Ciprian Nastase
@CipAIExplorer
๐๐ฅAI Innovator in #CustomerSupport ๐ค | Driving 80% ticket resolution via #AI | Sharing insights on #AIForSupport & tech transformations.
Next-level RAG๐คฏ: - multiple agents - web searches - LLM per-user memory - basic RAG I envision this integrated in AI support, with the LLM being able to provide not only responses backed up by real data sources, but also fine-tuned to each customer. Awesome demo @ashpreetbedi !
Introducing `Personalized Agentic RAG` - where the LLM remembers key details about the user and automatically chooses the tool for RAG. We'll build: ๐ ChatGPT like memory โ๏ธ Function calling ๐ง Multi-Agent orchestration code: git.new/personalization
Heck you can even throw your meta-commentary into the middle of a prompt in a parenthetical.
The most valuable skill in tech? Knowing when to stop coding and start thinking.
2023 - chatbots are here to replace us 2024 - RAG and fine-tuning will replace chatbots 2025 MCP and tool use - the undisputed future of work 2026 - computer environment and filesystem will revolutionise how we use AI
Composer 1 is crazy fast!
Introducing Cursor 2.0. Our first coding model and the best way to code with agents.
Did anyone else experience massive headaches today with both GPT and Claude, across different versions? I mean, inconsistent responses all across the board, with some of them plain idiotic
One big effect of AI agents is that they take previously scarce resources and skills and make them abundant and available to everyone. As a consequence, you can โself-serveโ many more of the hard problems that you used to have to go to others to solve. The product marketer canโฆ
An underdiscussed phenomena that is starting to happen in organizations: AI-driven role conflict. The roles in charge of design, product management, coding, marketing, etc. in a project used to have relatively clear lines. AI lets entrepreneurial workers extend into other roles.
Quick wins with AI feel great, until they stop scaling. One-off automations and prompt hacks wonโt cut it when AI agents start to outnumber humans. The shift? Move from experiments to tools. Not just chatbots or side projects, but real systems built into how teams work: โ โฆ
Google just consolidated 3,000 AI courses from Cloud, DeepMind, and their other divisions into one platform: Google Skills. The internet's drowning in "Get Rich Quick with AI" courses from people who got rich selling courses, not using AI. Meanwhile, actual resources from theโฆ
DeepSeek's OCR has been open sourced. It isn't just reading text better, it's compressing visual information into language tokens so efficiently that context windows might stop being the bottleneck we've been designing around for years.ย huggingface.co/deepseek-ai/Deโฆ
First Perplexity Comet, now ChatGPT Atlas. Waiting to see @AnthropicAI 's browser, and most importantly, waiting for Google to drop something (although that will probably be disappointing)
Meet our new browserโChatGPT Atlas. Available today on macOS: chatgpt.com/atlas
Meet our new browserโChatGPT Atlas. Available today on macOS: chatgpt.com/atlas
Has anyone come up with a disaster recovery backup from AWS to another hyper scaler thereโs gotta be a market for that.
Fascinating to see that so much of the internet relies on a single DNS record dynamodb.us-east-1.amazonaws.com which went missing today, causing widespread outages.
That wonderful moment when AWS has an outage and half of the internet is down ๐ซ
If you're banning AI chatbots from Whatsapp, why not go the full 9 yards and block Meta AI as well? Just saying... techcrunch.com/2025/10/18/whaโฆ
ChatGPT launched less than three years ago. When I talked about AI in mid-2024 for audiences ranging from professors to Silicon Valley leaders, only 10% of people had tried AI themselves. I am actually very surprised that AI adoption has been as fast as it has in enterprises.
In my most recent talks to companies, even though everyone is talking about AI, less than 10% of people have even tried GPT-4, and less than 2% have spent the required 10 or so hours with a frontier model. Twitter gives you a misleading impression, things are still very early.
It's interesting how people fixate on the mechanical aspects of coding while missing the metacognitive skills it develops. The ability to decompose complex problems, anticipate failure modes, and design robust solutions is what truly matters. These are timeless skills thatโฆ
The new Sora Pro feature that builds storyboards and executes them is really interesting. Here is the prompt โan ad for the abstract concept of the feeling you get after falling asleep on your armโ Notice the high character consistency, multiple shots, composition. All the AI.
AI is everywhere at work. Daily usage has doubled in the last year, and workers say it makes them 33% more productive. But... 96% of companies have NOT seen dramatic improvements in efficiency, innovation, or quality. Why? A new report from @Atlassian digs in.
We built AI directly into Twitter.๐ Connect Twitter โ It learns your voice โ Create personalised posts directly in X using AI No platform switching. No configuration. No learning curve. PrismX: prismx.ti.trilogy.com
United States Trends
- 1. Good Monday 27.4K posts
- 2. #RomanEmpireByBangChan 15.8K posts
- 3. ROMAN EMPIRE OUT NOW 13.6K posts
- 4. Jayden Daniels 25.6K posts
- 5. #MondayMotivation 26.5K posts
- 6. jungkook 613K posts
- 7. #STARDOM 5,311 posts
- 8. Dan Quinn 7,164 posts
- 9. $KITE 4,206 posts
- 10. Perle Labs 5,312 posts
- 11. #River 7,328 posts
- 12. Seahawks 39.1K posts
- 13. 60 Minutes 80.7K posts
- 14. Jake LaRavia 6,468 posts
- 15. Sam Darnold 15.3K posts
- 16. Commanders 51K posts
- 17. Godzilla 46.7K posts
- 18. #RaiseHail 8,907 posts
- 19. Bronny 15.9K posts
- 20. Snopes 3,486 posts
Something went wrong.
Something went wrong.