Skip to main content
These are the same tips shown in onboarding, grouped for easy reference.

Brand foundation tips

  • Lock your core narrative first
    A clear mission, audience, and promise makes every design and campaign decision faster.
  • Differentiate with proof
    Your “what makes us different” section is strongest when backed by concrete outcomes or examples.
  • One promise per page
    Multiple promises dilute trust. Pick the strongest one.
  • Constraints create identity
    Limiting fonts, colors, and tone strengthens memorability.

Chat tips

  • Use specific AI prompts
    Ask for outcomes like: “Rewrite this for social ads in a confident, friendly tone.”
  • Iterate in small passes
    Ask for one change at a time: tone, then structure, then length.
  • Provide constraints
    Tone, length, audience, and goal in one prompt produces sharper results.
  • Refine, don’t restart
    Editing a strong draft is faster than generating from zero.

Editing tips

  • Short beats clever
    Trim long sentences first. Clarity wins in landing pages and ads.
  • Cut filler words
    ”Very,” “really,” and “just” weaken authority. Remove them on the first pass.
  • Lead with outcomes
    Start sentences with results, not process.
  • Break dense blocks
    Short paragraphs increase reading completion.

Strategy tips

  • Define one primary audience
    Start with one audience segment first, then expand.
  • Own one idea
    Brands grow faster when they’re known for one clear thing.
  • Repetition builds recognition
    Saying the same core message consistently compounds over time.
  • Proof reduces friction
    Testimonials, metrics, or case examples shorten decision cycles.

Asset workflow tips

  • Save your best references Mark only high-signal assets as core brand assets.
  • Consistent colors compound Lock your palette early so new assets stay cohesive.
  • Tag by intent Organize assets by use-case (ads, landing, social), not only file type.
  • Archive with discipline Remove low-signal assets so your library stays useful.

Attachment tips

  • Upload reference images when you want the assistant to match a visual style or layout.
  • Add PDFs or docs like brand guidelines, briefs, or competitor materials to give the assistant stronger context.
  • Quality over quantity: One or two highly relevant files produce better results than five loosely related ones.
  • Paste article URLs to repurpose content. The assistant can summarize, rewrite, or turn articles into social posts.
  • Paste social post URLs to adapt existing content for a different platform. The assistant imports the post and rewrites it for your target audience.
Combine link pasting with a specific prompt for best results: “Here is a blog post. Turn it into 3 LinkedIn posts in our brand voice.”

Memory and corrections

  • The assistant remembers your preferences and past corrections within a brand workspace.
  • If the assistant gets your tone wrong, correct it once. It will apply that correction to future outputs.
  • Over time, this means less prompting and more consistent results.

Platform-specific prompting

  • X: Ask for concise posts under 280 characters. Mention if you want threads.
  • Instagram: Specify carousel vs. single image. Ask for hashtag suggestions.
  • LinkedIn: Request a professional tone. Mention if you want a personal story or thought leadership angle.
  • TikTok / YouTube: Ask for hook-first formats. Specify video length if relevant.

Iteration patterns

  • Refine in the same thread when you are adjusting tone, length, or details on existing content. The assistant keeps the full context.
  • Start a new thread when you are switching to a completely different topic or campaign. A fresh thread avoids confusion from prior context.
  • Watch the context limit indicator. Each thread has a circular context meter visible in the chat. As the conversation grows, the circle fills up. When it reaches the limit, start a new thread. The assistant loses accuracy in long threads once context is full. Your brand context and corrections carry over automatically, so you only need to repeat the task-specific details.
When the context limit circle is full, start a new thread. Continuing in a full thread will produce lower-quality responses.