AMA: Atlassian Sr. Director of Product Marketing, Jeremy Hemsworth on AI and Product Marketing

I'll be honest — I'm all over the place. But it's intentional, I swear! There's a maturity curve for AI adoption: glorified spellcheck at one end, a full AI employee at the other. I'm on every rung of it at once, depending on the task: Leadership updates — a glorified spellcheck. They need precision, so I keep the help tight and the judgment mine. Reporting — a Rovo agent surfaces metrics and project status automatically, so I'm not digging through Tableau. Content — Automated workflows that gen ...Read More

This is a hard one. AI-forward companies are shipping faster than anyone can keep up with. Elena Verna at Lovable has talked on a few podcasts about how fast they ship — to the point that people can't absorb all the new capabilities. I feel it weekly: I open Claude or Codex and there's a new feature, or the old one's been renamed. The balance I've landed on: stop treating collateral as one-off assets and start treating it as a system. That starts with a clean source of truth. The launch brief, p ...Read More

There are really two answers here: one for startups, one for big companies. For startups, the change is already underway. "Founding PMMs" are staying solo for longer because they can. I've seen plenty of org charts with "teams of agents," and I think that's largely right — but it's not that new jobs are appearing. It's that instead of a PMM putting on their content hat to write the blog, an agent writes the draft and the PMM reviews it. PMMs have always been orchestrators; now they can put the i ...Read More

I'll answer for PMM, the function I know best. The core human job is deciding which story is true, differentiated, and worth aligning the company around. AI can help with the work around it (e.g., research, ideation, first drafts), but the judgment — which customer truth matters, which tradeoff to make, which narrative actually moves the market — still has to come from a person. AI makes everyone competent because it's great at pattern matching. When every competitor has the same tools, trained ...Read More

Controversial take, and maybe a cynical one: right now, almost every AI capability deserves top-of-funnel billing. I wouldn't hold much back for late-stage. Here's why. AI budget is largely coming out of SaaS budgets — Kyle Poyar published a good piece on this recently, and it matches what I'm seeing. Leaders also have a mandate to adopt AI. I've heard it firsthand ("I have budget to buy an AI tool, any new spend has to have AI, etc."). So as much as I dislike it, if you don't have top billing i ...Read More

Three areas. Learn to build agents and workflows — APIs, MCPs, the basics of how systems connect. The PMMs who stay relevant in the AI era will be technical enough to build software. Not because they have a CS degree, but because AI has dropped the price of entry. I was recently on a call with engineers and worked out a solution with Claude's help — not because I can code, but because I can frame the business problem and the outcome I want. They poked holes in it and it needed more work, but I w ...Read More

I touched on this earlier, but it's worth spelling out. Start with the frontier models — Claude (Cowork or Code), ChatGPT/Codex, and Gemini are non-negotiable. The must-have skill isn't just chatting with them; it's using them to build artifacts, automate research, and prototype workflows. From there, I'm getting real value from tools built for specific jobs: Visuals and decks — Gamma, Reve, and Moda. Voice and audio — ElevenLabs for voice, Descript for podcast and webinar editing, Synthesia for ...Read More