The graveyard of cloud apps
Google has killed over 290 products and services. Not experiments — products that people depended on. Google Reader had millions of active users when it was shut down. Google Stadia promised a gaming revolution and lasted three years. Google Inbox reimagined email, earned devoted fans, and was quietly buried. Each time, users were told to export their data — if an export was even possible — and find somewhere else to go.
Google isn't alone. Evernote was once the default note-taking app for an entire generation of knowledge workers. It had 225 million users at its peak. Then it changed ownership, raised prices, stripped features, and became something its original users barely recognized. People who had built years of notes, research, and personal knowledge inside Evernote found themselves trapped in a degrading product with no clean way out.
Notion, for all its strengths, has had its share of outages — some lasting hours — during which millions of users simply could not access their work. Not their social media. Not their entertainment. Their work. The notes they needed for a meeting happening right now. The project plan they were presenting in twenty minutes. Gone, behind a loading spinner and a status page.
The fundamental problem
The issue isn't that these are bad companies or bad products. The issue is architectural. When your data lives exclusively on someone else's servers, you've made a bet. You've bet that the company will continue to exist, that it will continue to offer the product, that it won't change the pricing to something you can't afford, that it won't be acquired by someone who guts it, and that their servers will be available every single time you need them.
That's a lot of bets. And the historical record shows they don't always pay off.
Cloud-dependent tools create three specific risks:
- Availability risk. You can't access your data when the server is down. If you're on a plane, in a poor-connectivity area, or the service is having an outage, your work is unreachable.
- Continuity risk. The company can shut down, pivot, get acquired, or change terms at any time. Your data is only as durable as the business model behind it.
- Portability risk. Most cloud tools store data in proprietary formats. Even when they offer export, you get a dump of JSON or HTML that no other tool can meaningfully import. Your years of structured thinking become a pile of flat files.
What local-first actually means
Local-first is not the same as offline-only. It doesn't mean going back to the 1990s. It means your data lives on your device first, and syncing to the cloud is optional, additive, and under your control.
In a local-first architecture:
- Your files exist on your hard drive in formats you can read and open with other tools
- The app works fully offline — it doesn't need a server to function
- Sync, when available, is peer-to-peer or end-to-end encrypted — the server never needs to read your data
- If the company disappears tomorrow, your files are still right there on your machine, readable and intact
This isn't a theoretical framework. Obsidian proved that local-first can work at scale — it stores everything as plain Markdown files in a folder you control. The tradeoff is real (collaboration is harder, sync requires more engineering), but the benefit is profound: your data is yours.
How Depli handles your data
Depli stores all workspace data locally on your machine. Your notes, canvases, modules, and project structures live in readable file formats in a directory you can find, open, and back up yourself. No proprietary database you can't inspect. No binary blob that only Depli can parse.
This means several things in practice:
- Depli works offline. No loading spinner. No "reconnecting..." banner. Open the app and your work is there, instantly, because it's already on your machine.
- You can back up your data however you want. Copy the folder to an external drive. Put it in Dropbox. Use git. It's just files.
- If Depli disappears, your work doesn't. We hope that doesn't happen, but the honest truth is: no startup can guarantee its own future. What we can guarantee is that your data doesn't depend on ours.
- Your data stays private. Nothing is sent to a server unless you explicitly choose sync. The AI features in Depli can run with local models, and when cloud models are used, your workspace data isn't stored on our servers.
Open formats are a promise
Using open, readable formats isn't just a technical decision. It's a promise to the people who trust us with their thinking. It says: we believe your work belongs to you, not to us. We believe you should be able to leave at any time, taking everything with you, with zero friction.
This runs counter to standard startup advice. Lock-in is usually considered a feature, not a bug — it's how you retain users and build moats. But we think that logic is backwards. If the only thing keeping people in your product is the difficulty of leaving, you've already lost. People should stay because the tool is genuinely the best place for their work, not because their data is trapped.
Software comes and goes. Your thinking should endure.
The tools we use to think are among the most intimate software we interact with. They hold our half-formed ideas, our research, our plans, our creative work. That data deserves more than a terms-of-service agreement and a prayer that the servers stay on. It deserves to live on your machine, in your control, in formats that will be readable long after any individual app is gone.
That's the bet Depli makes. Not on our own permanence — but on yours.