xAI’s Grok Adds Memory to Rival ChatGPT and Gemini

✨TodayOnAI’s Daily Drop

• xAI’s Grok Adds Memory to Rival ChatGPT and Gemini

• Deck Raises $12M to Unlock the Web's Hidden Data

• OpenAI Launches Codex CLI, an Open-Source AI Agent for Your Terminal

• OpenAI’s New o3 Model Can Reason with Sketches, Diagrams, and More

• 💬 Let’s Fix This Prompt

• 🧰 Today’s AI Toolbox Pick

xAI

xAI’s Grok Adds Memory to Rival ChatGPT and Gemini

🚀 TodayOnAI Insight: xAI has introduced a memory feature to its Grok chatbot, marking a key step toward matching the personalized capabilities of ChatGPT and Gemini. The update allows Grok to recall past interactions, tailoring responses based on user preferences.

🔍 Key Takeaways:

• Grok now uses memory to deliver more personalized answers by remembering previous conversations.

• Users can view, manage, or delete specific memories directly from the chat interface.

• The feature is in beta on Grok.com and the mobile apps, but is not available in the EU or U.K.

• Memory can be toggled off in Data Controls, offering transparency and user control.

• xAI plans to expand the feature to Grok’s integration on X (formerly Twitter).

💡 Why This Stands Out: xAI’s move into persistent memory reflects a growing trend: AI chatbots are shifting from session-based tools to long-term, adaptive assistants. With user data privacy in the spotlight, Grok’s emphasis on transparency and control may help differentiate it in a crowded market. The question now is whether Grok’s pace can keep up with OpenAI and Google’s rapidly evolving ecosystems.

WEB

Deck Raises $12M to Unlock the Web's Hidden Data

🚀 TodayOnAI Insight: Montreal-based startup Deck has raised $12 million in Series A funding to expand its platform that enables user-permissioned access to data from websites without APIs. Positioned as a “Plaid for the rest of the internet,” Deck aims to unlock fragmented, siloed online data across industries using AI-powered browser automation.

🔍 Key Takeaways:

• Core tech: Deck’s AI agents mimic human browsing to log in, navigate, and extract structured data from any web portal—no API required.

• User control: Operates with explicit user consent and supports dual authorization from both users and client platforms.

• Growth: Connections surged 120% month-over-month in February; pricing is based on successful API calls.

• Use cases: Initially focused on utilities—connecting to over 100,000 providers in 40+ countries—but expanding into sectors like e-commerce, payroll, and government.

• Funding & team: Raised $16.5M to date; founders previously built Flinks, acquired for $140M. Infinity Ventures led the round, with participation from Intact Ventures and others.

💡 Why This Stands Out: As generative AI scales, data access is becoming a critical bottleneck. Deck's infrastructure tackles this by transforming the open web into a developer-friendly platform, filling gaps where APIs fall short. With rising pressure for data portability and transparency, Deck’s model suggests a broader shift: the future of AI may depend not just on models—but on unlocking the fragmented web they learn from.

OPENAI

OpenAI Launches Codex CLI, an Open-Source AI Agent for Your Terminal

🚀 TodayOnAI Insight: OpenAI has launched Codex CLI, a lightweight, open-source AI coding agent that runs locally in the terminal. The tool bridges OpenAI’s models with command-line interfaces to automate coding tasks directly on a user’s machine—part of a broader push toward “agentic” software development.

🔍 Key Takeaways:

• Local execution: Codex CLI connects OpenAI’s models to your desktop environment, enabling them to write, edit, and act on code through the terminal.

• Multimodal input: The tool supports screenshots and sketches for reasoning, combining visual and code-based inputs in one interface.

• Model integration: Codex CLI will integrate with the latest OpenAI models, including o3 and o4-mini.

• Open-source release: Developers can explore and customize the tool freely.

• Funding initiative: OpenAI is offering $1M in API grants ($25K each) to projects building with Codex CLI.

💡 Why This Stands Out: Codex CLI reflects OpenAI’s growing ambition to embed AI deeper into software engineering workflows—not just as assistants, but as autonomous agents. While still early-stage, this step hints at a future where AI can build and test entire applications from brief descriptions. The move also underscores the shift toward local, transparent tools—will developers trust AI more if it runs on their own machines?

OPEN AI

OpenAI’s New o3 Model Can Reason with Sketches, Diagrams, and More

🚀 TodayOnAI Insight: OpenAI has unveiled its most advanced AI model yet, o3, designed to analyze and reason with images—including low-quality sketches and whiteboards—marking a major step toward multimodal reasoning. A smaller, faster variant called o4-mini also launched, as the company seeks to outpace rivals like Google, Anthropic, and xAI in the generative AI race.

🔍 Key Takeaways:

• New model capabilities: o3 can interpret and reason through visual inputs like sketches, diagrams, and whiteboards—even when the image quality is poor.

• Multimodal integration: The model independently uses all ChatGPT tools, including browsing, Python, image understanding, and generation, to solve complex problems.

• Model variants: o4-mini offers faster performance at a lower cost, while o3 is optimized for math, coding, and scientific analysis.

• Access and availability: Both models are now available to ChatGPT Plus, Pro, and Team subscribers.

• Safety and scrutiny: OpenAI claims both models were vetted under its most intensive safety protocols to date, though recent policy changes have drawn criticism.

💡 Why This Stands Out: OpenAI’s push into visual reasoning shows its ambition to create AI that doesn’t just generate content, but truly understands it across modalities. As the company moves faster than ever—sometimes ahead of transparency norms—it raises important questions: Can safety keep pace with innovation? And who sets the standards when everyone’s racing to lead?

✨ See how a simple prompt upgrade can unlock better AI output.

🔹 The Original Prompt

At first glance, this prompt might seem okay. But it's too broad — and that limits the quality of AI-generated results. Let’s improve it using prompt engineering best practices.

✅ The Improved Prompt

💡 Why It's Better

• Specifies the audience and niche (marketing agency for SMBs)

• Adds content themes to ensure relevance and variety

• Clarifies the desired structure for easy scanning and execution

• Focuses on value-driven content (education, tools, strategies)

🛠️ Learn how to adapt this prompt for SaaS, AI tools, dev teams & more →
Read the full PromptPilot breakdown

💡 Bonus Tool: Want to generate and master prompts instantly?
👉 Try PromptPilot by TodayOnAI (Free to use)

📰 More from the AI World

• Trump administration reportedly considers a US DeepSeek ban

• OpenAI’s latest AI models have a new safeguard to prevent biorisks

• OpenAI is reportedly in talks to buy Windsurf for $3B, with news expected later this week

• OpenAI partner says it had relatively little time to test the company’s o3 AI model

🧰 Today’s AI Toolbox Pick

• 🏄Surfer SEO (SEO Tool): Skyrockets your organic traffic.

• 📱Langbase (Apps Tool): Builds, deploys, and manages generative AI apps.

• 💻Replit AI (Coding Tool): Automates the repetitive parts of coding.