A conversation Venky had with a product person on our team who's running four projects and can't find time for AI. He gave her — and now us — the actual sequence. Names changed.
Venky brought this back the same week we published the first piece in this corner. It was clean enough that we wanted it on a page, mostly in his words. — Sree
Continuing the same thread — change happening at the grassroots, not top-down — I just had a fabulous discussion with one of our product people. Let's call her Aanya. We were talking about a big launch we're trying to ship faster, and I was helping her with the art of influencing the business team so they move quickly. And, of course, we drifted into her AI journey. She'd already done the Claude Code building sessions with the development team — the dark-mode pairing sessions where you sit shoulder-to-shoulder. But she was very candid:
You know what, Venky, I'm not getting much time to dabble with it. I'm running four projects at the same time right now.
That was the opening I'd been waiting for. So I pivoted into the conversation we keep having about people taking the lead in embracing AI — and what the nudge actually has to be.
Forget Claude Code, for now
Aanya, I said — forget about your Claude Code journey for the time being. Claude Code is always going to live in pair mode. I don't expect you, as a business analyst or a product owner, to be independently capable of spinning up Claude Code, cloning the repo, picking the right model, building a feature. There's a real dependency on the engineering team. One small error, one misconfigured message, and you're halted.
Take the four projects as the challenge
Since you have four projects on your plate right now — why don't you take that as the challenge itself? Same way you did your projects back in college. You never complained to a professor when they asked you to put one together over a weekend. You took it up because you actually enjoyed it. Bring that posture back. If you have four projects, you need to start learning, trying, and doing AI around the work, not inside the work. Because the work has its own demands and its own clock.
One hour a day. Or three hours, twice a weekend.
What I'd suggest. Every day, spend one hour. On a weekend, depending on when you get up — ideally start from six to nine in the morning. Get up at 5:30. Have a good coffee. Sit with three uninterrupted hours from six to nine a.m. Don't do anything beyond that. The rest of the day is completely free — party, family, beach, sleep. Whatever you want.
In that window, this is what I want you to do. Start with Claude Cowork.
What Cowork actually is
Claude Cowork is built by the Claude team for business users and non-tech people. It's different from Claude chat. Claude chat is unorganised, by design — a little disorganised, in fact. You are a function of your prompt and whatever you attach as context. And it's amnesiac: every new chat forgets the last one.
Cowork is the opposite. Cowork is a structured compartment for a specific project. Every session is the project. You have four projects? Create four Claude Cowork projects. Then, slowly, start building context around each one. Attach the right documents. Talk to it in your own voice — orally if you want — and give it the context only you have. Use the right tools — a skill, a brainstorming mode. Strengthen the local folder that lives attached to that project.
The compounding from week one to week four
Now look at what happens. It's a daily practice. Within a week, suddenly the local folder for one project has become superhuman. So capable, so versatile, so much in it. And now the four projects are no longer a challenge. They're a breeze.
You also build a habit. A new piece of work arrives, you spin up a Cowork project for it on day one, and all the past work keeps adding on top. That is how you become AI-first. You move faster, cleaner, better. Your job slowly shifts to being a reviewer — which is, by the way, exactly what a developer becomes inside Claude Code too. So the shape of the work is the same. Only the entry point is different.
It needs a little structure. Cowork gives you the structure. The rest is the daily ritual — try, learn, make mistakes, repeat. The power of compounding sets in from week one to week four. By the end of that, you will be as good as some of the best in the world at doing what you do — with AI on your shoulder.
And then she tells the others
Aanya accepted this whole-heartedly. The next thing I want — and we should make this happen — is to have her tell this story to the product cohort, in her own words. Let it spread sideways, not top-down. The most powerful version of this advice doesn't come from me. It comes from someone in the same chair, six weeks in, showing the others what a Cowork project looks like when it has started to compound.
That, more than anything, is the shift: doing something on your own time, on your own laptop, instead of waiting to spend time during work hours.
— Venky, retold for the corner. Sree typing.