Weekly AI Highlights: February 14 - 21, 2026
What Small Business Owners Need to Know Right Now
Reader, what a week!
Unless you’ve been under a rock, you’ve read Matt Schumer’s “Something Big is Happening” article. It has created such a buzz I had to weigh in. I recorded this podcast episode today but it won’t be promoted until Tuesday. However, you can listen to it here early, because you are a VIP. 🙂
And the week was jam-packed with other big stuff too….
Anthropic made Claude Sonnet 4.6 strong enough to handle most of your daily knowledge work, even with long documents and bigger context windows. Google is pushing reasoning forward with Gemini 3 Deep Think, built for the kind of complex problems where accuracy really matters.
But I have to tell you… trying Manus' AI agent (inside Telegram) honestly blew me away this week.
1. Manus rolls out personal agents in Telegram
Manus has launched personal AI agents inside Telegram, meaning you can now message your agent directly and assign it real tasks from your phone.
Instead of logging into a dashboard, you just send a message. The agent can research, browse the web, build documents, redesign pages, send emails, and work on multiple tasks at once. It’s designed to feel less like using a tool and more like texting an assistant.
…And they’ve made the setup SO EASY. So if you have been itching to try Clawdbot/OpenClaw and the like but have felt completely confused by the ten million steps you have to go through to set it up, you are going to love this!
Source Link: https://manus.im/blog/manus-agents-telegram
MY TAKE
This Telegram integration is wild!!!
I tested it myself and started simple. The first thing I asked it to do was find thoughtful counterarguments to a trending AI article so I could see both sides before forming my own opinion. It pulled them almost instantly.
Then I pushed it further. I asked it to rebrand one of my sales pages using my updated branding and linked to my Notion branding doc. It rebuilt the page and looked beautiful. Clean. On brand. Look at it here:
https://frontrowaiclub.com/
While that was running, I gave it another task: create my upcoming Claude training, including a full slide deck and supplemental materials based on an email we sent. It handled both at the same time. And it delivered everything faster than it estimated.
I also asked it to find and complete a speaker application for an upcoming conference…and a few other tasks.
It feels less like prompting and more like delegating. Take a look at the screenshots below so you can see exactly how this played out.
(You’ll notice I call it “Oliver”. That’s because when you set it up it asks “What do I call you? What will you call me?” And I’ve always loved the name Oliver 😀)
ACTION STEP (DO THIS WEEK)
If you already have a Manus account, connect it to Telegram. Telegram is free, and this is where the personal agent actually becomes useful.
Once it’s connected, give it one real task you’d normally spend time on. Not a test. Something meaningful, like researching a topic, reworking a page, or outlining a training. That’s when you’ll see what it can actually do. (Read all security risks before proceeding!)
|
2. Anthropic Launches Claude Sonnet 4.6 Offering Opus-Like Results at Lower Cost
Anthropic just released Claude Sonnet 4.6, and there are a few updates you should know about.
- It now supports a 1 million token context window, which means it can handle much longer documents and conversations without losing track.
- It will take longer to hit usage limits. That alone makes it more practical for daily work.
- It’s clearly optimized for knowledge workers and Claude Cowork usage. It’s smarter at handling multi-step tasks inside Cowork, including the computer-use capabilities that let Claude take action in Chrome.
Opus 4.6 is still stronger for very complex workflows and heavy coding. But for most non-coding work, Sonnet 4.6 is a noticeable step up from 4.5.
It’s also cheaper, which matters if you’re using the Anthropic API.
Source:
https://www.anthropic.com/news/claude-sonnet-4-6
MY TAKE
This is probably the new default. If you’re not doing heavy coding or extremely complex workflows, Sonnet 4.6 gives you better performance, longer usage, and lower cost. That’s a solid upgrade.
ACTION STEP (DO THIS WEEK)
If you use Claude Cowork, switch your default model to Sonnet 4.6 and run one real task through it. Pay attention to how long it lasts before hitting limits and how well it handles longer context. If you’re not coding, this may become your go-to model.
|
3. Gemini 3 Deep Think Is Here
Google has introduced Gemini 3 Deep Think, a new reasoning mode designed for more complex, multi-step thinking. Instead of giving you a quick answer, Deep Think spends more time working through harder problems before responding.
This model is built for advanced reasoning tasks like math, science, coding, and structured analysis. It’s meant to handle deeper, more technical challenges where accuracy matters more than speed.
Source Link:
https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-deep-think/
MY TAKE
This is clearly aimed at serious problem-solving. For everyday prompts, you probably won’t notice a huge difference. But if you’re working through something complex where reasoning actually matters, this is the direction the models are heading.
NEW TRAINING: Comprehensive Claude Training
February 25, 2026 at 12 PM ET | 9 AM PT | 5 PM UK
Everything you didn’t know you needed to know about Claude!
YOU'LL LEARN:
- Why Claude is different from other AI tools (and why that matters for your business)
- How to set up Claude the right way so you get the best results from day one
- The basics of Claude Code — what it is, how it works, and how to start using it even if you're not technical
- How to use Claude Cowork to automate tasks right from your desktop
- How to use Projects, Styles, and Memory to streamline your workflow
- Real examples and live demos so you walk away ready to put Claude to work immediately
(Just click. No need to type in your email, we already have it.)
PROMPT OF THE WEEK
The “Do Less” Optimization Prompt
This prompt helps you get better results by removing over-instructions like “be thorough” or “try harder.” It’s designed to stop the model from overthinking and looping so you get cleaner, more direct output.
Anthropic just dropped updated prompting best practices for its new Claude 4.6 models, and the biggest takeaway is counterintuitive: do less.
If you've been stuffing your prompts with phrases like "be thorough," "think carefully," or "you MUST use this tool," it's time to delete them. Those bandaid fixes for older models now actively hurt performance on 4.6. As one Anthropic engineer put it: "I saw huge leaps in intelligence as soon as I removed anti-laziness prompts."
Here's the new playbook:
- Remove "try harder" language. Phrases like "be thorough" or "do not be lazy" cause the model to overthink and loop. 4.6 is already proactive; telling it to try harder is like yelling "RUN FASTER" at someone already sprinting.
- Soften tool instructions. Swap "You MUST use this tool" for "Use this tool when it would help." The model triggers tools appropriately now without being threatened into it.
- Be explicit about actions. "Can you suggest changes?" will get you suggestions. "Make these changes" will get you changes. Say what you mean.
- Use effort settings as your main dial. Instead of prompt-engineering your way to better results, adjust the effort parameter (low, medium, high) to match your task.
Source: theneuron.ai
|
The Bottom Line
This week belongs to AI agents that actually execute. Sonnet 4.6 is a strong, practical upgrade for daily work, and Gemini Deep Think looks promising. But Manus inside Telegram is what has me really feeling a certain kind of way. I’m dazzled, disoriented, and a little out of breath, to be honest. It genuinely feels like working with a REAL assistant.
Comments
If these updates made you think differently about how you want to work with AI, hit reply and let me know. I read every message, and your questions help shape what I cover and teach next.
Have a great weekend,
Founder, Front Row AI Club
PS. Was this newsletter forwarded to you? Sign up to get it in your inbox.
You can find my previous newsletter articles here: The Front Row AI Report
© 2026, Front Row AI Club. All rights reserved.