Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.hyprcore.ai/llms.txt

Use this file to discover all available pages before exploring further.

When a meeting ends, Hyprcore runs the meeting template against the full transcript and produces:
  • A title (you can rename it).
  • A summary (one or two paragraphs describing what happened).
  • Action items (bulleted list, with owners when the model can identify them).
  • Decisions (key conclusions, when applicable).
Everything renders into the meeting page, where it sits above the transcript and notes. You can edit any of it freely.

​
Default templates

Hyprcore ships several templates. Pick one when you start a meeting, or set a default in Settings → Meetings → Default template.
TemplateBest forWhat it produces
DefaultGeneral meetingsTitle, summary, action items
StandupDaily standupsPer-person updates and blockers
1:1Manager conversationsTopics discussed, follow-ups, agreements
Customer callSales / CS callsCustomer pain points, asks, next steps
InterviewHiring interviewsStrengths, concerns, decision recommendation

​
Custom templates

Open Settings → Meetings → Templates to make your own. A template has:
  • A name and description.
  • A system prompt that defines the voice and structure.
  • A user prompt with {{TRANSCRIPT}} and {{NOTES}} placeholders.
  • The output format—plain markdown, structured sections, or a fixed schema.
Example: a “research interview” template that always produces Quotes, Themes, and Open questions. Or a “podcast” template that produces show notes with timestamps.

​
Re-run a summary

If the first summary doesn’t capture what mattered:
  1. Open the meeting page.
  2. Click Summary → Regenerate.
  3. Optionally pick a different template before regenerating.
The transcript stays the same—only the summary section is replaced. You can do this as many times as you want.

​
Choosing the LLM

Summary quality scales with the LLM. Configure in Settings → Models → Post-processing:
  • Hyprcore Cloud. Easiest. Uses Claude or GPT under the hood depending on the model you pick. Costs credits.
  • Anthropic Claude (Sonnet or Haiku). Best at structured output. Bring your own API key.
  • OpenAI (GPT-4o or 4o-mini). Solid all-around. BYOK.
  • Google Gemini Flash / Pro. Fast and cheap.
  • Apple Intelligence. Free, on-device, smaller models. Good for short meetings.
  • Local Ollama. Fully private. Llama 3, Mistral, or any model you’ve pulled.
For a 30-minute meeting, expect the summary to take 5–30 seconds depending on the provider.

​
Action item tracking

Action items are first-class. After a meeting, they appear in the page and (if you’re signed in) in your Action items view in the dashboard. Check them off as you complete them. They stay portable as Markdown checkboxes, so you can copy them anywhere.