News & Trends: OpenAI’s Big Moves in Fall 2025
Sora 2 Stuns with Cinematic AI Video Breakthrough – OpenAI’s Sora 2 introduces near-photorealistic text-to-video generation with synchronised audio and advanced scene control. The companion Sora app adds a TikTok-style interface for sharing AI-generated clips, signalling a shift in how creative industries approach video production. View Details
DevDay Unleashes AgentKit and Apps SDK, Redefining AI as a Platform – OpenAI’s DevDay unveiled AgentKit and the Apps SDK, letting developers build, deploy, and monetise AI agents directly within ChatGPT. It positions ChatGPT as an AI-native platform comparable to an operating system. Learn more
Apps in ChatGPT Open a New Chapter for AI-Native Experiences – OpenAI has introduced full app functionality inside ChatGPT, allowing developers to build and monetise interactive apps directly within the chat interface. Users can now browse, install, and use apps seamlessly alongside conversations, transforming ChatGPT into an all-in-one AI workspace. View Details
Instant Checkout Transforms ChatGPT into an E-Commerce Hub – The new Instant Checkout feature allows users to complete purchases directly within ChatGPT. It’s part of OpenAI’s strategy to make conversational AI a seamless gateway for shopping and transactions. Learn more
ChatGPT Pulse Enhances Team Collaboration in Real Time – ChatGPT Pulse adds real-time collaboration and proactive updates for teams, integrating chat history and external tools to deliver daily insights. It moves ChatGPT closer to functioning as an intelligent workplace assistant. View Details
GPT Realtime Brings Affordable, Low-Latency Voice AI – The GPT realtime API offers low-latency streaming for natural voice interactions at reduced cost, opening the door for lightweight, voice-driven applications. View Details
Step by Step Guide on How to Use ChatGPT Agent Mode
OpenAI has introduced ChatGPT Agent Mode, a powerful evolution of the ChatGPT interface that allows the model not just to think and respond, but to act in your digital environment. Instead of you giving a single prompt and waiting for text output, Agent Mode lets ChatGPT take over more of the execution interacting with web pages, running code, navigating interfaces, and producing final deliverables.
This built-in mode integrates capabilities from OpenAI’s previous tools (such as “Operator” and “Deep Research”) into one unified system that can go from research to real-world action.
In this edition, we’ll explain what Agent Mode can do, how you get started, and share examples and prompts you can try immediately. What ChatGPT Agent Mode Can Do
ChatGPT Agent Mode expands the model’s abilities beyond conversation. It allows the AI to perform structured actions such as browsing the web, analysing data, and creating files, giving professionals a hands-on way to automate research and task execution.
Key Capabilities
- Autonomous browsing – Navigates websites, searches information, and gathers data without manual input.
- Research and synthesis – Conducts multi-step research and summarises findings into concise reports.
- Code execution and analysis – Runs code or processes data for technical and analytical tasks.
- Document generation – Creates structured outputs such as reports, spreadsheets, or presentations.
- App connectors – Accesses read-only data from tools like Gmail or Google Drive for added context.
- Safety controls – Requires confirmation for sensitive actions and allows users to monitor or pause activity.
ChatGPT Agent Mode is available only to paid users in supported regions. It does not retain memory between sessions, so previous context is not remembered. Visual outputs may need minor formatting adjustments, and high-risk actions such as logins or payments require user confirmation. These safeguards help maintain security while keeping users in control.
Watch How to Use ChatGPT Agent Mode
Getting started with ChatGPT Agent Mode is straightforward. Once logged into ChatGPT (via browser or desktop app), open the Tools menu and select Agent Mode. You can also activate it using the /agent command in the chat bar.
For a step-by-step walkthrough, watch our short tutorial video:
Watch how to use ChatGPT Agent Mode. Three Interesting Things You Can Use Agent Mode For
- Automated competitor landscape briefing Ask the agent to explore a sector, collect data on top competitors (financials, product features, news mentions), and deliver a slide deck comparing them.
- Inbox triage & action planning Let the agent read your Gmail (read-only), categorise by theme (e.g. “urgent,” “follow-up,” “info”), and produce a one-pager with action items or a prioritised task list.
- Report refresh & dashboard update Ask it to pull quarterly data from spreadsheets or databases, compute deltas vs prior periods, and produce an updated report or summary document.
These are just starting points. Agent Mode is flexible once you define a clear scope and structure.
Sample Prompts to Try
Below are prompts you can paste into ChatGPT (after enabling Agent Mode) to see it in action:
- “Summarise my inbox by topic over the past week and draft a one-page action plan with next steps.” (Uses Gmail connector for context; produces an organised action summary.)
- “Compare three competitors in the AI consulting space, summarise their offerings, pricing, and recent announcements, then build a 5-slide deck with insights and recommendations.”
- “In the Google Drive folder named ‘Q3 Deliverables’, extract key metrics from the reports, compute growth vs Q2, and deliver a table with observations and a narrative summary.”
Connect with GenFutures Lab
At GenFutures Lab, we help organisations harness AI responsibly and effectively from strategy and adoption to workforce upskilling.
If you’re exploring how AI agents, automation, or training could transform your organisation, we’d love to hear from you. 👉 Book a consultation or connect with us