Trent Gillespie
June 25, 2026
On Tuesday, Anthropic launched Claude Tag, a version of Claude that lives inside Slack as a persistent teammate. You tag it in a channel, hand it work, and it picks up where the last person left off. It remembers what's been discussed. It can act on its own when you turn that on. Anthropic says 65 percent of its own product team's code is now written by an internal version of the same tool. Impressive!
I see Tag as a bet on a particular way of using AI: as a coworker inside the place your team already chats, rather than as a separate tool like ChatGPT or Claude that you open in another app. That's not entirely new. Claude has been in Slack for a while. But Tag takes it further with memory, independence, and the ability to reach into your other systems.
I wouldn't recommend most businesses move serious work into it yet. For Tag to actually do work, it has to reach reliably into the systems where the work happens, and that part needs more control and governance. Slack is also a famously loose place. People discuss things there that they wouldn't put anywhere else. Dropping an autonomous AI into that environment, with limited controls today, is a risk most companies aren't ready to manage. And once your team's work lives in Tag, you're depending on two vendors at once, Slack and Anthropic, either of whom can change pricing or features at any time.
I'd watch Tag closely. The approach is interesting and other vendors will copy it. For most businesses, the right move right now is a small pilot in a low-stakes channel, not a serious commitment.
But zooming out, Tag is just one of a myriad of AI launches happening right now. You've heard about both the Claude Fable announcement and following removal from the market. Claude Sonnet 5, OpenAI GPT-5.6, and Google Gemini 3.5 Pro are all expected within days. Perplexity, Slack, Glean, Microsoft and all other SaaS companies are each making a different bet on where AI lives in your business, with continual changes hitting your team. The pace doesn't let up, and the competition for the layer your team works inside is as intense as anything in tech right now.
Few businesses can absorb new capability at this pace, let alone calculate ROI on it reliably. Which has me thinking about a different measure entirely. I've started calling it AI Velocity, and I think it might be the more useful thing to aim at.
The top question I get from every executive I talk with is the same. What's the highest-ROI AI opportunity in my business right now?
At Stellis, we help them find it. We're working hard to make it easier than ever for a leader to identify, prioritize, and capture exactly that.
I've also come to think it's the wrong question.
A few weeks ago Verizon's CEO Dan Schulman told the Bloomberg Tech conference that AI will replace a large percentage of his company's customer service work. He also said AI agents are already producing customer satisfaction scores significantly better than the human teams they're replacing. By the standard scorecard, that's a clean win. Lower cost. Higher satisfaction. ROI delivered.
Step back a second and ask the more important question: why did all those customers need to call Verizon in the first place?
Because the bill was confusing. Because the promotion they signed up for didn't match the statement. Because the plan change they tried in the app failed. Every one of those calls is evidence that something else in the company doesn't work. Automating the call makes the symptom cheaper. I'm not sure it fixes the underlying problem.
A company aiming at ROI optimizes the call, and then hopes to measure and show positive results in 6 months or a year. A company aiming to be more innovative than its competitors asks why the call happened, and rebuilds the upstream process so the customer never has to make it. Both companies can show you a number. Only one of them has actually reinvented itself. That's the move I'm watching for, across every business I work with: not faster versions of the work you already do, but reinvention of what the work is.
At Amazon, we had a saying: velocity is more important than speed. I think that is the perfect way to think about AI right now. Speed is how fast a single thing moves. The market is changing faster than it ever has. Employees can move faster than ever before.
But velocity is the rate of progress in a chosen direction over time. A team can be fast and going nowhere. A team with velocity might look slower in any single moment but the work compounds.
What I see, and recommend, is that you stop worrying about every new model, capability or platform. It makes you think about speed.
That's why I keep coming back to AI Velocity. ROI is a lagging indicator, and a slow one. AI Velocity tells you whether your team is actually moving in the right direction, fast enough to matter. It's also much easier to measure.
I've been working on what the right scorecard looks like for a leader who wants to track this, to start measuring our implementation efforts against. Here's where I've landed, at least for the moment.
If velocity is what you're after, two measures seem to matter more than ROI.
Workflows improved per month. Workflows actually changed and running differently than they were thirty days ago, with a manager's name on each one. If there's no owner, it's a pilot, not a workflow change. This is the velocity number. If it's flat, my read is the company isn't changing.
Active workflow portfolio. The cumulative count of changed workflows currently running. The evidence that your company is actually a different company than it was a year ago.
If you do these, you will get ROI, but more importantly will have reinvented processes, increased the pace of innovation, and be outperforming your competitors who are stuck in the past.
I think the question your CEO should be taking to the board isn't "what's our return on AI." It's "are we changing faster than the companies we compete with."
An AI agent preparing your sales team for every meeting looks like this.
Pulls a one-page brief on every account on tomorrow's calendar, with the company's recent news, financial position, hiring patterns, and competitive moves.
Profiles the people in the meeting from LinkedIn, podcast appearances, and prior conversations with your team.
Surfaces the prospect's likely priorities based on their industry, their role, and what's changed at their company in the last ninety days.
Drafts the discovery questions your best rep would ask, with the supporting talking points and case studies ready to pull up.
Updates the CRM after the call with notes, next steps, and the signals worth tracking on that account going forward.
Salesforce's State of Sales report, which surveys more than 5,000 sales professionals globally, found that reps spend roughly 28 percent of their week actually selling. The rest goes to data entry, internal meetings, and the prep that should already be done before they walk into the room. McKinsey has independently confirmed the same range.
A variety of tools on the market handle parts of this. The companies I see succeeding aren't picking one and standardizing on it. They're connecting several together and treating the integration as the actual work.
Simply: start to measure AI Velocity. Right now, write down the number of workflows you have changed with AI. Not just had someone test, but actually changed, standardized, documented, and owned by a manager. Count them up. That's where you are today.
Tell everyone you are counting and expect to see AI applied to workflows in July. It could be simple, it could be a reinvention.
And then re-measure. That's your velocity.
Hit reply and tell me how many workflows your company has changed in the last 30 days. I'm curious whether anyone is keeping count.
-Trent
Trent Gillespie is CEO of Stellis AI and a keynote speaker helping business leaders understand and operationalize AI in their companies. He spent almost nine years leading global innovation efforts at Amazon before leaving to help other companies build the capabilities they need to compete. Book Trent to speak to your group or book a call to discuss using AI within your business.
Tell us what you thought of today's email. |
Did someone forward this newsletter to you? If you're not already signed up, you can subscribe to AI SPRINT™ for free here.