> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rocket.new/llms.txt
> Use this file to discover all available pages before exploring further.

# Tips and tricks

> Power-user tips and common mistakes to avoid when building with Rocket.new. Learn how to get faster results from Solve, Build, and Intelligence. Try free.

export const LlmsDirective = () => <blockquote className="llms-directive">
    For the complete documentation index, see <a href="/llms.txt">llms.txt</a>.
    For a lightweight markdown version of this page, append .md to the URL.
  </blockquote>;

<LlmsDirective />

These tips come from watching thousands of users get the most (and least) out of Rocket. Whether you're working with Solve, Build, or Intelligence, these patterns will help you move faster and avoid common pitfalls.

## General tips

Applicable to every task type in Rocket.

<AccordionGroup>
  <Accordion title="Be specific about what you want" icon="bullseye-arrow">
    Vague prompts produce vague results. The more context you give Rocket, the more targeted the output.

    **Instead of:**

    ```plaintext wrap theme={null}
    Analyze the market.
    ```

    **Try:**

    ```plaintext wrap theme={null}
    Analyze the U.S. meal-kit delivery market for a startup targeting single professionals in urban areas. Focus on market size, top 5 competitors, pricing models, and underserved segments.
    ```

    This applies to Solve, Build, and Intelligence equally. Specificity is the single biggest lever you have.
  </Accordion>

  <Accordion title="Use follow-ups instead of starting over" icon="arrows-rotate">
    Every task in Rocket maintains context. If the first result isn't quite right, follow up in the same task rather than creating a new one. Follow-ups are faster because Rocket already understands what you're working on.

    ```plaintext wrap theme={null}
    That's good, but dig deeper into the pricing comparison. Include a table showing per-seat costs at 10, 25, and 50 users for each competitor.
    ```
  </Accordion>

  <Accordion title="Name your tasks clearly" icon="tag">
    You'll accumulate tasks quickly. Give each one a descriptive name so you can find it later - especially within projects where multiple Solve and Build tasks coexist.

    Good: "Q1 competitor pricing analysis" or "MVP dashboard v2"

    Bad: "Research" or "My app"
  </Accordion>

  <Accordion title="Use projects to stay organized" icon="folder-open">
    If you're working on something with more than two or three tasks, put them in a [project](/getting-started/project/overview). Projects give your tasks shared context, keep files organized, and make it easy to pick up where you left off.
  </Accordion>
</AccordionGroup>

## Solve tips

Get better research outputs with sharper question framing.

<AccordionGroup>
  <Accordion title="Frame questions around decisions" icon="scale-balanced">
    The best Solve prompts are tied to a real decision. Instead of asking Rocket to "research CRM software," tell it what decision the research will inform.

    **Instead of:**

    ```plaintext wrap theme={null}
    Research CRM software.
    ```

    **Try:**

    ```plaintext wrap theme={null}
    We're choosing a CRM for a 20-person sales team selling B2B SaaS. Compare HubSpot, Salesforce, and Pipedrive on pricing, onboarding time, and integration with Slack and Gmail. Recommend the best fit and explain why.
    ```
  </Accordion>

  <Accordion title="Specify your audience and format" icon="users">
    Tell Rocket who will read the report and how they'll use it. This shapes the depth, tone, and structure of the output.

    ```plaintext wrap theme={null}
    This report is for our CEO - keep it to a 2-page executive summary with a recommendation at the top and supporting data below.
    ```

    Or:

    ```plaintext wrap theme={null}
    This is for our product team. Include detailed feature comparisons and link to source material where possible.
    ```
  </Accordion>

  <Accordion title="Break complex research into stages" icon="stairs">
    For large research questions, run multiple Solve tasks in sequence. Start broad, then go deep on the areas that matter most.

    1. **First task:** "Map the competitive landscape for online tutoring platforms in the U.S."
    2. **Second task:** "Deep-dive into Wyzant and Varsity Tutors - compare their tutor acquisition strategies, pricing models, and retention rates."
    3. **Third task:** "Based on the gaps identified, what positioning would work for a new entrant targeting STEM subjects for high schoolers?"
  </Accordion>

  <Accordion title="Ask for specific output formats" icon="table">
    Rocket can structure output in many ways. If you want a comparison table, a SWOT analysis, a ranked list, or a pros/cons breakdown, say so explicitly.

    ```plaintext wrap theme={null}
    Present the comparison as a table with competitors as rows and evaluation criteria as columns. Include a 1-5 rating for each cell with a brief justification.
    ```
  </Accordion>
</AccordionGroup>

## Build tips

Get cleaner apps and faster iteration cycles.

<AccordionGroup>
  <Accordion title="Define structure before features" icon="sitemap">
    Start by telling Rocket the pages/screens, navigation flow, and data model. Then add features one by one. Apps with clear structure from the start are easier to iterate on.

    **Good first prompt:**

    ```plaintext wrap theme={null}
    Build a recipe app with these screens: Home (featured recipes), Search (with filters for cuisine and difficulty), Recipe Detail (ingredients, steps, nutrition), My Recipes (saved favorites), and Profile. Use a bottom tab navigation.
    ```

    Then follow up:

    ```plaintext wrap theme={null}
    Add a shopping list feature. Users should be able to tap "Add to list" on any recipe's ingredient list, and see a combined shopping list on a new tab.
    ```
  </Accordion>

  <Accordion title="Iterate in small steps" icon="shoe-prints">
    Resist the urge to ask for five things at once. Small, focused prompts produce better results and are easier to test.

    **Instead of:**

    ```plaintext wrap theme={null}
    Add authentication, a payment system, a dashboard, email notifications, and an admin panel.
    ```

    **Try:**

    ```plaintext wrap theme={null}
    Add email/password authentication with a login page, signup page, and password reset flow.
    ```

    Ship that, test it, then move to the next feature.
  </Accordion>

  <Accordion title="Use Visual Edit for quick tweaks" icon="paintbrush">
    For cosmetic changes - colors, spacing, text, element order - use [Visual Edit](/build/editor/visual-edit) instead of chat. It's faster and more precise for targeted tweaks.

    Save chat prompts for structural or logic changes where you need Rocket to write new code.
  </Accordion>

  <Accordion title="Use Redesign for existing sites" icon="rotate">
    If you already have a website and want to modernize it, don't start from scratch. Start a Build task with the URL, and Rocket will use the existing site as a foundation. You keep the content and get a fresh design.

    ```plaintext wrap theme={null}
    Redesign this website: https://example.com. Make it modern and minimal with better mobile responsiveness. Keep all existing content but improve the visual hierarchy.
    ```
  </Accordion>

  <Accordion title="Start from templates when possible" icon="layer-plus">
    [Templates](/build/overview) give you a tested structure to build on. Even if the template isn't exactly what you need, starting from one is faster than starting from a blank prompt - and the code structure tends to be cleaner.
  </Accordion>

  <Accordion title="Connect integrations after core functionality works" icon="puzzle-piece">
    Get your app's screens, navigation, and basic logic working before connecting Supabase, Stripe, or other services. This avoids debugging integration issues alongside layout problems.
  </Accordion>
</AccordionGroup>

## Intelligence tips

Configure the Intelligence wizard well and get useful signals from Intelligence.

<AccordionGroup>
  <Accordion title="Write detailed context in the wizard" icon="filter">
    The "Your context" step in the Intelligence wizard is the most important part of setup. Intelligence uses your company description to decide what signals matter and how to frame them.

    **Weak:** "We make project management software."

    **Strong:** "We build project management software for design agencies. Our differentiator is client-facing portals. We're expanding from freelancers into agency teams. I care most about pricing changes, new collaboration features, and any competitors pushing into the agency segment."

    The more specific your context, the more relevant your daily briefs.
  </Accordion>

  <Accordion title="Start with fewer competitors, add more later" icon="crosshairs">
    Adding 20 competitors upfront floods your brief with noise. Start with three to five direct competitors you actively compete with. You can add more competitors from the Intelligence dashboard at any time.

    Each focused monitor is more useful than a broad list.
  </Accordion>

  <Accordion title="Pair Intelligence with Solve for deeper analysis" icon="magnifying-glass">
    When Intelligence surfaces a significant change, don't just read the signal - use it as input for a Solve task. This turns a raw alert into strategic insight.

    ```plaintext wrap theme={null}
    My competitor just dropped their enterprise pricing by 20%. Analyze what this likely means for their strategy and how we should respond.
    ```
  </Accordion>

  <Accordion title="Review signals and adjust frequency regularly" icon="broom">
    After a few weeks, review which signal types are most useful.

    * **Too many low-relevance alerts?** Switch from daily to weekly for certain competitors.
    * **Missing important signals?** Add channels or increase frequency.

    Manage everything from the Intelligence dashboard.
  </Accordion>
</AccordionGroup>

## Cross-task patterns

The real power of Rocket is using Solve, Build, and Intelligence together.

<AccordionGroup>
  <Accordion title="Solve → Build: Research-driven development" icon="arrow-right">
    Use Solve to validate an idea before building it. This pattern saves you from building something nobody needs.

    1. **Solve:** "Analyze the market for AI-powered resume builders. Who are the top players, what do users complain about, and where are the gaps?"
    2. **Review** the report and identify your angle.
    3. **Build:** "Build an AI resume builder that focuses on \[gap identified in research]. Target \[audience]. Include \[specific features that address user complaints]."

    The Build task benefits from the clarity that Solve provided.
  </Accordion>

  <Accordion title="Intelligence → Solve: Signal-driven research" icon="arrow-right">
    Run the Intelligence setup wizard once to set up monitoring, then use Solve to analyze significant signals when they arrive.

    1. **Intelligence wizard:** Set up monitoring for three competitors' pricing pages and product changes.
    2. **Intelligence** surfaces a signal: Competitor A launched a free tier.
    3. **Solve:** "Competitor A just launched a free tier. Analyze the likely impact on our mid-market segment and recommend whether we should respond with a pricing change or a feature differentiation strategy."
  </Accordion>

  <Accordion title="Intelligence → Build: Competitive response" icon="arrow-right">
    Use Intelligence signals to trigger Build tasks that keep your product competitive.

    1. **Intelligence** surfaces: Competitor B added AI-powered search.
    2. **Solve:** Analyze what this means for your roadmap.
    3. **Build:** "Add an AI-powered search feature to my app. Connect to OpenAI and search across all content types. Make it faster and more intuitive than a basic keyword search."
  </Accordion>

  <Accordion title="The full loop: Solve → Build → Intelligence" icon="arrows-spin">
    For new products, the full loop is the most powerful pattern:

    1. **Solve** the market question: What should we build and why?
    2. **Build** the product using those insights.
    3. **Intelligence** - run the wizard once so Intelligence monitors your competitors as you grow.

    Each capability feeds the next, and you stay informed at every stage.
  </Accordion>
</AccordionGroup>

## Common mistakes

Patterns to avoid across all task types.

<AccordionGroup>
  <Accordion title="Starting too broad" icon="circle-exclamation">
    **The mistake:** "Analyze the SaaS market" or "Build me an app" with no further detail.

    **The fix:** Include who it's for, what decision it informs (Solve), what screens and features it needs (Build), or what specific changes to watch for (Intelligence). Rocket performs best with focused, well-scoped prompts.
  </Accordion>

  <Accordion title="Asking for everything at once in Build" icon="circle-exclamation">
    **The mistake:** One massive prompt that requests authentication, payments, dashboards, admin panels, email notifications, and AI features all at the same time.

    **The fix:** Build incrementally. Get the core screens and navigation right first, then add features one at a time. Test each addition before moving on.
  </Accordion>

  <Accordion title="Ignoring follow-ups" icon="circle-exclamation">
    **The mistake:** Creating a new task every time you want a different angle on the same topic.

    **The fix:** Use follow-up messages in the same task. Rocket maintains full context, so follow-ups are faster, cheaper, and more coherent.
  </Accordion>

  <Accordion title="Monitoring too broadly with Intelligence" icon="circle-exclamation">
    **The mistake:** Selecting every channel and adding every competitor you can think of during wizard setup.

    **The fix:** Start with your three to five most direct competitors and the two or three channels most relevant to your strategy. Add more over time as you learn which signals are actually useful. The Intelligence dashboard makes it easy to add competitors and adjust channels at any time.
  </Accordion>

  <Accordion title="Confusing Intelligence and Solve" icon="circle-exclamation">
    **The mistake:** Using Solve for ongoing monitoring (Solve answers one question and stops), or expecting Intelligence to produce research reports.

    **The fix:** Use **Solve** for point-in-time research (questions with answers). Use **Intelligence** for continuous monitoring (watching for changes over time). Intelligence is a one-time workspace setup - run the wizard once and Intelligence handles the rest from the dashboard. If you're not sure, ask yourself: "Do I need an answer now, or do I need to watch for changes?"
  </Accordion>

  <Accordion title="Connecting integrations too early in Build" icon="circle-exclamation">
    **The mistake:** Connecting Supabase or Stripe before the app's core UI and flow are working.

    **The fix:** Get the screens, navigation, and basic logic right first. Then connect external services once you're happy with the structure. This avoids debugging integration issues and layout problems simultaneously.
  </Accordion>
</AccordionGroup>

## What's next?

<CardGroup cols={2}>
  <Card title="Solve best practices" icon="magnifying-glass" href="/solve/best-practices">
    Deep-dive into research strategies and prompt patterns for Solve.
  </Card>

  <Card title="Build best practices" icon="hammer" href="/build/best-practices">
    Detailed guidance on app creation workflows and iteration strategies.
  </Card>

  <Card title="Intelligence best practices" icon="radar" href="/intelligence/best-practices">
    Get the most out of your monitors and the Intelligence dashboard.
  </Card>

  <Card title="Quick start" icon="bolt" href="/getting-started/quick-start">
    Haven't tried Rocket yet? Get your first result in under five minutes.
  </Card>
</CardGroup>