do.next
A North Star vision for the next decade of DigitalOcean, told as four vignettes about the people who use it
do.next is the ten-year vision for DigitalOcean. It is a seven-minute video, four customer composites built from real evidence, a leadership playbook, and a phased plan that turns each concept into a shipping bet. Phase one of the new shell is live in the Control Panel, and the plan is what teams ship from.
Every company lives between what they dream of and what they can build today. DigitalOcean closed that gap once by making the cloud feel simple. Then more choices arrived, and what was simple got harder to see. do.next brings that clarity back.
I led design strategy, research synthesis, art direction, and motion direction, with three designers building it alongside me.
Vision video, available for review on request
The seven-minute video is shared privately with recruiters and clients on request. The stills below cover the major beats.
The foundation
Built from four customers.
Before we wrote a frame of the video, we wrote four customers. Each is a composite from real evidence: customer interviews, Ideas Portal, call notes, NPS comments, support patterns. The video is built from short scenes drawn from these customers, snapshots of what they need. Together they show what do.next was built to make possible.
LucentAI is a small AI research lab leaving a hyperscaler. Cost matters, but what they're really looking for is control.
StudioMosaic is a design-engineering studio in Mumbai running thirty client environments. Their growth problem is administrative overhead, and they want the platform to stay invisible.
Kindred Systems is a seven-person AI startup shipping at startup speed. They want speed they can trust: fast and reliable together.
Orion Labs is a well-funded enterprise training big generative models. They have capacity, but what's missing is transparency. Their users aren't asking for more power. They're asking for an explanation.
The starting point
What we had to move past.
The DigitalOcean a developer used to know was a form. Choose an image. Choose a size. Choose a region. Choose options. Add a key. Click create. The form was famously simple, and that simplicity was the company's first growth engine. It is also where the work stopped. The platform waited, patient and unhelpful, for a developer to know what they wanted before they ever asked for help knowing it.
The unlock
Type anything. Launch, scale, ask.
The center of the new platform is one input: a monospace prompt that runs across the top of every page and waits, cursor blinking, for a sentence. Launch a droplet in NYC. Scale my postgres cluster. Search the docs for managed Redis. Investigate the latency on api-gateway-prod. The product no longer asks a developer to remember which tool answers which question. The prompt does the routing.
Vignette one · StudioMosaic
A latency spike at the end of the day.
StudioMosaic is a design-engineering studio in Mumbai whose work keeps their clients' worlds online. Late in the day a Slack alert lands: latency is spiking. Maya and Serena tell the platform to run an investigation, and it answers in plain language. In the control panel the investigation is already waiting, and a quick rollback holds the line for the client while they look deeper. The root cause is obvious once they see it: set the service to autoscale, the change DO recommends. They preview it, test it, and everything stabilizes. The client never knew how close they came to an outage.
Vignette two · Orion Labs
Anjali wanted a report.
Orion Labs trains huge generative models, and every experiment racks up GPU minutes. Anjali, the CFO, needs a clear view of last month's spend and doesn't expect the dashboard to deliver it, since tools like this usually break the moment you need them. It's there, clean and tailored for her, with one thing she didn't ask for: a suggestion from DO about how to spend less. She runs the simulation, sees the new bill at eighteen percent in savings, approves the change, and keeps a 48-hour rollback in her pocket. She thought she wanted a report. She wanted a decision.
Vignette three · StudioMosaic again
Ninety seconds. That's all it took.
StudioMosaic is growing: a new client, a new request, another chance to move fast. Serena opens the Copilot bar. Add monitoring for Client A. The Marketplace slides in right where she's working. The defaults look right. The price looks right. She installs it. A moment later, monitoring is added to all of Client A's projects. It takes ninety seconds, start to finish. Serena walks away feeling like she can take on the whole week.
Vignette four · One conversation
One conversation, across every surface.
Serena opens a support thread in the Control Panel Copilot. Hours later she switches to the CLI. The same conversation is right there waiting. Continue troubleshooting? She does. The dialogue flows through Console, CLI, and Support, uninterrupted, aware of everything that came before. When a DO engineer joins, they see the full trail: logs, commands, outcomes. The issue gets resolved without anyone repeating themselves.
The platform that knows you
Welcome back, by name.
The dashboard greets the developer the way a colleague would. It already read the activity feed; it already knows the migration finished; it already saw the spend trend. The opening line is a paragraph in plain English with the next two actions ready to click.
From vision to rollout
A video, then the work.
A vision is only useful when other people can use it as a compass. The video is the headline; the work is the rollout. We paired the do.next vision with a phased plan that converts each concept into a shipping bet.
The shell and navigation rolled out in phases through the Control Panel: we collapsed the legacy left rail, introduced the universal prompt, layered in outcome-based menus, then added a contextual workspace mode and dark theme. Phase one shipped first; the rest follow on a calendar. I keep a public-facing rebuild of that shell on this site as Seashell UX; it uses the same vocabulary and the same phased thinking, kept generic so anyone can poke at it.
The customer-empathy packet was the pre-read for a Shared Services leadership offsite, with four composites and an evidence ledger, so any product team can trace a concept back to a real customer ask. The product story ran in roadmap reviews, and the shared archetypes gave product, engineering, design, and go-to-market one vocabulary that held when it crossed functions.
A component library and tooling plan under Shared Services Design and UI Engineering keeps the surfaces in the video converging on one shell instead of rebuilding the same patterns locally.
A cloud that listens, learns, and works alongside the people using it.
Outcome
Across LucentAI, Kindred, StudioMosaic, and Orion Labs, the work looked different but the pattern was the same. Each team got back something it had been missing. They all began in the same place we opened with; that tension between what they imagine and what they can actually do today. do.next closes that gap again, so a team can get from the idea to the built thing with far less standing in the way.
Takeaways
- DigitalOcean
- Cloud and Developer Tools
- AI
- Design Leadership
- Long-Term Vision
- Product Strategy
- Customer Research
- Product Vision
- UX Design
- Visual Storytelling
- Motion Direction
- AJ Zichella (Design)
- Soyun Park (Design)
- Isabel Shic (Design)
Up next
