DeepL CMO Steve Rotter: Marketing Teams Are Underestimating AI
Most teams still treat AI like a handy assistant for single tasks. Steve Rotter thinks that misses the real upside: experimentation at scale. His stance is simple and hard to argue with-use AI to move faster, but keep the human judgment that makes work resonate. Dashboards are helpful; decisions still need people.
From startup garage to language AI
Rotter's path runs through a startup he co-founded that was acquired by Adobe, then Brightcove during the early online video boom, followed by CMO roles across enterprise software. Today he's at DeepL, where the common thread is hypergrowth. He builds structure without slowing momentum, takes products global, and rallies teams around priorities that actually move the needle.
What DeepL offers now
DeepL started in 2017 with translation and now serves 200,000+ mostly B2B customers. The suite includes DeepL Write for refining copy and DeepL Voice for real-time voice translation, which is set to integrate with Zoom meetings. Real-time speech translation is hard-context shifts, acronyms pop up, and grammar differs across languages-yet that's exactly where business conversations live.
In a UK survey, 60% of respondents said AI translation tools improved their international operations. The practical win: clearer communication, fewer mistakes, and faster execution across borders.
PLG roots, enterprise scale
DeepL grew on product-led adoption, then layered on an enterprise motion over the last three years. Rotter's team found "pockets" inside large companies-20 users in one region, 20 in another-without a coordinated rollout. The next phase is straightforward: unify the sprawl, integrate with existing systems, and get thousands more using the tools where they work.
The pro-agentic shift
Rotter sees AI agents moving into the martech stack as digital co-workers. They log into systems, collect and enrich data, and automate workflows end to end. DeepL's trends research found 80% of leaders reporting AI-driven ROI gains (productivity, innovation, revenue), and 42% plan to increase usage next year.
The strategy follows suit: move beyond translation as a feature and deliver workflow-ready AI that fits into current toolchains. Less stitching tools together, more getting real work done.
Team design that actually ships
Rotter runs "flat and fast." Less hierarchy, fewer silos. The mantra: always be challenging. Every other week, someone on the team demos what they built with AI-from campaigns to field tactics to technical workflows. Wins get shared publicly, because almost every win is an orchestra, not a solo.
Creative or data? Neither. Be human.
Rotter rejects the false choice. "There's no such thing as neutral content." Spam your audience with auto-generated fluff and you hurt your brand. Make the work relevant and human first, then let data prove if it worked.
He's blunt about metrics blindness. Moving email click-through from 8% to 10% sounds good-until you admit 90% still ignored the message. Data without judgment invites small bets and smaller thinking.
What marketers get wrong about AI
They underestimate it. AI won't just remove jobs; it will reshape them. The big miss is experimentation. Most teams think in single interactions or single agents. Rotter argues for swarm-style thinking-create content or experiences for many distinct personas, industries, and contexts at once.
Ask AI to build the blog post, then produce 20 credible variants for 20 real roles, with the right jargon and objections. That's how you scale relevance without scaling headcount linearly.
Multilingual customer experience: words matter
Rotter tells a story from Tokyo: a journalist used DeepL on a high-stakes finance story while a colleague used a different tool-and got key details wrong. Accuracy changed careers that day. Not every translation carries that weight, but for contracts, support, or e-commerce, precision isn't optional.
One practical example: in Dutch, "water-resistant" and "waterproof" look similar. Get that wrong on a product page and you won't just deal with returns-you'll deal with trust erosion.
Has AI plateaued?
Rotter doesn't buy it. Compute keeps advancing, multimodal is here, and agent-driven workflows are moving from fun experiments to real work. Soon you won't click through a checklist in your marketing apps-you'll tell your computer to run the play across tools, then review the output.
A myth worth retiring
Rotter says the pendulum swung too far into funnel math. Brand is back at the center. Yes, products are easier to spin up and sites are faster to launch. But durable growth tilts toward brands people trust and remember.
Advice he'd give his younger self
Do time at an agency or inside a B2C brand. At DeepL, consumer DNA now blends with enterprise SaaS rigor-scale and experimentation connected to segmentation, personas, and buying teams. That mix makes better marketers.
How he unwinds
Trail running in the Blue Ridge Mountains. It forces focus-trip once and you're on the ground. More than fitness, it's time with no notifications. Creative work needs empty space; "always on" quickly turns into "always numb."
What to remember about marketing
Move fast and don't fear failure. Fear kills initiative and learning. Things won't always work, but you'll get smarter if you keep shipping.
Playbook: put this to work now
- Run a biweekly AI show-and-tell. Reward useful experiments, not perfect decks.
- Pilot agentic workflows on one process (e.g., lead routing, content QA) and measure time saved.
- Switch from single-output prompts to swarm prompts: 1 asset, 20 persona-specific variants.
- Audit multilingual touchpoints where precision matters most: support macros, product pages, contracts.
- Reframe email success: add a "did this help?" signal and track replies or downstream actions.
- Find "pockets" of adoption inside large accounts and formalize an expansion plan.
- Create a failure budget. If no tests fail this quarter, you're moving too slow.
Level up your team's AI skills
If you lead a marketing org and want structured training, consider this practical path: AI Certification for Marketing Specialists. It helps teams apply AI to real workflows, not just talk about it.
Your membership also unlocks:
