Creating Your First App
Let's build something! This walkthrough covers everything from opening Vulcan to having a live app you can share.
Step 1 — Open Vulcan
Go to vulcan.shipveho.com. You'll see two options:
- Research first — opens the Research Workspace, where you can clarify what you're building with an AI analyst before writing any code. Good if you're not sure exactly what you want yet. → Research Workspace guide
- Start building — opens the Project Workspace and takes you straight to the AI builder. Good if you already know what you want.
If you're new, click Start building to get into the Project Workspace. You'll see a set of template cards — Kanban Board, Analytics Dashboard, CRM, and a few others. Clicking one is the fastest way to get something live.
Step 2 — Start a new project
In the Project Workspace, click New App or click one of the template cards. You'll see a text box asking what you want to build.
Type a description of your app. The more specific you are, the better the starting point. Compare these:
| Vague | Specific |
|---|---|
| "A task tracker" | "A task tracker for my team where we can add tasks, assign them to people, mark them done, and filter by status" |
| "A dashboard" | "A daily ops dashboard showing open delivery exceptions, grouped by region, with a button to mark each one resolved" |
| "Something for inventory" | "An inventory tracker for our warehouse — items have a name, SKU, quantity on hand, and reorder threshold. Alert me when stock is low." |
Don't stress about being perfect — you can always refine it in the chat after. But a good first prompt saves a few rounds of back-and-forth.
Prompts that work well
Here are some real prompts to use as inspiration or copy-paste starting points:
For ops teams:
"A stop completion tracker where dispatchers can look up a driver by name and see which stops they've completed vs pending today. Let them mark stops complete and add notes."
"A delivery exception log — we can add exceptions with a driver name, route, issue type, and timestamp. Exceptions stay open until someone marks them resolved."
For internal tools:
"An employee directory with search. Each person has a name, team, role, and Slack handle. Admins can add/edit/remove people."
"A simple approval queue — anyone can submit a request with a title and description, and managers can approve or deny it with a note."
For tracking & reporting:
"A daily standup tracker — each team member can log their status (what they did, what they're doing, any blockers) and I can see everyone's entries in one view."
"A budget tracker for our team. We can log expenses with amount, category, and description. Show a running total by category."
Step 3 — Pick a template (optional)
After entering your prompt, Vulcan automatically matches your description against its template library and suggests a starting point if one fits. Templates are pre-built starting points — fully working apps the AI customizes to match what you described. They skip a lot of setup and usually produce better first results than building from scratch.
For example: describe a kanban board and it selects the kanban template. Describe a developer productivity dashboard and it loads a full GitHub + Linear + PagerDuty dashboard template. You'll see the suggested template in the chat before any code is written.
If nothing fits, the AI builds from a minimal default scaffold instead.
Step 4 — Orient yourself
If this is your first time in the App Builder, take a moment to look around. The left panel is the chat. The right panel is the workspace — it starts on the Preview tab, which will show your live app once it's deployed. The toolbar at the top-right has the Deploy to Production and Save Version buttons.
Step 4 — Watch the AI build
The chat panel opens and the AI starts working. It will:
- Ask a few clarifying questions (if it needs to)
- Write the code for your app
- Deploy a live preview automatically
You'll see the preview panel on the right update when the deploy finishes. This usually takes 30–60 seconds.
If the AI asks you questions before writing code — answer them. It's not stalling. A few seconds of clarification saves minutes of fixing later.
Step 5 — Try your app and give feedback
Click around in the preview panel. Pretend you're a real user. Try to break it. The faster you find problems, the easier they are to fix.
When something's off, describe it concretely in the chat:
- "The button color should be blue, not gray"
- "Add a search bar above the table"
- "The table should sort by date by default, newest first"
- "I need a way to delete items — right now I can only add them"
- "When I save the form it clears — I'd like it to stay open so I can add more"
- "The filter doesn't seem to work — when I select 'Pending' it still shows all items"
The AI will update the code and redeploy the preview after each message.
When the output isn't what you expected
The AI doesn't always get it right on the first try. That's normal. A few things that help:
Describe what you expected vs what you got:
"I expected to see a table of all employees, but the page is empty even though I added some. Something's off with the data loading."
Be specific about which part is wrong:
"The form at the top works fine. The problem is the table below — it's not showing any of the items I submitted."
If it looks completely off, say so directly:
"This isn't quite what I had in mind. Can we start over? I want something simpler — just a list with an add button, nothing fancy."
If the AI went in the wrong direction:
"Let's undo that last change and go back to how it was before. Then I want to try a different approach."
→ For more on handling mistakes and bad output, see The Chat Interface — When things go wrong.
Step 6 — Deploy to production
Once you're happy with how things look and work, click Deploy to Production. Your app gets a permanent URL you can bookmark or share with your team.
That's it! Your app is live.
Tips for a great first app
- Don't describe the whole app in one message. The AI works best when it builds one thing at a time. Start with the core feature, check the preview, then add the next thing. A 10-sentence first prompt usually produces worse results than a 2-sentence one followed by follow-up messages. → How to build incrementally
- Expect to iterate. The AI will get things wrong. That's normal — the workflow is designed around it. Preview after every turn, undo if something goes wrong, keep going.
- Describe the problem, not the solution. "I need to track which drivers have completed their stops each day" gives better results than "a tracking app with a table."
- Describe who uses it. "Just me" vs "my whole team of 20 dispatchers" changes how the AI designs the app.
- Check the preview after each change. Catching a broken layout early is much faster than fixing it after 5 more changes on top.
- Save a version before a big change. If you're about to ask for a major redesign, save a version first so you can roll back if it goes sideways.