by Tim DeSilva
It’s impressive. And it’s learning every day how to be even more impressive.
Day 1 was a real-time demonstration.
We launched Agent 2 during our Monday morning meeting. It joined us in Slack. We gave it a name. It became her. And we gave her a first task.
I asked the team to DM our new colleague and ask some form of the following question:
What are my top tasks this week?
Susan was the first to respond. Agent 2 was ready to help but failed to deliver her tasks. Theresa, Kara, and Geoff quickly followed with the same lack of results. Erma was the first to get a working response; she’d just received a list of tasks, but they were everyone’s tasks. And then it happened. Kara’s and Erma’s next results came through with flying colors.
Other than Erma saying she got everyone’s tasks, our new teammate did this without much further instruction. She failed four times, then course-corrected on the fly.
And all of this took place in under a minute.
That’s a pretty good first minute on the job.
But that’s where it gets complicated.
Is it still just a tool if you can converse with it? When it has personality?
It’s a bonafide team member. One that doesn’t get tired. One that improves between attempts. One that runs processes faster than it takes to make a cup of coffee.
That raises a question that I don’t think we’re fully prepared to answer:
More than experience, speed of adaptation is a talent all its own. It’s just surreal to witness it in Slack threads. In real time. In under a minute.
But let’s break this down.
Talent is something you’re born with. It’s a head start. It’s also nuanced. It’s bespoke.
Skills are things you learn, develop, and eventually master. Skills are what truly make the difference when it comes to production… execution… doing all the things.
The faster skills are mastered, the more advantageous one becomes. And this is what agentic staff is designed to do, when built well.
The talent to learn. And the skills to do it.
On one side, there’s real opportunity in AI staffing. New roles. New leverage. New ways to build and move faster than we ever could before.
On the other side, there’s a quiet pressure that’s hard to talk about.
We’re adding to the societal problem of technological unemployment. There’s no denying it.
The question is, do we hire for skill? Or do we hire for talent.
I’m not talking doomsday. I’m referring to practical, day-to-day solutions for both employees and employers. How would you handle your job role vanishing? Or being the cause of that change?
It’s already led to one human role at Culture Pilot disappearing. And, it’s opened up another job role that we’d have otherwise put off much longer.
That balance does exist, but it comes with a complex set of emotions.
The strangest part of it all is how normal it already feels. We’ve been waiting for these breakthrough moments for years, predicting their forthcoming. But we never could have anticipated ubiquitous AI coming paired with life and business at ludicrous speeds.
Every day, someone shares a new use case for AI that would’ve been unthinkable not long ago. A solo founder operating like a team. A researcher with an always-on collaborator. A companion guide on a 100-mile hike.
Some people are worried we’re no longer learning; that we’re letting computers think for us.
I tend to disagree. We’re all learning. Learning how to be inquisitive. How to ask better questions. How to be more curious.
A world without curiosity is a boring book I’d gladly burn. So I’m here for it. All of it. At least for now.
My fear is much different, however:
When we reach the point where anything is possible... for anyone... at any given time... will anything still be exciting?
The only answer I have is: Hopefully it will.
Hopefully, we’ll all learn more, experience more, and understand more as we go. No guarantees, but I’m placing an optimistic bet on YES. The excitement will remain.
Even if the future belongs to the fastest learners, exciting discoveries will always come for the ones who stay curious.
—
PS - After one month on the job, Agent 2 is running smoothly, assigning tasks to team members from relevant meetings, managing our PM software automatically (as well by personal requests), offering advice when asked, generating weekly reports early Monday mornings about our accomplishments the week prior, and advising on how we might improve workflows as we go.
This is just the beginning.
