Hi Synapticians!

As you might already know, we're pretty big fans of ChatGPT around here, let's be honest, they've simply built the best AI product on the market! So when OpenAI drops new features, we perk up and listen.

Their latest updates are pretty exciting. Remember DeepSearch? The quality was mind-blowing, but sharing results was a real pain, until now! You can finally export those brilliant search insights as PDFs. Finally!

Speaking of updates, OpenAI just announced SharePoint integration (yes, you heard that right). This is interesting timing, especially with all the whispers about tensions between OpenAI and Microsoft. Makes you wonder what this means for Copilot, doesn't it?

Here's the thing: Microsoft's secret weapon has always been putting their AI directly into the tools you already use every day. That's what makes it so practical for businesses : no app-switching, no extra steps, just AI where you need it.

Happy reading!

Top AI news

1. OpenAI's PDF Export: Transforming AI Research for Enterprises
OpenAI introduced a PDF export feature for ChatGPT's Deep Research, allowing users to download well-formatted reports, including tables and clickable citations. This update targets professional sectors like finance and consulting, where verifiable documents are crucial. It marks a strategic focus on enterprise customers, addressing a practical pain point in sharing AI-generated insights. The move is part of the competitive landscape in AI research tools, where integration and speed are vital for adoption.

2. ChatGPT integrates with SharePoint to analyze company data
ChatGPT's new beta feature allows users to access company data through a SharePoint connector, enabling them to analyze and summarize content from multiple SharePoint sites with source references. This feature is available for ChatGPT Plus, Pro, and Team users and ensures data privacy by only accessing content users have permission to view. OpenAI plans to extend support to Enterprise customers soon.

3. Guardian agents aim to reduce AI hallucinations below 1%
Vectara introduces guardian agents to tackle AI hallucinations by not just detecting but correcting them, achieving error rates below 1% for smaller models. These agents take precise actions to maintain content integrity, especially in contexts where deviations might be intentional, like creative fields. The approach, supported by the HCMBench evaluation toolkit, offers enterprises a middle path to leverage AI while ensuring accuracy, thus expanding potential AI use cases.

Bonus. AI's Free Will: Importance of a Moral Compass
The article by Frank Martela examines whether AI meets philosophical conditions for free will. Martela argues that AI exhibits goal-directed agency, genuine choice, and control, making it ethically imperative to equip AI with a moral compass. This shift in responsibility from developers to AI agents highlights the need for deep ethical training among AI developers, especially as AI approaches complex moral dilemmas. Recent events, like the withdrawal of ChatGPT's update, underscore the urgency of addressing these ethical challenges.

Meme of the Day

Theme of the Week

Small language model - Hugging face SLM overview

Need a primer on SLMs? This blog walks through the spectrum from 1 M to 10 B parameters in plain English. It compares token budgets, inference latency, and energy draw against GPT-style behemoths. Tables of open-source checkpoints show which tasks each model family nails today. Perfect briefing-slide material for anyone new to the “smaller is better” movement.

Stay Connected

Feel free to contact us with any feedback or suggestions—we’d love to hear from you !