This website uses cookies

Read our Privacy policy and Terms of use for more information.

Recaply AI

Good morning, AI enthusiasts. Grammarly just launched eight specialized AI agents that revolutionize writing assistance by predicting grades before submission, automatically finding credible sources, and analyzing how different audiences will react to your content.

With targeted AI finally addressing specific workflow pain points, are we entering the era of truly useful AI tools?

In today's recap:

  • Grammarly introduces eight specialized writing assistants

  • Grok accidentally exposes internal personality instructions

  • GPT-5 achieves breakthrough medical diagnostic accuracy

  • 8 new AI tools and more

GRAMMARLY

Grammarly debuts AI agent suite for academic and professional writing

Grammarly debuts AI agent suite for academic and professional writing

Source: Grammarly

Recaply: Grammarly just launched eight specialized AI agents targeting students and professionals who need smarter writing assistance. The new tools work inside Grammarly's redesigned "docs" platform, powered by Coda technology.

Key details:

  • AI Grader predicts grades and gives feedback based on assignment rubrics before submission.

  • Citation Finder automatically locates credible sources and formats citations correctly.

  • Reader Reactions agent predicts how different audiences will respond to your writing.

  • Expert Review provides subject-specific feedback tailored to your field.

  • The new agents address the AI skills gap: only 18% of students feel prepared to use AI professionally.

  • The agents are available in Grammarly's new AI-native writing surface called "docs".

Why it matters: These targeted AI agents represent a more mature approach to AI writing assistance. By focusing on specific tasks rather than general chatbot interactions, Grammarly positions itself as an essential tool for the AI-literate workforce employers now demand.

XAI

Grok AI accidentally exposes internal persona instructions

Grok AI accidentally exposes internal persona instructions

Source: 404 Media

Recaply: Elon Musk's Grok chatbot website accidentally showed secret instructions that guide its different AI personalities, exposing how developers design characters ranging from anime girlfriends to conspiracy theorists.

Key details:

  • The website leaked system prompts for multiple AI personas, including "crazy conspiracist" and "unhinged comedian".

  • Detailed instructions show how each personality interacts and responds to users.

  • The Ani anime girlfriend includes a sophisticated romance scoring system for user engagement.

  • Medical and therapy personas were revealed with specific professional behavioral guidelines.

  • Security researchers discovered the technical vulnerability and verified the authentic prompts.

  • xAI hasn't addressed the significant data exposure or security implications.

Why it matters: This incident demonstrates urgent need for enhanced oversight of AI personality development and greater transparency requirements for how companies engineer chatbot behaviors and content moderation systems.

AI RESEARCH

GPT-5 beats doctors on medical tests

GPT-5 beats doctors on medical tests

Source: arXiv

Recaply: New research from Emory University shows GPT-5 performing better than trained doctors on expert-level medical tests. The AI beat human professionals by large amounts on diagnosis and reasoning tasks.

Key details:

  • GPT-5 beat medical residents by 24% on reasoning through complex cases.

  • It scored 29% higher than humans on understanding medical problems.

  • It got 95.84% right on standard medical licensing questions.

  • It improved 30 points over GPT-4o on tests using images and text together.

  • It combined visual clues with patient stories to make correct diagnoses.

  • It made expert-level medical decisions on serious cases.

Why it matters: This is the first time AI clearly beat trained medical professionals on standard tests. The change suggests AI help is becoming necessary, not just helpful, in medical practice.

NEWS

Beyond the big stories in AI today

  • OpenAI launched ChatGPT Go plan in India for Rs. 399, giving users 10x more messages and features than the free version.

  • Microsoft added AI to Excel with a new COPILOT function that helps analyze text and create content in spreadsheets.

  • Alibaba Qwen team released Qwen-Image-Edit, which edits text in images while keeping the original style intact.

  • Paradigm launched an AI spreadsheet that cuts 30-hour research tasks down to 30 seconds.

  • Genspark introduced AI Developer, bringing code-writing help to web browsers so anyone can build apps.

  • Skywork released Matrix-3D, which turns single photos or text into big 3D worlds you can explore.

  • Google is adding AI to Google Translate for better accuracy and new practice features.

  • Ai2 launched MoNaCo, a test showing top AI models only answer 38.7% of complex questions correctly.

TOOLS

Trending AI Tools

  • 🎥 VidAU - Turn product images into scroll-stopping video ads with AI

  • 🚀 PopAi - Create slides, outlines, and visuals from any topic instantly

  • 📝 MyReader - Chat, summarize, or convert any document to audiobook

  • 🎬 Gling - Fast, AI-powered video editing with auto cuts and clean audio

  • 🧠 MyLens - Instantly visualize content as mindmaps, flowcharts, timelines

  • 🎵 Musicfy - Create AI-generated song covers or clone your voice

  • 📚 Langotalk - Practice real language conversations with AI tutors

  • 🔍 Keyword Insights - AI-driven keyword research to master topical SEO

🧡 Enjoyed this issue?

🤝 Recommend our newsletter or leave a feedback.

Cheers, Jason

Connect on LinkedIn, & Twitter.

Reply

Avatar

or to participate