• WKND AI
  • Posts
  • 2025 The Year Of 'Agents'🤖

2025 The Year Of 'Agents'🤖

+Meta Acquires Manus

Hello WKND AI Warriors!

AI agents finally move from demos to real products, with companies embedding them into research, support, and operations and demanding clear ROI in 2026.

Also, Meta says it will cut Chinese ties and shut down China operations as scrutiny grows around its Manus acquisition.

Plus, SoftBank completes a $40B+ investment in OpenAI, valuing the company near $500B and making it one of the fund’s biggest bets ever.

Oh yeah, and Instagram says AI content will become so common that platforms may need to fingerprint real media instead of fake to signal authenticity.

Today’s newsletter includes:

  • 📰 AI NEWS RECAP

  • 🤿 AI DEEP DIVE

  • 📝 AI PROMPT OF THE WEEK

  • 🎨 AI IMAGE OF THE WEEK

🎓AI COURSE OF THE WEEK

Most AI content is built for clicks, not people with real jobs.

This training is for professionals who are smart, busy, and just late to AI.

In one weekend, you’ll build prompts and workflows for your job, not generic “10 best prompts” screenshots. ​

Crash course, deep dive, prompt library, tool stack…done.

👉 Get instant access here.

📰 AI NEWS RECAP

2025 The Year Of 'Agents'🤖

AI agents had a breakout year in 2025.
But they were not problem free.

Fast Company just looked back at the year in AI agents, the tools that do work for you instead of just answering questions.

Here’s what stood out:

Agents got useful fast
People used them to automate tasks that used to take real effort, like data collection, scheduling, or coordinating steps across apps.

But the bugs showed up too
Agents misunderstood instructions.
They acted inappropriately.
They made decisions people did not expect.

What everyone learned

AI agents work best when they have clear boundaries and human supervision.
Unchecked autonomy caused more confusion than progress.

Here’s the lesson for professionals

  1. Treat AI agents like apprentices, not managers
    Let them do the repetitive work, but check results before decisions are made.

  2. Define steps, not goals only
    The clearer the instructions, the better the outcomes.

  3. Build verification into your process
    Humans in the loop are not overhead. They are risk control.

Agents are here to stay, but they need structure.
In 2026, the leaders will be the ones who can deploy responsibly, guide clearly, and verify consistently.

AI might take on tasks.
But humans still decide the direction.

Meta x Manus
Meta is buying Manus for about $2 billion after the startup hit millions of users and over $100 million in ARR, and plans to keep it running independently while stripping out Chinese ownership and shutting its China operations.

SoftBank x OpenAI
SoftBank has now fully funded roughly $40–41 billion into OpenAI for around a 10–11 percent stake, making it one of SoftBank’s biggest AI bets and valuing OpenAI near $500 billion.

Instagram and authenticity
Instagram chief Adam Mosseri says AI is becoming so common that it will be more practical to “fingerprint” real media than fake, pushing platforms to highlight raw, less-polished content as a signal of authenticity.

Grok for business
xAI launched Grok Business and Enterprise as workplace assistants that search internal docs, analyze data, and help teams collaborate, with higher tiers adding security, audit, and a private vault so company data stays isolated.

OpenAI’s audio push
OpenAI is doubling down on voice, building a much more conversational audio model and a 2026 audio-first device aimed at replacing some screen time with always-available, talk-first assistants.

LeCun vs Wang
Yann LeCun called new Meta AI leader Alexandr Wang “inexperienced,” hinting at doubts about his research depth and suggesting Mark Zuckerberg is unhappy with recent Llama drama and the company’s AI trajectory.

🤿 AI DEEP DIVE

Still haven’t tried Google’s NotebookLM?
Now it makes videos.
Here’s your chance, watch this!

📝 AI PROMPT OF THE WEEK

Copy and paste this into your favorite chatbot.

USE CASE: Research brief with sources and gaps.

PERSONA: Analyst.

CONTEXT: Topic is [topic], audience is [audience], deadline is [date].

TASK: Create a research brief with key questions, source list, and gaps to close.

FORMAT: Objective, Questions to Answer, Key Terms, Sources table, Gaps and Close Plan, Five Key Takeaways.

TONE: Neutral, scholarly.

Master prompting with the PCTFT framework.

It turns generic AI outputs into laser focused insights.

Why it works?

Persona: Assigns a specific role to the AI.​

Context: Provides background information.​

Task: Defines the specific action or response needed.​

Format: Specifies the structure of the output.​

Tone: Indicates the desired style or mood.​

🎨 AI IMAGE OF THE WEEK

Give AI image creation a try for yourself.

Copy the text below into Google’s AI image creator!

Ultra-realistic colorful macarons filled with cream and lavender jam, dynamic explosion of macarons and cream filling splashing outwards, crumbs and powdered sugar flying in the air, cinematic dark red gloomy background matching the series of luxurious desserts, 8K ultra-detailed, shallow depth of field, luxurious food photography, frozen moving splash, dramatic studio lighting, elegant composition, soft bokeh, high-quality advertising style 

How'd you like this newsletter?

Love it or hate it? Let us know why!

Login or Subscribe to participate in polls.

How can you help?

Refer my newsletter to help others learn AI.

Missed last week’s edition?