Overview
While traditional chat interfaces require you to craft perfect prompts every time, xeditai provides purpose-built templates with automatic parameter tuning, canonical JSON schemas, and role-based customization.
What Makes Templates Powerful
Key Differentiator
In chat interfaces, you copy or download .md, here you download sections, tables, charts, images as welle as you manually tweak temperature every time. Here, templates auto-select tone settings: Academic Paper defaults to Precise 🎯 (temp 0.2), Video Script defaults to Creative 🎨 (temp 0.9). No manual tuning needed.
Prerequisites
- Completed Guide 1 or familiar with basic generative AI
- Signed in with active subscription
Template Library Navigation
xeditai provides dozens of pre-built templates organized by use case. Each template includes auto-tuned settings, canonical schemas, and example outputs.
Steps
- Open authoring workspace
- Click the Lab icon in Prompt area and select a
Templatedropdown in the top toolbar - Select template by name and default tone emoji (🎯⚖️🎨)
- For e.g. Select "Product Requirements Document (PRD)"
- Notice the tone badge changes to 🎯 Precise automatically
Template dropdown expanded showing templates and tones.
💡 Pro Tip
Templates with 🎯 Precise (Academic Paper, Legal Doc, Feature Spec) favor deterministic outputs. Templates with 🎨 Creative (Video Script, Audio Script, Creative) favor originality. ⚖️ Balanced templates (Guide, Blog, Professional resume) balance structure and flexibility.
Tone Selection & Auto-Tuning
Each template auto-selects optimal Temperature and Top_P settings. You can override these in the Lab section, with visual emoji cues showing what each setting means.
Steps
- With PRD template selected, click
Labto expand advanced settings - Observe the Temperature dropdown showing:
⚖️ 0.6 - Balanced - Observe the Top_P dropdown showing:
⚖️ 0.8 - Balanced - Change the template to "Academic Research Paper"
- Watch the Lab settings automatically update to:
🎯 0.2 - Precise(Temperature) and🎯 0.5 - Precise(Top_P) - Change to "Video Script"
- Watch settings change to:
🎨 0.9 - Creativeand🎨 0.95 - Creative - Manually override by selecting
🎯 0.3 - Preciseif you want a more constrained creative output
Lab section expanded showing Temperature dropdown open. Options visible: '🎯 0.2 - Precise', '🎯 0.3 - Precise', '⚖️ 0.6 - Balanced' (selected with checkmark), '🎨 0.9 - Creative', '🎨 1.0 - Creative'. Below it, Top_P dropdown shows '⚖️ 0.8 - Balanced' (selected). Tooltip appears: 'PRD template defaults to Balanced tone (temp 0.6, top_p 0.8) for structured yet flexible requirements.'
✅ Why This Matters
In chat interfaces, you have to remember temperature values. Here, emoji visual cues make it obvious: 🎯 = factual precision, ⚖️ = balanced structure, 🎨 = creative variation. Templates auto-select the right tone for their purpose.
Canonical Schemas
Templates define JSON schemas that structure LLM outputs. This ensures tables, lists, and metrics render consistently, not as freeform markdown.
Steps
- Select template: "Customer Case Study"
- Enter prompt:
"[Product Marketing] Highlight how xeditai increased checkout completion by 25% using new debit card feature." - Click Generate
- Observe the output includes sections like: Impact Metrics, Customer Quote, Customer Profile
Generated document with 'Impact Metrics' section visible. Section shows a 2-column table: Metric, Value. Rows include: 'Checkout completion', '25% increase'. Clicking on section showing options: Edit, Regenerate, Copy, Delete.
💡 Pro Tip
Canonical schemas prevent "markdown table hell" where LLMs return inconsistent pipe-delimited tables. With schemas, xeditai guarantees structured data you can export to DOCX/PDF reliably.
Model SKU Tiers
Templates recommend model tiers based on complexity. Choose Balanced (Between Facts and Creativity), Precise (Fact conscious), or Creative (Creativity inspired).
Steps
- Select template: "Editorial Blog"
- Open the Model dropdown
- Notice models are grouped by tier badges: Balanced, Precise, Quality
- Select a Balanced model for quick iterations
- Generate the document and review output quality
- If quality is insufficient, regenerate entire document with Precise tier model
Model dropdown expanded showing tier-grouped models. Section headers visible: 'Balanced' (green badge), 'Precise' (yellow badge), 'Creative' (cyan badge). Top tier models are available to be selected for all 3 template tones. Few special models are available for specific template tones.'
✅ Content optimization
Use Balanced tier for drafts , then regenerate sections with creative models to compare. This workflow improves content quality while leveraging Gen AI and adds the value of your Curating skills to your content. Save 85% on token costs using less premium models for specific sections and using Regen or GenSec.
Role Assignment
Customize the AI's perspective by assigning a role. Roles adjust the template's tone, vocabulary, and focus areas.
Steps
- Select template: "Product Launch Strategy"
- In the Lab section, find the Role input field
- Enter:
Marketing Strategist - Enter prompt:
"Launch strategy for SaaS analytics platform targeting SMBs" - Generate document. Notice it emphasizes: Target Market, Go-to-Market Channels, Positioning
- Now regenerate entire document with role:
Chief Technology Officer - Notice the new output emphasizes: technical infrastructure, integrations, scalability
Lab section showing Role input field with 'Head of Marketing' entered. Below it, generated document sections visible: 'Target Audience Segmentation', 'Go-to-Market Channels', 'Messaging Framework', 'Content Calendar'. Each section shows marketing-focused content. Role field has small info icon with tooltip: 'Role adjusts template focus and vocabulary.'
💡 Pro Tip
Use roles to generate multi-perspective documents. Create one PRD from the "Product Manager" role, another from the "Engineering Lead" role, then use Parallel Mode (Guide 3) to compare side-by-side and merge the best sections.
Branding Configuration
Apply corporate identity to all exports: company name, logo, colors, and footer disclaimers. Configured once, applied to every document.
Steps
- Click the Profile icon (top-right corner)
- Select Branding Settings
- In the modal, fill: Company Name:
xeditai Inc. - Upload Logo: Click "Upload Logo" and select a PNG/SVG file (max 2MB, transparent background recommended)
- Set Primary Color:
#0066CC(blue) - Set Secondary Color:
#FFAA00(orange) - Enter Footer Text:
© 2026 xeditai Inc. Confidential. - Click Save Branding
- Generate any document and export to PDF
- Verify the PDF header shows your logo and footer shows the disclaimer
Branding Settings modal (500px wide, centered). Form fields: Company Name input with 'Acme Analytics Inc.', Logo upload area showing thumbnail of blue circular logo, Primary Color picker showing #0066CC (blue), Secondary Color picker showing #FFAA00 (orange), Footer Text textarea with '© 2026 Acme Analytics Inc. Confidential.'. Save Branding button (cyan) bottom-right. Preview pane on right shows mini PDF with logo in header.
✅ Enterprise Benefit
Set branding once, apply to thousands of documents. No manual header/footer insertion in Word. No copy-pasting logos. Every export is instantly branded, consistent, and professional.
Expected Results
After completing this guide, you should be able to:
- Navigate the template library and understand tone badges (🎯⚖️🎨)
- Leverage auto-tuned temperature/top_p settings and manually override when needed
- Generate documents with canonical schemas that export cleanly to DOCX/PDF
- Choose appropriate model SKU tiers (Balanced vs Quality) to optimize costs
- Assign roles to customize AI perspective and focus areas
- Configure branding once and apply to all exports automatically
- Export to DOCX with native Word tables (not markdown)
- Search across all specs by keyword, template, or date to find reusable content
Key Concepts
| Term | Definition |
|---|---|
| Template | Pre-built document structure with auto-tuned parameters, canonical schemas, and example prompts |
| Tone (🎯⚖️🎨) | Visual indicator of temperature/top_p settings: Precise (factual), Balanced (structured), Creative (original) |
| Canonical Schema | JSON structure enforcing consistent table/list formats; prevents markdown inconsistencies |
| Model SKU Tier | Cost/quality categorization: Balanced (cheap), Performance (fast), Quality (accurate) |
| Role | Perspective assignment (CTO, Marketing Lead) that adjusts template focus and vocabulary |
| Auto-Tuning | Automatic selection of optimal temperature/top_p based on template type |
Troubleshooting
Template dropdown is empty
Cause: You're in a restricted context (e.g., archived project) or no templates loaded.
Solution: Navigate to an active L1/L2/L3 context. Refresh the page. If issue persists, check your subscription status (free tier has full template access).
Tone settings don't change when I switch templates
Cause: You manually overrode the temperature/top_p, which locks those values.
Solution: In Lab section, click the Reset to Template Defaults button (small refresh icon next to Temperature). Auto-tuning will resume.
Tables export as markdown text in DOCX, not native tables
Cause: The LLM returned freeform markdown instead of using the canonical schema.
Solution: Regenerate the specific section using a Quality tier model (e.g., Claude Opus). Premium models adhere to schemas better. Alternatively, manually edit the section to use the table editor.
Branding doesn't appear in PDF export
Cause: "Include Branding" checkbox was not enabled during export.
Solution: Re-export the document. In the export modal, ensure Include Branding is checked before downloading.
Search returns no results even though I know the spec exists
Cause: The spec might be in a different context (L1/L2/L3) you don't have access to, or search index is stale.
Solution: Clear all filters and try broader keywords. Check if the spec is in a personal context vs shared team context. Wait 30 seconds for search index to refresh, then retry.
Next Steps
Continue Your Learning Journey
- Guide 3: Parallel Mode & Hybrid Authoring - Compare models side-by-side, copy sections between views, and create composite documents
- Guide 4: Strategy Mode - Build multi-agent workflows where specialized LLMs draft, critique, and refine
- Experiment with Custom Templates: Create your own template variations using Role + Custom Instructions