🤝 Meta's Agents for Its AI Brain

Hey folks,

Happy New Year 🌴 I’ve had a week off in Vietnam to relax and bring in the New Year and pumped to get back into 2026 with lots of plans. I hope you managed to reset your mind, goals and your passwords!

Did you hear that thousands of people stood freezing waiting for Brooklyn Bridge fireworks that didn't exist? AI-generated TikTok and Instagram posts had promised a spectacular display. ChatGPT gave "tips" on the best viewing spots. Even Time Out New York and Secret NYC listed it as a top destination. When the clock struck midnight there was silence. No fireworks, no announcement, just confused crowds realizing they'd been duped by AI slop. It's a perfect metaphor for 2026, the age of information is also the age of misinformation.

Today in AI: 

→ Meta's $2B bet on autonomous agents
→ OpenAI's audio-first future (no screens required)
→ How to use AI to generate a menu sample
→ Prompt of the day: Turn complex technical decisions into recommendations

Let’s dive into 2026 🚀

🤝 Meta Just Bought the Agents for Its AI Brain

$2 billion. Ten days to close. That's how badly Meta wanted Manus.

The Singapore-based AI agent startup hit $100M ARR in just eight months, processing over 147 trillion tokens and powering 80 million virtual computers. Now it's Meta's third-largest acquisition ever, behind only WhatsApp and Scale AI.

What Manus actually does:

• Executes complex multi-step tasks: market research, coding, data analysis

• Runs autonomous agents that work across applications

• Processes requests at scale most startups can't touch

Meta's plan: integrate Manus agents directly into Facebook, Instagram, and WhatsApp where Meta AI chatbot already serves billions. Now Meta will complete tasks.

But here's what's strategically interesting: Meta immediately cut all Chinese ties. No Chinese investors remain. No China operations continue. Meta's clearly positioning this for enterprise and regulatory approval.

For your team: If you're building internal AI agents, watch how Meta deploys Manus across its ecosystem. The integration patterns will become templates for enterprise agent architectures.

Together with Gladly

Get the investor view on AI in customer experience

Customer experience is undergoing a seismic shift, and Gladly is leading the charge with The Gladly Brief.

It’s a monthly breakdown of market insights, brand data, and investor-level analysis on how AI and CX are converging.

Learn why short-term cost plays are eroding lifetime value, and how Gladly’s approach is creating compounding returns for brands and investors alike.

Join the readership of founders, analysts, and operators tracking the next phase of CX innovation.

Subscribe here to stay informed and ahead of the curve.

Subscribe now

🎧 OpenAI Is Building What Comes After Screens

Forget chatbots. OpenAI is betting its hardware future on a world where you never look at a screen.

The company has unified engineering, product, and research teams to overhaul its audio models, all preparing for an audio-first device launching late 2026 or early 2027. Internally codenamed "Gumdrop," the device is shaped like a pen (think iPod Shuffle size) with no dedicated screen.

The technical shift:

• New audio model architecture produces responses that sound more natural and emotive

• Can speak simultaneously with users. Current models can't do this

• Better interruption handling and contextual awareness via cameras and microphones

• The device could be part of a broader family including smart glasses and screenless speakers.

After acquiring Jony Ive's io Products for $6.5B last May, OpenAI has the design firepower to actually pull this off. Ive's vision is less addictive, more ambient computing.

For your team: Start thinking about voice-first interfaces for internal tools. If OpenAI's bet pays off, audio-based AI interaction becomes the new default UX and your apps need to work without screens.

🎯 How To Order From a Chinese Menu

This would have been better sent before the holiday period, but very useful when you want to know what’s on a menu that’s not in English. Try this

1. Go to ChatGPT

2. Upload a photo of the menu

3. Prompt the following

Turn this menu into English and make images of each dish like on a menu.

🎯 Prompt Of The Day

Turn complex technical decisions into clear recommendations for stakeholders:

I need to evaluate [TECHNOLOGY/TOOL] for our team. Create a decision framework with: 

1) Key evaluation criteria weighted by importance
2) Comparison against our current solution and 2 alternatives
3) Total cost of ownership over 2 years including migration 4) Risk assessment with mitigation strategies
5) A clear recommendation with confidence level. 

Format as a one-page executive brief.

🐝 AI Buzz Bits

💰 Nvidia's $20B Groq deal is an acquisition that isn't one. Structured as a "non-exclusive licensing agreement" to avoid regulators, 90% of Groq employees (including CEO Jonathan Ross) are joining Nvidia. The chip giant gets Groq's LPU inference technology while Groq technically remains "independent." Antitrust theater at its finest.

📸 Instagram's boss says it's easier to prove what's real than chase what's fake. Adam Mosseri's year-end memo dropped a bombshell: with AI-generated images now indistinguishable from photos, Instagram will shift to "fingerprinting real media" instead of detecting fakes. His solution? Camera manufacturers should cryptographically sign images at capture. The polished feed aesthetic is "dead"—raw, unflattering content becomes proof of humanity.

📈 Anthropic targets $26B revenue in 2026. That's nearly triple their 2025 numbers and more than OpenAI's projected $20B. They're eyeing a $300B+ IPO valuation, potentially larger than OpenAI's. The Claude maker expects $70B revenue and $17B positive free cash flow by 2028.

🛠️ Tool Spotlight

1. Manus — General-purpose AI agent that executes complex tasks like market research, coding, and data analysis autonomously; now part of Meta's ecosystem.

2. MCP SDK — The official Model Context Protocol SDK for building servers that connect AI agents to your tools, databases, and APIs.

3. Dimension — Proactive AI assistant for engineering teams that catches failed Vercel deploys at 2am, checks logs, and fixes issues with context from your PR history.

4. Traycer — Plan first AI coding for large codebases; creates structured specs before handing off to your agent (Cursor, Claude Code, Windsurf), then verifies changes and catches regressions. Free tier available.

For a full list of 1500+ AI tools, visit our Supertool Directory