Hi {{first_name}}!
Happy Thursday,
I missed last week's newsletter, since I was in Arizona for a conference… So this one's packed with extra insights to make up for it. On that note, if you have a moment to provide your feedback on your preferred length and frequency for this newsletter I’d greatly appreciate it!
This week the big news is OpenAI’s new Atlas browser.
Beyond covering Atlas and other noteworthy updates I’ll also be diving into how the big four frontier models (aka the large general purpose AI models - ChatGPT, Gemini, Claude and Grok) stack up against one another. I’ll be sharing which model works best for different business tasks. I still find that most people I talk to at my AI trainings and presentations are still only using ChatGPT for almost everything, which is something I’m hoping I can help you change!
So, this week we're covering:
New and Noteworthy Updates - Haiku 4.5 crushing costs, Slack becoming an AI hub, and more
AI Model Comparison - Where each of big 4 frontier models shine
ChatGPT Atlas Browser - OpenAI's new browser that integrates AI directly into your browsing experience
OK, let's get into it!
New and Noteworthy
Recent AI Developments
Slack's AI Ecosystem: Slack is evolving beyond a single Slackbot into a comprehensive AI ecosystem powered by OpenAI, Anthropic, and Perplexity. The beta (now at 70,000 users) lets you use ChatGPT, Claude, and other assistants directly in channels or DMs. The AI pulls from live Slack data (messages, files, workflows) without breaking permissions. You can tag @Codex to write code or pull data from Google Drive and Salesforce without leaving Slack. This feels like the start of every workflow living right where the conversation begins.
Claude Haiku 4.5 Launch: Anthropic just dropped a game-changer. Haiku 4.5 does Sonnet 4's job for one-third the cost and twice the speed. On real coding tasks, it even beats Sonnet 4 at some things. The economics of AI just shifted. You can now get enterprise-level AI performance without the enterprise price tag.
Gemini's "Help Me Schedule" in Gmail: Google added an AI scheduling tool in Gmail that suggests meeting times from your Google Calendar. It appears under the email compose screen, lets you edit time options, and adds meetings automatically once confirmed. Making scheduling quicker and smarter.
Gemini Gems become sharable: Google Gems have always been a powerful feature of Gemini (similar to CustomGPT’s) and they are a great way to save time on repetitive prompts. However, up until recently it wasn’t possible to share Gems. Luckily that has changed and you can now share Gems (with individual people or via a public link that anyone can access) just like you would share a file in Google drive.
Microsoft's MAI-Image-1: Microsoft launched an image generation model designed for photorealistic images. They worked with creative professionals to avoid repetitive or generic outputs. The model processes prompts faster than larger, slower models and currently ranks in the top 10 on LMArena's image-quality leaderboard.
ChatGPT Personality Updates: Sam Altman announced personality improvements for ChatGPT, The first update brings back the conversational style people loved about ChatGPT-4o, with additional features rolling out by December. The AI companion market is huge, and OpenAI isn't letting competitors take the lead.
AI Model Comparison: Finding Your Perfect Match
I get asked this constantly: "Should I use ChatGPT or Claude or Gemini?" Here's my honest take after experimenting with all of them for work purposes. There's no single "best" AI model. Each one has superpowers, and blind spots.
Here’s My Take on the Big 4 (and when to use each)
Best for: All-around business and personal tasks
This is your reliable workhorse. Strong at writing, solid at coding, great at reasoning and research. If you're new to AI or need one tool that handles most situations well, start here. Assuming you use this regularly it is 100% worth the cost of upgrading from the free plan to the $20/mo (personal) or $25/mo (teams) plan for a multitude of reasons (security, privacy, features, usage limits… etc). I use it often reviewing emails and documents, brainstorming, researching etc and it has effectively become my google replacement in many situations.
Best for: Deep thinking, project based work, analysis of long documents
Claude is my go-to for serious business work. It handles nuance better than other models, processes massive documents, and actually "thinks through" complex problems. I use it for strategy sessions, analyzing contracts, and anything requiring careful reasoning. The Projects feature is unmatched for organizing work by client or project. I also love Claude’s ability to fine tune your output to match your tone and voice.
Best for: Real-time research, integrated image and video generation and google workspace integration.
When I need facts from the web or recent developments, Gemini delivers (as it should, being a Google product). Its search integration means you get accurate, up-to-date information without the hallucinations. Great for market research, competitor analysis, or anything where accuracy matters. I also find myself using Gemini often with the integration of their Nano Banana image generation tool and their Veo3 video generation tool.
Best for: Real-time insights and unfiltered responses
Grok connects directly to X (formerly Twitter) for the most current information and trending conversations. It's built to be more conversational and less filtered than other models, helpful when you want straight talk instead of corporate-safe responses. Worth testing for social listening and trend spotting.
My actual workflow:
Editing emails and quick questions: ChatGPT
Strategy, projects and deep work: Claude
Research, image and video generation: Gemini
Social trends and real-time intel: Grok
Over the next few weeks, I'll go deeper into each of these models inside the Ampra Circle community, sharing specific use cases, prompting strategies, and real examples of how I use each one.
Tool Spotlight: ChatGPT Atlas Browser
OpenAI just dropped something big: ChatGPT Atlas, their first web browser that puts AI at the center of your browsing experience.
Here's what makes it different: Instead of opening a browser and then switching to ChatGPT when you need help, Atlas integrates AI directly into your browsing. It sees what page you're on, can answer questions about the content in real-time, and helps you take action without leaving your current tab.
Key features that matter for business:
Context-aware assistance: Atlas knows what webpage you're viewing and can answer questions about it instantly via the sidebar with the context of your history.
Smart suggestions: Get AI-powered recommendations while typing anywhere on the web - emails, forms, documents
Tab management: Ask Atlas to open, close, bookmark, or revisit tabs using voice or text commands
Agent mode: In preview for Plus, Pro, and Business users, Atlas can complete multi-step tasks as you browse like researching services, comparing options, and even making purchases
Real business applications:
Research competitors while Atlas summarizes their websites and pulls key insights
Draft responses to customer emails with context from your current conversation thread
Navigate complex workflows by asking Atlas to handle the clicking and form-filling
Review contracts or documentation with instant AI analysis of what you're reading
Edit an email directly in your web based email client (ie Gmail) by selecting your message draft and prompting it in-line. This replaced the very common flow that most people have of copying an email draft from gmail into ChatGPT then back into Gmail once it’s proofed or edited.
How it compares to Chrome's Gemini integration: Google just rolled out Gemini directly into Chrome as well, and both are impressive. Chrome's implementation works more like a built-in assistant that activates when you need it, which is great for quick questions about web content. Atlas goes deeper with its agent mode, actually taking actions on your behalf across multiple steps. Think of Chrome + Gemini as your research assistant, while Atlas is more like a virtual employee who can complete tasks. Both are free to try (Atlas requires ChatGPT Plus), so test them both and see which workflow fits you better.
We're watching the browser wars evolve fast. OpenAI is positioning Atlas as a direct competitor to Chrome, Perplexity, and Arc… all browsers that already have massive user bases.
Atlas is currently available for macOS only, with broader rollout expected soon. Check it out at chatgpt.com/atlas
That's it for this week! I'm curious which AI model you're currently using most (outside of ChatGPT) and whether this breakdown changes your approach. Hit reply and let me know what's working (or not working) in your daily workflow.
Looking forward to hearing from you,
Julien
PS: If you found this comparison helpful, feel free to forward it to someone on your team who's still trying to figure out which AI tool to use. They can sign up for the newsletter here for free.