Friday, June 6, 2025
Read time: 5-6 min
Read this article online
Hi, it’s Chad. Every Friday, I serve as your AI guide to help you navigate a rapidly evolving landscape, discern signals from noise and transform cutting-edge insights into practical leadership wisdom. Here’s what you need to know:
1. Algorithmic Musings. Claude’s Artifacts Gallery: Bringing Fire from the Mountain
Claude has increasingly become the AI tool I turn to for working with natural language. I use it quite heavily, which is likely why the good folks at Anthropic sent a surprise my way this week.
Picture this: It’s 7pm the night before I’m scheduled to deliver an AI workshop. I fire up Claude and – boom – there’s a brand-new feature staring back at me. I briefly panicked, thinking I needed to master this feature before my 9am presentation. (I really didn’t. It was just my overclocked brain playing tricks on me.)
The new feature was real, though. Turns out I have access to an advance preview that they’re calling “Artifacts Gallery,” previously known as “Artifacts Studio.” Apparently, being a minor power user has its privileges.
What Are Artifacts, Anyway?
For those who don’t regularly work within Claude, Artifacts are interactive, editable content blocks that appear alongside your conversation. Think of them as workspaces for substantial content like code, documents, creative writing, or visualizations—stuff you might want to reference, modify, or use outside of a chat.
Instead of burying code or long text in the conversation flow, Claude creates these artifacts that you can view, copy, run (for code), or iterate on together. They’re essentially collaborative workspaces for content creation.
My Hot Take: A Step Toward Unleashing Creativity
Claude’s Artifacts Gallery isn’t a great technological leap forward, but it is a substantial step toward unleashing the power and creativity of individual users. Here’s what I mean.
Most users aren’t leveraging frontier AI models to even 10% of their full capabilities. Why? The technology is still so new that both credible use cases and qualified trainers remain insufficient to meet market demands. (Shameless plug: Call me for either need – I’m happy to assist you.)
Claude’s Artifacts Gallery attempts to remedy that problem through a curated collection of examples. On the surface, it’s spiritually akin to ChatGPT’s “Explore GPTs” offering, albeit with more thought given to UX and practicality. There are also community features, but I’ll save that deep dive for another week.
The Real Value: Bringing Fire to the Valley
When you dig deeper, you can see the gallery’s curators have given serious thought to use cases that showcase Claude’s true abilities. In my view, anything that helps bring fire from the mountain of the gods to the valley-dwellers below is a win.
While this may not be quite a Prometheus-style moment, it’s absolutely worth checking out once it’s made available to the entire user base. The gallery doesn’t just show you what you can do – it demonstrates how to think about leveraging AI for substantial, practical work.
Until it reaches general availability, I’ll keep playing with my advance preview. Got questions about what I’m learning? Drop me a line. I’m always up for a good conversation about the evolving landscape of AI tools.
2. Sound Waves: Podcast Highlights
Our latest episode featuring my conversation with Jan Lehman & Rafe Johnson from CTC Productivity dropped this past Monday. Tune in to learn how scattered third-party subscriptions are killing your ROI and how leveraging Copilot may be the right play for your organization. Subscribe for free today on your listening platform of choice to ensure you never miss a beat.
New episodes release every two weeks.
3. Research Roundup: What the Data Tells Us
AI Sentiment Analysis: Your Competitive Edge in Financial Decisions
New research comparing nine AI models just answered a question I get constantly: “Should we trust AI for financial analysis?” The short answer: Yes, but with guardrails.
The numbers that matter: Microsoft Copilot achieved 82% accuracy in analyzing financial sentiment versus 45% for traditional tools like TextBlob. That’s not just an improvement—it’s a competitive advantage.
What this means for your Monday morning: If you’re still relying on basic sentiment tools to analyze earnings calls, customer feedback, or market research, you’re leaving money on the table. AI can spot nuanced language and hedged statements that traditional tools miss entirely.
The catch: Even the best AI models hit an 85% accuracy ceiling and occasionally hallucinate. Your analysts still need to review flagged items, but now they’re focusing on the 15% that actually requires human judgment instead of manually processing everything.
Action item: Ask your finance team what tools they’re using for sentiment analysis. If it’s not AI-powered, that’s a quick win waiting to happen.
Small Models, Big Savings: The SLM Opportunity
Here’s validation for something I’ve been telling clients all year: you don’t always need the biggest, most expensive AI model to get the job done.
Fresh research confirms that small language models (SLMs) can handle 40-70% of typical AI tasks while cutting costs by 10-30x. Think of it like hiring a specialist instead of a generalist—sometimes you just need someone who’s really good at one thing.
Where this works best: Tool calling, code generation, and routine data processing. Tasks where you need reliability and speed, not creativity or complex reasoning.
The business case: One client reduced their AI infrastructure costs by 60% by switching routine customer service responses to SLMs while keeping complex inquiries on larger models. Same quality, fraction of the cost.
Your next move: Audit your current AI spending this month. Map out which tasks actually require frontier models versus those that could run on smaller, cheaper alternatives. My belief is that 50-70% of most organizations’ AI workload could be handled by SLMs.
Vendor reality check: Companies like Anthropic, OpenAI, and others offer smaller models alongside their flagship products. Your current provider probably has options you haven’t explored.
Read our full analysis of each of these research papers at AI for the C Suite.
4. Radar Hits: What’s Worth Your Attention
Phonely’s new AI agents hit 99% accuracy—and customers can’t tell they’re not human. This matters because customer service is often the first place executives see ROI from AI investments. If you’re piloting AI customer support, 99% accuracy should be your benchmark.
Someone built the iPhone AI agent Apple should’ve made by now. The takeaway: if your organization is waiting for tech giants to solve your specific problems, you’re thinking about this wrong. The tools exist today to build custom solutions.
5. Elevate Your Leadership with AI for the C Suite
Ready to turn these insights into action? I’m working with three new clients next quarter on AI strategy and implementation. If you want to move beyond pilots to real business impact, reply with ‘STRATEGY’ and let’s talk.
As we navigate this unprecedented fusion of human and machine intelligence, remember: the best leaders aren’t just adapting to change – they’re actively shaping it. Until next week, keep pushing boundaries.
Chad