The Huddle: Issue 04

A biweekly look at what's changing in AI, ways to apply it today, resources you didn't know you had, and sharing wins.

Too Good For The Public

The models are getting more capable faster than most teams are learning to use them. Anthropic reminded us of that this week with Claude Mythos, a model so powerful they decided not to sell it to the public. While that’s the headline, there are two other stories worth sharing from the world of AI empowerment that I think may light a fire for our readers.

The model Anthropic won't sell you

Last week Anthropic announced Claude Mythos Preview, their most capable model yet, with a twist. They are not making it generally available. Instead, they handed it to a coalition including AWS, Apple, Microsoft, Google, CrowdStrike, and Palo Alto Networks through an initiative called Project Glasswing, with the mission of finding and patching vulnerabilities in the software the entire internet runs on.

In early internal testing, Mythos autonomously discovered thousands of previously unknown vulnerabilities across every major operating system and web browser. Then yesterday, one day shy of this newsletter going out, Anthropic released Claude Opus 4.7. That's the model you actually get to use. Most of the Mythos gains made it into 4.7 for everyday reasoning, coding, and tool use, with built-in safeguards that detect and block high-risk cybersecurity prompts. Same price as 4.6. Available in Claude products, the API, and the major cloud providers.

So what is a small-to-mid-size business leader supposed to do with all of this?

Probably not much, directly. Mythos isn't for you. Opus 4.7 is a real upgrade you'll feel over time without doing anything different. And if you’re living in the ChatGPT universe, you have a new model coming before the next time we publish a newsletter (fear not).

It’s worth noting - a frontier model just became so capable at one category of work that its maker decided to treat it as a regulated instrument rather than a product. It tells you the competitive dynamics between OpenAI, Google, and Anthropic are starting to wrestle with outputs that are qualitatively different from the consumer-chatbot era we've been living in.

There is also going to be a wave of security patches hitting widely-used software in the coming months as Glasswing partners disclose and fix what Mythos has found. If your business relies on older software, embedded devices, or vendors who are slow to update, this is a nudge to ask whether your patching process can keep up. Your IT lead doesn't need to panic. They just need a current picture of what is running in your environment today (we're talking weeks of work here, not months).

The gap that matters the most is the one between what these models can do and what most teams are doing with them. That gap is getting wider, not narrower. Opus 4.7 is the quiet upgrade in your chat window. Whether your business feels any of that isn't going to come down to the model. It comes down to whether your people are building the muscle to use these tools when it actually matters.

Which is a good segue, because I've been watching that play out up close this month.

CLIENT WIN: LB'S 90 DAYS

A Trailblaze client “LB” is a project manager and integration specialist at a company we've been working with for almost a year. When she started exploring the world of AI, she would have told you (as she told me on our first call) that she had zero technical ability. She was not exaggerating. Formulas in Excel made her nervous. Words like "API" and "oAuth" were gibberish, not concepts.

We started going through our curriculum and she was showing up to 1:1 coaching sessions with questions. Then she started scheduling extra sessions - she was getting curious at a rate faster than her peers. Somewhere in there she moved from asking to building. Ninety days later, she was leading the rebuild of her company's corporate website. Not sitting in meetings about the rebuild, leading it. She was coordinating design, writing the requirements, scoping the phases, and changing what was possible for their web presence.

The best part is she didn’t stop there, she kept building. She had a vision to solve a fleet management problem for 100+ company vehicles. So she rallied the team, asked thoughtful questions, and built a working web application that is scaling across the organization.

Last time we caught up, she told me, "I am a completely different person and problem solver than I was just four months ago." That’s music to our ears, because she is not a flash in the pan.

What made LB unique (but not singular) is that at some quiet moment nobody witnessed, decided she was going to stop watching the wave and learn to surf. We handed her a board and pointed at the water. She did the paddling.

We’ve seen this arc happen with people at every level of technical comfort (and some of our clients have seen it play out four or five times on their own teams by now). What separates the LBs from everyone else isn't aptitude. It's a decision, and it usually doesn't even look like one. It’s asking the dumb question in the 1:1 and coming back next week instead of quitting. It’s believing you can reinvent the way you work.

If you've got someone on your team who is curious and a little stuck, that's the ingredient. We'd love to help you turn it into momentum.

HAVE YOU HAD A HACKATHON YET?

One team we work with cleared their calendar last week and put 12 people in a room for a day, and said "build something that matters." We got a front row seat to be a coach in the room, setting the structure, guiding over shoulders when someone hit a wall, debating which ideas were worth pursuing and which ones were going to step on each other.

They walked out with six live applications ready to deploy. A few examples:

• Freight Photo app. Built for iPad, it lets the team document every angle of a shipment before it leaves the dock door. When a warranty claim lands, they pull up the exact photos from the moment it left their hands and trace it back. Hundreds of shipments will be impacted and a five figure savings is projected.
  ​
• Employee review platform. Built by the leader of their People Team to replace a $5K/year piece of software they used and hated. The new version works the way their review process actually works, not the way a generic vendor assumed it should.
  ​
• Sales analytics dashboard. Distributing weekly performance across 20 markets and 100+ reps, with a gamified layer that's pulling the sales org's attention in a way a spreadsheet never did. The old system was a mess nobody opened. This one the team is checking on their own.

Three of six. I'd show you the others but this is a newsletter and we promise to keep it bite-sized.

Here is what the hackathon format does that a regular AI training session doesn't. It forces people to build in public, debate tradeoffs out loud, and walk out with something they can point at. It collapses the timeline between "I have an idea" and "I have a working tool" into one day. And the social dynamic, watching the person next to you push through the thing you're scared of, is what makes it stick after everyone goes back to their desks.

If you've been looking for a way to turn your team's curiosity into momentum, a build day (or hackathon) is one of the most concentrated ways we've found to do it. If you want to talk about what one could look like at your company, we'd love to have that conversation.

THE BUZZER

I started this issue on Claude Mythos because it's the story everyone is going to send you this week. It absolutely matters because it’s where this is all headed.

But the stories I actually wanted to tell you are LB's and the one about the room full of people (and snacks) that ended with six tools their company didn't have that morning.

The ceiling on what's possible keeps getting higher. What's underrated is how much the floor can rise too, when a team decides to stop watching and start building.

Thanks for huddling with us to start the weekend,

Bryce Stuckenschneider

Founder, Trailblaze Labs

ABOUT TRAILBLAZE LABS

We started Trailblaze Labs because we saw a gap forming between what AI could do and what most businesses knew how to do with it. Turns out, that gap is where we live now. We help business leaders set real AI strategy, train their teams to use these tools with confidence, and build the workflows that actually move the needle. Trailblaze translates AI to the language of your business.

If something in this newsletter sparked a question, gave you an idea, or made you want to push back on something I said, I want to hear about it. Seriously. Reply to this email, connect with me on LinkedIn, or just say hi. I read everything and I love a good friendly argument about where this is all headed.

Let's keep building.