Do's & Don'ts
A practical companion to good Helium use. Keep these close when you're working — they're the habits that separate consistently good output from inconsistent results.
Working with Helium
✓ Do
- ✓Name the outcome, audience, and constraints in your prompt.
- ✓Feed AIM your brand, data, and context — it sharpens every output.
- ✓Verify figures and claims that drive real decisions.
- ✓Ship a rough draft, then refine with follow-up prompts in the same thread.
- ✓State the format up front: "a 10-slide deck", "a single HTML file", "a 3-paragraph summary".
- ✓Give one example of tone or style you want rather than describing it abstractly.
- ✓Read the privacy policy before uploading sensitive or client data.
- ✓Treat the first output as a strong starting point — stay close, challenge it.
✕ Don't
- ✕Paste a one-word topic and expect a finished deliverable.
- ✕Treat the first output as a verified final answer — it's a strong draft.
- ✕Upload confidential client or regulated data without reading the privacy policy first.
- ✕Use Helium to deceive, plagiarise, or misrepresent people.
- ✕Let AI replace the thinking that makes your work yours.
- ✕Describe steps to Helium — describe the outcome instead.
- ✕Skip verifying numbers or claims that will drive real business decisions.
- ✕Accept output that doesn't sound like you — rewrite what needs it.
Prompt quality — vague vs. specific
A good prompt names the outcome, the audience, and the constraints:
✕ Vague
- "Make me a website."
- "Analyse this data."
- "Write something about our product."
✓ Specific
- "Build a one-page site for a wedding photographer, dark theme, with a gallery and a contact form."
- "From this CSV, chart monthly revenue by region and flag the top three outliers."
- "Write a 200-word LinkedIn post for founders, in our confident-but-plain voice."
Using autonomous Bees? There's a separate Do's & Don'ts for deploying agent workflows — see Bees Do's & Don'ts in the Hermes guide.