Claudius Gramse

Joined the @perplexity_ai Business Fellowship! 🚀

Direct access to their mind-blowing AI research tools = better insights for you. Deep Research is a game-changer.

#AI #DeepResearch

KI-Personalisierung: Es geht nicht um Manipulation, sondern um *Ermächtigung* 💪.

Gib deinen Kunden die Informationen und Werkzeuge, die sie *wirklich* brauchen.
Wie setzt du das um?

#KI #CustomerEmpowerment #Mehrwert 🛠️🔍

🧵 6/6
KEY TAKEAWAY:
Innovation needs both freedom and clear boundaries. Let's shape the future through open dialogue. It's not about restrictions, but smart guardrails that balance creativity and responsibility.
#AI #Innovation #Ethics #DigitalTransformation

🧵 5/6
ETHICAL CONSIDERATIONS:
While opportunities are vast, we can't ignore the risks 7. Autonomy means responsibility. We need clear guidelines to prevent misuse and ensure AI aligns with our values.

🧵 4/6
RESEARCH & DEVELOPMENT:
Accelerated innovation through autonomous experimentation and analysis. AI can discover new drugs, improve climate models, and optimize algorithms - saving time and resources.

🧵 3/6
CUSTOMER EXPERIENCE:
Real-time personalized customer service. AI adapts to individual needs, from intelligent chatbots to tailored marketing campaigns, significantly improving customer satisfaction.

🧵 2/6
EFFICIENCY:
Self-replicating AI can process tasks in parallel, reduce costs, and optimize resources. Think automated processes running 24/7 across manufacturing, healthcare, and financial services.

Solitary figure contemplating the future of self-replicating AI, set against a minimalist turquoise landscape - # nextGenAI by Evonomics

🧵 1/6
Meta and Alibaba have demonstrated it: AI can self-replicate. Models like Llama3.1 and Qwen2.5 showcase this capability 2. But what does this mean? Let's cut through the hype and look at the facts:

AI Personalization: It's not about being creepy, it's about being relevant 🤖

It's about empowering your customers with the information and tools they need 💪

Learn about the 5 Pillars: zurl.co/vLA12 #AI #Personalization #CX 🌟

"Personalisierung ist wichtig." - Jeder Marketing-Manager, überall.

Aber was bedeutet das *wirklich*? Geht es nur darum, deinen Vornamen in eine E-Mail zu schreiben? 🤔

KI kann mehr. Viel mehr. #KI #Personalisierung #Marketing

Generic marketing is OVER. 👋

Customers expect *personalized* experiences. AI makes it possible, at scale.

I've written a deep dive on the 5 pillars of AI-driven personalization. Check it out! zurl.co/mqaxk

#AI #Personalization #Marketing

Hibiki Samples - a Hugging Face Space by kyutai Discover amazing ML apps made by the community

Hibiki: Real-time, offline, open-source speech translation.

Will AI replace human translators? Especially within the EU, with its massive translation needs? 🤔

Let's get real. #AI #Translation #Disruption #Jobs #EU zurl.co/VwFHa

Remember when AI needed constant human guidance? Those days are ending.

Agentic RAG systems are learning to think for themselves. They refine, adapt, improve - autonomously.

Scary? Exciting? Both? Let's discuss. #AI #RAG #Future

Babel numérique | Le code a changé | ARTE Disponible jusqu'au 14/03/27Si, il y a quelques années, on se moquait des traductions fournies par Google Translate, les progrès de l'intelligence artificiel...

Real-time, offline speech translation on an iPhone? 🤯

The open-source Hibiki project is making it happen (French -> English for now).

Game-changer for international business?
zurl.co/6O4Zu

#AI #OpenSource #Translation #Hibiki

The future of RAG is here:

📌 HTML structure preservation
📌 Multimodal capabilities
📌 Autonomous refinement

Which trend will have the biggest impact?

Read the full analysis: zurl.co/YEBtq #AI #RAG #Future

Agentic RAG: The next evolution in AI systems

• Autonomous query refinement
• Self-optimizing results
• Intelligent adaptation

Full analysis in my latest article! zurl.co/EfJb5
#AI #RAG #Future

Real-world applications of modern RAG:

🔹 Document processing
🔹 Visual data integration
🔹 Autonomous refinement

Which excites you most?

Full breakdown in the article! zurl.co/YJXro
#AI #RAG #Innovation

Multimodal RAG systems are changing the game!

Combining text + visual data leads to:
• Better understanding
• More accurate responses
• Broader applications

Full analysis in my latest article 🔍 zurl.co/cebcG
#AI #RAG #Innovation

Deep diving into HtmlRAG today! It's fascinating how preserving HTML structure can make such a difference in data retrieval.

Check out my analysis of why structure matters in modern RAG systems. zurl.co/THpla
#AI #RAG #TechTalk

All the AI Builders, I just found out!

There are 16 of them!

1. Cursor
2. Bolt
3. v0
4. Windsurf
5. Replit Agent
6. Loveable
7. Devin
8. Pear AI
9. Github Copilot
10. Github Spark
11. IDX by Google
12. webdraw .ai
13. UIzard
14. Amazon Q Developer
15. ChatGPT inside Mac app
16. Softgen AI

Modern Types of RAG systems: From HTML to Agentic Systems

What's the deal with traditional RAG systems? I've just published an article to explain it all!
Let me know what you think about the limitations of standard RAG systems, and if you have experienced them yourself! [zurl.co/oNvU7]

#RAG #KI #InformationRetrieval

📲 Open Source Social Media Agent

For the past three weeks, we've been using an ambient agent to draft and schedule social media posts highlighting community content

Think of it as an "AI Social media intern"

Today we're releasing it as an open source repo 🧵

The GPT-4 barrier was comprehensively broken Some of those GPT-4 models run on my laptop LLM prices crashed, thanks to competition and increased efficiency Multimodal vision is common, audio and video are starting to emerge Voice and live camera mode are science fiction come to life Prompt driven app generation is a commodity already Universal access to the best models lasted for just a few short months “Agents” still haven’t really happened yet Evals really matter Apple Intelligence is bad, Apple’s MLX library is excellent The rise of inference-scaling “reasoning” models Was the best currently available LLM trained in China for less than $6m? The environmental impact got better The environmental impact got much, much worse The year of slop Synthetic training data works great LLMs somehow got even harder to use Knowledge is incredibly unevenly distributed LLMs need better criticism Everything tagged “llms” on my blog in 2024

Here's my end-of-year review of things we learned out about LLMs in 2024 - we learned a LOT of things simonwillison.net/2024/Dec/31/...

Table of contents: