|
|
|
Overwhelmingly Overwhelming
I recently watched an interview with Alex Hormozi, who, when asked about AI said, "I'm ashamed at my uses cases... I know I should be using it better."
And I get it!
With all the news about agentic AI and the billions of dollars being spent there's a feeling that you're falling behind, and it's moving too fast to catch up.
So this week's newsletter is designed to addresses this FOMO.
My suggestion is to take one AI application and learn it well. Zero in on small, actionable steps you can take right now to expand your knowledge of that one application without watching hours of YouTube videos or trying the latest and greatest AI application (which in my experience is usually not what it's cracked up to be anyway). Don't worry about complex strategies or automated workflows -- just simple improvements that will expand your base knowledge of the AI space -- and will compound over time.
|
|
Start at square one. If you jumped straight in to AI without learning some of the basics, or you're new to AI chatbots, build confidence by understanding the foundation and features
|
|
Simple prompting techniques like starting with a role cue ("you're a financial analyst") or providing sample output will dramatically improve the AI's response quality
|
|
Create your specialized AI. Instead of retyping the same prompts, build custom AI tools that remember your specific instructions, context, and preferred communication style for recurring tasks
|
|
|
Getting the Fundamentals Right
Most people jump straight into using AI without knowing how to make an AI conversation effective. This guide from Zapier suggests that successful AI interactions start with three principles: clear context, consistent communication preferences, and building on AI responses iteratively. The guide emphasizes that AI systems work best when you treat them like knowledgeable assistants who need basic orientation rather than mind-readers.
Why This Matters: Without these fundamentals, you're basically asking AI to guess what you want, every single time. Establish basic context and preferences upfront to eliminate vague responses and create a foundation for more sophisticated AI interactions.
|
Think Problems, Not Prompts
MIT's research on effective prompting reveals that improvements come from simple techniques rather than complex prompt engineering. The key insight is that AI systems understand conversational language, so adding basic context like "you're a financial analyst" or "respond like you're explaining to a new employee" immediately focuses the output. The research goes on to show that providing sample formats ("give me three bullet points like this example") or specifying your role and preferred communication style works better than elaborate instructions.
Small step: Start your next AI conversation with two elements: a role assignment ("you're a [specific expert]") and a format request ("give me this as [specific format]"). This simple framing will immediately improve response relevance and usability.
Read MIT's Findings →
Stop Retyping Over and Over Again (VIDEO)
Gemini's Gems solves one of the most frustrating aspects of AI usage: having to retype the same detailed prompts every time you need similar outputs. With Gems you can create your own "AI assistants" that remember your exact requirements. You define the AI's role, communication style, knowledge base, and operational constraints once, then access that specialized assistant whenever needed. The advantage is consistency - your custom Gem maintains the same expertise level, tone, and approach across all interactions.
Small step: Create your first Gem for one repetitive task you do weekly. Go to gemini.google.com, click "Explore Gems," then "New Gem." Give it a clear role (like "meeting notes formatter" or "email responder") and specific instructions about tone, format, and what to include or avoid.
Watch a Gems Tutorial →
|
Quick Hits
|
Learn Any Skill in 15-Minute Chunks
ChatGPT's Study Mode can break any skill into bite-sized daily lessons. Whether you need to understand industry jargon for a meeting or learn a new software feature, AI can structure learning into guided and manageable time blocks.
Small step: Have ChatGPT design a week-long crash course on a software tool you need to learn, broken into 15-minute daily sessions with specific features to practice and mini-exercises to complete.
|
|
PowerPoint, Word, Excel and PDFs from a Chat
Claude's new file creation feature lets you describe the documents you want to create. Instead of copying AI responses into other applications, Claude can return ready-to-use Excel spreadsheets, Word documents, PowerPoint presentations, and PDFs.
Small step:Ask Claude to create a spreadsheet or document for a task you do weekly (like "create a monthly budget tracker with automatic calculations" or "turn this data into a presentation").
|
|
AI Now Remembers Your Past Conversations
Gemini now has automatic memory to recall previous conversations, allowing it to reference your interests and preferences. To address privacy concerns, Google also introduced "Temporary Chats" that aren't saved, and users can disable the memory feature entirely.
Small step: Check your Gemini settings to decide whether you want automatic memory enabled - go to "Personal Context" and toggle "Your past chats with Gemini" based on your privacy preferences.
|
|
|
Industry Developments
|
The Web's New Licensing Standard
Publishers and tech companies including Reddit, Yahoo, and O'Reilly Media have developed Really Simple Licensing (RSL), a new web protocol that lets content creators specify licensing terms for AI crawlers. RSL enables machine-readable licensing that can demand attribution, payment per crawl, or compensation every time AI uses content to generate responses.
|
|
Microsoft Reduces OpenAI Dependence
Microsoft will integrate Anthropic's AI models into Office 365 platforms including Word, Excel, and PowerPoint, marking a strategic shift away from exclusive reliance on OpenAI. The partnership reflects growing tension between Microsoft and OpenAI as both companies pursue competing priorities.
|
|
|
|
