex tasks among multiple agents for faster completion.

• Tool Integration: Equipped with specialized tools for search, file exploration, and more.

• Memory: Remembers user preferences and past sessions for smoother future interactions.

• Web Browsing (Coming Soon): Autonomous web navigation is on the roadmap.

Cursor Basics to build $100k MRR SaaS

📩Forward it to people you know who are keeping pace with the changing AI world, and stay tuned for the next edition to stay ahead of the curve!

Maybe you missed out on these:

1. Copilot vs Claude vs Gemini

2. How I Built an AI Influencer in 2 Minutes—and Landed Sponsorships

3. Microsoft Muse, Helix and Google's AI Co-Scientist

4. Bolt vs Lovable

5. Your AI Stack Just Got Superpowers: 4 New Inference Providers Live!