BUSINESSES NEED MORE CAPACITY
Business optimism is continuing to improve, with a couple pretty big asterisks. According to this NFIB survey, 27% of owners named work quality and availability as their single biggest problem. More than one-third said they had job openings they couldn’t fill.
Businesses are hopeful, but at the highest level they don’t have enough people on board and a quarter of people running these companies are worried the team they have isn’t up to snuff.
But here at Trailblaze, we have a unique position, being trusted to work alongside about a dozen businesses (of vastly different sizes and industries). Most businesses we work with aren’t suffering from too many people doing clearly defined jobs… They’re suffering from too many jobs that aren’t clearly defined at all.
The more questions we ask, the more we hear common things like “Nobody officially owns that.” And often, “No system consistently tracks that.” There’s no clear starting gun and no agreed-upon definition of “done.”
It just waits until somebody responsible notices. Usually around 5:03.
AI IS LEAVING THE CHAT WINDOW
I read a tweet this week from a team member at OpenAI that made me laugh (but I also haven’t forgotten)...
Imagine telling someone 5 years ago that one day everyone would have an artificial intelligence in their pocket that could answer almost any question.
They’d ask, “So everyone is a genius?”
And you’d say, “No, they’re idiots.”
For the last few years, most AI has behaved like an incredibly smart person sitting in a conference room waiting for you to walk in and ask a question. It’s a passive genius, and the superpower is knowing what to ask for.
I compare it sometimes to when we finally all got “the internet in our pocket.” The vast majority of factual knowledge in the world is on a rectangle in your pocket - but people spend more time on TikTok or Candy Crush than accessing it. It’s because the power is in the hands of your creativity (or lack thereof). Maybe that’s harsh, but AI is an amplified version of that.
So the major AI companies are trying to move from assistance to execution. OpenAI used those exact terms in a recent report about how companies are putting AI to work (and yes, they have an incentive to get you to use it for work - the more you use it, the more valuable their company is - that part isn’t lost on me).
But I figured it was worth sharing the real ways AI is getting work done now, with a peek behind the curtain at Trailblaze. Hopefully there is inspiration that takes you “out of the chat window” or helps you realize you don’t have to visit the conference room with the genius in it. You can put him in your pocket, bring him alongside your tasks, and worry less when 5:03 hits the clock.
Operations - we built our own CRM to keep track of existing clients, a client pipeline, development projects in-motion, and more. Now, that tool has a dashboard for our weekly team meetings - so our advisors can run down their client list, give high level updates of what’s changed, anywhere they’re stuck, or something they’re excited about. Even better - we connected the CRM to our note taking apps and our Slack, so it autonomously grabs a ton of that so we aren’t hand-keying things in (at least, eliminating 90% of it).
Finance - we’re platform agnostic (equal balance of Claude and ChatGPT for our team). So when connectors or plugins are available on both, we get excited. The QuickBooks plug-in recently has been amazing for us. From my phone, I can check the status of every open invoice, ask questions about which clients pay on time (or late… cough cough). I can ask it to give me estimates on quarterly tax payments, give me a second pair of eyes as we run payroll, or even resolve pesky fees that are creeping into our accounting books.
Sales and Marketing - while we’d love a clear-cut way to run the client acquisition side of things, that’s never the case. Every client is different. They’ve got different executive sponsors, different goals, time horizons, availability in the first 90 days, and more. Some clients want education and enablement work, some want strategy, some want us to build the next great idea… and the majority want all of those things. So using AI to (nearly) automate our proposal process has been a lifesaver. We get the boilerplate stuff blessed by a lawyer, and then have our tool review all discovery call notes to get us to a first draft. The result is often 3x better and 10x faster than doing it the old way.
DON’T START WITH THE TOOL
A caution we give to clients or prospective clients in early conversations is to zoom out before you jump to the most common questions. Because most engagements start with these questions:
- Should we use ChatGPT?
- But our engineers want Claude?
- Copilot is basically included, isn’t that good enough?
- What about agents? Are those different? Do they cost more?
Those are reasonable questions, but asking those is like standing in the power-tool aisle at Home Depot, scratching your head about which drill you should buy before deciding what you’re building.
So we’ve been redirecting those questions with something simple.. Let’s start with the work.
What work isn’t getting done? What’s left at 5:03 when you really want to make it home for dinner on time? If you had a brilliant new hire with unlimited potential, where would you point them? Those questions start to help you pick a tool.
Take something as ordinary as customer follow-up.
“Help with customer follow-up” is not a job.
Here’s a job:
Every Thursday afternoon, review the customers who received proposals but haven’t responded. Check the CRM and email for recent activity. Draft a personalized follow-up for each one. Put the drafts in a review folder. Do not send anything until I’ve had a chance to read, touch them up, and verify everything looks good.
Now we have something an AI can attempt. We also have something a person can inspect, improve, and eventually trust. Only then does it make sense to talk about a tool.
The magic isn’t really the prompt - it’s the clarity of the work you want done.