OpenAI Super App: Old wine, New bottle

OpenAI Inc. is said to be planning the biggest overhaul of ChatGPT since the chatbot burst onto the scene in November 2022. According to Financial Times, the company aims to transform ChatGPT into a “super app” that integrates AI agents, coding tools and third-party services.

The reported move comes as OpenAI has confidentially filed an S-1 with the US Securities and Exchange Commission, formally taking its first step toward a public listing. SpaceX Inc. and Anthropic have also filed S-1s, while Perplexity has publicly disclosed its IPO ambitions.

In this week’s edition of Tech Talk:

• AI IPOs: SpaceX, OpenAI & Anthropic
• Fable 5, Claude Mythos in a safe avatar
• AI tool of the week: Gemini Daily Brief
• Anthropic says AI may eventually build itself
• TCS aims to have as many AI agents as humans

A ‘super app’, and an IPO

OpenAI, valued at roughly $850 billion and eyeing a listing, is burning through cash faster than it earns. Sam Altman may need such clever rebranding to push sales by transforming ChatGPT from a conversational chatbot into a full-stack “super app”. A senior OpenAI employee recently told the Financial Times that “chat is dead”. What the company needs is a platform that does things: books flights, writes code, designs presentations, places orders—all without the user leaving the ChatGPT interface.

To that end, OpenAI has unveiled an Apps SDK that embeds third-party services like Booking.com, Canva, Spotify, and Expedia directly into ChatGPT’s chat window. Codex, its AI coding assistant with over five million weekly active users, is also being pushed to the foreground. Autonomous agents capable of executing multi-step tasks are being positioned as the product’s new centre of gravity.

OpenAI needs to show Wall Street a credible path to profitability before it goes public. A super app, if it works, would deepen user engagement, justify premium enterprise subscriptions, and extract a toll from every third-party transaction conducted on its platform. The ambition, in short, is to become the operating system of the AI era.

History, however, counsels caution. Web portals tried precisely this in the late 1990s. Yahoo bundled search, email, news, finance, maps and social features into a single destination—and was undone not by a rival bundle, but by Google’s ruthless focus on doing one thing better than anyone else. When a superior alternative exists for any individual service within a bundle, users defect.

The Chinese super app experience is another case in point. WeChat succeeded because it controlled payments, owned the social graph, and operated within a relatively closed mobile ecosystem that favoured incumbency. Baidu attempted something similar and was steadily outflanked by rivals who controlled distribution channels it could not access.

OpenAI enjoys none of WeChat’s structural advantages. Apple and Google control the mobile OS layer in its key markets. European regulators have already restricted its third-party integrations across the UK, EEA and Switzerland with no committed timeline for resolution.

The early commercial evidence, too, is sobering. When Walmart piloted Instant Checkout inside ChatGPT, conversion rates fell well below its own channels. Walmart promptly replaced it with a proprietary assistant embedded in ChatGPT, retaining customer data and the transaction relationship while using OpenAI merely for distribution.

That said, what distinguishes OpenAI’s bet from its predecessors is the genuine novelty of AI agency—the ability to act across the internet on a user’s behalf, not merely retrieve and display. If that agentic capability proves transformative enough, it may generate the loyalty that bundling strategies alone never could. That remains, for now, an empirical wager rather than a proven proposition. The super app is the oldest trick in the technology playbook. OpenAI is staking its valuation on AI to make it work.

AI IPOs: A Deja Vu for investors

Investors have been here before. A transformative technology arrives, valuations detach from earnings and markets start pricing the future today. Capital markets financed booms in railways, dotcoms, social media and crypto plays, only to watch many companies collapse once reality crept up on what former Fed Chair Alan Greenspan called “irrational exuberance”.

Investors are now being asked to bankroll a new crop of companies whose sky-high valuations depend less on what they earn today than on what they might achieve someday. Think of the interplanetary ambitions of SpaceX or the quest for AGI at OpenAI and Anthropic.

With IPOs lined up in the US, those three American companies are expected to seek $220-250 billion from investors. Since only thin slices of equity will be listed on the stock market for trading, their valuations would be eye-popping: SpaceX worth almost $1.8 trillion, OpenAI and Anthropic at $1 trillion each. A sudden expansion in market capitalisation of such magnitude, even if only a fraction is free-float, would need a lot by way of business prospects to justify.

Here’s the full article.

AI TOOL OF THE WEEK

By AI&Beyond, with Jaspreet Bindra and Anuj Magazine

The AI capability unlocked today is Gemini Daily Brief

What problem does it solve? Most professionals start the day the same way: open email, check the calendar, scroll through yesterday’s threads, and spend 20 minutes figuring out what actually matters. The information is all there in Gmail and Google Calendar, but nobody is connecting the dots for you.

The cost is not dramatic. It is the slow daily tax of orientation time, the minutes you spend excavating context before you can do anything useful. Gemini Daily Brief works as a background orchestrator, reading across your inbox, calendar, and previous Gemini conversations to produce one consolidated, prioritised snapshot of your day.

How to access: https://gemini.google.com/daily-brief available to Gemini Ultra, Pro, and Plus subscribers.

Gemini Daily Brief can help you:

• Front-load your focus: Get a prioritised view of your entire day before opening a single inbox tab or calendar view.
• Act without switching: Reply to emails, set reminders, and schedule events directly inside the brief without leaving the view.
• Tune it to your needs: Give feedback on any item and Gemini adapts the next brief to better match your working style.

Example: A senior HR manager has back-to-back interviews starting at 9:00 am and needs a clear picture of the morning before opening a single app.

• Open the brief: Navigate to gemini.google.com/daily-brief. It presents a prioritised snapshot drawn from inbox, calendar, and past chats.
• Review the schedule: It surfaces the interview block with prep context pulled from earlier Gemini conversations on the same candidates.
• Catch the urgent item: It flags a department head’s email ranked above everything else in a 200-message inbox.
• Draft a reply: Prompt “Write a brief acknowledgement saying I’ll respond after my morning interviews”, and Gemini drafts it inline.
• Set a reminder: Add a follow-up task directly from the brief without opening a separate app.
• Narrow the view: Ask “Show me only this morning’s calendar items” for a faster scan before the first call.

What makes Gemini Daily Brief special?

• Cross-source intelligence: It connects Gmail, Google Calendar, and Gemini—three inputs most AI assistants treat as silos.
• Act inside the view: Unlike a passive summary, you can reply, schedule, and set reminders without leaving the brief itself.
• Adaptive over time: Tell it what to prioritise differently and it adjusts the next brief, not just the current one.

Note: The tools and analysis featured in this section demonstrated clear value based on our internal testing. Our recommendations are entirely independent and not influenced by the tool creators.

AI BITS & BYTES

Fable 5, Claude Mythos in a safe avatar

Anthropic PBC has unveiled Fable 5, a new Mythos-class AI model that it says is safe for general use, while outperforming all previous Claude models such as Opus, as well as rival models of OpenAI and Google Gemini on multiple benchmarks.

The Claude maker has also introduced Mythos 5, a version of the same underlying model with certain safeguards relaxed for a limited group of cyber defenders and infrastructure providers through Project Glasswing.

Still, Anthropic warns that these AI models could significantly lower the cost and complexity of cyberattacks through advanced vulnerability discovery and agentic hacking capabilities. Read more.

Anthropic warns AI may eventually build itself

AI is likely to eventually reach a stage where it build itself—a concept known as recursive self-improvement, Anthropic noted in a blog.

“We are not there yet, and recursive self-improvement is not inevitable. But it could come sooner than most institutions are prepared for.”

The length of tasks that AI models can reliably complete on their own is doubling roughly every four months, versus seven months earlier. If this trend holds, tasks that take a skilled person days could come into range this year. In 2027, AI systems could be capable of tasks that take a person weeks.

Incidentally, more than 80% of the code merged into Anthropic’s codebase was authored by Claude as of May. Read more.

TCS wants as many AI agents as human workers

Tata Consultancy Services Ltd. wants to become the world’s largest AI services firm—a goal that would require as much artificial intelligence as human. At least that’s what Chairman Natarajan Chandrasekaran would have us believe.

A slowdown in business—TCS’s annual revenue shrunk for the first time since listing in FY26—is a thing of the past, Chandra said at the company’s thirty-first annual general meeting, outlining a future where the IT bellwether could deploy as many AI agents as employees, and embed AI in all revenue streams.

“I predict that over the next three years, TCS will have as many AI agents as humans… In all, the day is not very far when TCS will have an equal number of AI agents, or AI workers, as it has physical workers.”

That, from a company that undertook its biggest layoffs last year. TCS ended FY26 with 584,519 employees, down by 23,460 compared to the previous year.

“Will AI lead to decrease in hiring? Absolutely. The company will not be hiring as it used to, Chandra said, “because certain portions of the work…will go to agents.”

🔗 The Sword of Damocles that’s Anthropic Claude

Meta + Reliance’s AI data centre in India

Meta Platforms Inc. has partnered with Reliance Industries Ltd. to lease its first AI-enabled data centre in India. RIL will build a 168 MW capacity data centre in Gujarat’s Jamnagar, which will then be leased by Meta to power its products and AI. That follows Google’s plans to set up a $15 billion, 1 GW AI hub in Andhra Pradesh’s Visakhapatnam.

According to IMARC Group, India’s data centre market is projected to nearly double to $13.11 billion by 2034, fuelled by increasing cloud adoption, digital transformation initiatives, and rising demand for AI computing capacity.

IN CHARTS

In May 2026, global cyberattack activity eased from April’s sharp rebound, but organisations experienced an average of 2,055 weekly cyberattacks—a 2% increase annually and a 7% decrease monthly.

While the monthly decline may read as stabilisation, ransomware activity surged to its highest year-on-year growth rate of 2026, and Gen AI-driven data exposure risks continued to deepen across enterprise environments.

ALSO READS

• Siri AI to iOS 27: Every new feature Apple announced at WWDC 2026
• Pentagon blacklists Baidu, BYD weeks after Trump-Xi summit
• Here’s how long it will take for AI to reach its potential
• Does workplace loyalty still matter when AI can take your job?
• The AI revolution worth betting on isn’t in your portfolio

Tech Talk is a weekly newsletter by Leslie D'Monte on everything happening in the world of technology and AI. Want this newsletter delivered to your inbox? Subscribe here.