Agents Studio

Agents Studio allows you to customize Claudia’s behavior through AI Playbooks and Brand Voice settings. These features give you complete control over how Claudia interacts with your customers across different phases of their shopping journey.

Accessing Agents Studio
Navigate to Agent Studio
  1. Log in to your Claudery backend
  2. In the left sidebar, locate Agent Studio (Claudery Agent Studio icon)
  3. Click on Agent Studio to access the main menu
AI Playbooks Configuration

AI Playbooks customize Claudia’s responses based on the customer’s shopping phase (Discovery, Consideration, Decision, Post-Purchase) and sentiment (Neutral, Excited, Confused, Frustrated).


Accessing AI Playbooks
  1. From Agent Studio menu, click on Settings
  2. Navigate to the AI Playbooks tab
  3. Alternatively, go directly to Agent Studio → AI Playbooks if available as a separate menu item
Understanding Playbook Phases

Claudia adapts her behavior based on four shopping phases:

  • Discovery – Customer is browsing and exploring products without commitment
  • Consideration – Customer is comparing products and gathering detailed information
  • Decision – Customer is ready to purchase and asking final questions
  • Post-Purchase – Customer has placed an order and asking about delivery or support


Understanding Playbook Sentiments

Claudia adjusts her tone based on four customer sentiments:

  • Neutral – Professional and friendly standard tone
  • Excited – Enthusiastic and energetic, matching customer’s energy
  • Confused – Step-by-step guidance with clear explanations
  • Frustrated – Immediate empathy and priority support
Configuring Custom Playbooks
  1. In the AI Playbooks settings, you’ll see a grid of 16 combinations (4 phases × 4 sentiments)
  2. Click on any combination to customize the prompt for that specific phase and sentiment
  3. Enter your custom instructions in the text editor
  4. Include specific guidance such as:
    • Product recommendations to emphasize
    • Brand messaging and voice
    • Policies and disclaimers to mention
    • Specific tactics for that phase/sentiment combination
  5. Click Save to apply your custom playbook

Note: Custom playbooks are validated against Brand Voice rules. If your playbook violates any Brand Voice compliance rules (forbidden words, competitor mentions, etc.), you’ll be prompted to adjust it before saving.

Using AI Playbook Suggestions
  1. Click the 💡 Suggestions button in the AI Playbooks page
  2. Claudery will analyze your store and suggest playbook improvements
  3. Review the suggestions and apply those that match your brand
  4. Suggestions are tailored to your store’s industry and product catalog
Understanding Playbook Impact on Claudia

When you customize AI Playbooks:

  • Claudia automatically detects the customer’s shopping phase and sentiment
  • She selects the appropriate playbook prompt based on the detected phase and sentiment
  • Custom playbooks override default prompts, allowing complete customization
  • Changes take effect immediately after saving
  • Playbooks work across all interaction types (product searches, cart operations, general questions)
Brand Voice Configuration

Brand Voice sets absolute compliance rules that Claudia must follow, regardless of phase or sentiment. These rules ensure consistency with your brand guidelines and legal requirements.

Accessing Brand Voice Settings
  1. From Agent Studio menu, click on Brand Voice
  2. You’ll see the Brand Voice configuration page
Enabling Brand Voice
  1. Toggle the Enable Brand Voice Compliance switch at the top
  2. When enabled, all brand voice rules become active
  3. When disabled, Claudia uses standard prompts without brand restrictions
Voice Tone Setting
  1. Select your preferred voice tone from the dropdown:
    • Professional & Formal – Corporate, business-like tone
    • Conversational & Friendly – Casual, approachable tone
    • Luxury & Exclusive – Premium, sophisticated tone
    • Playful & Fun – Lighthearted, energetic tone
  2. This tone is maintained across all phases and sentiments
  3. Voice tone overrides phase-based sentiment adjustments
Forbidden Words
  1. Enter words or phrases that Claudia must never use
  2. Separate multiple words with commas (e.g., cheap, discount, clearance, bargain)
  3. These words are blocked from all responses
  4. Useful for maintaining brand positioning (e.g., avoiding “cheap” for luxury brands)
Competitor Restrictions
  1. List competitor names that Claudia must never mention
  2. Separate multiple competitors with commas (e.g., Nike, Adidas, Amazon)
  3. Prevents Claudia from referencing competitors in product recommendations or comparisons
  4. Helps maintain brand focus and avoid promoting competitors
Legal Restrictions
  1. Enter any legal compliance requirements specific to your industry
  2. Examples:
    • No health claims for supplements
    • FDA disclaimer required for certain products
    • Cannot promise specific results
  3. These rules are enforced in all responses
  4. Critical for regulated industries (healthcare, finance, supplements)
Required Disclaimers
  1. List disclaimers that must be mentioned when relevant
  2. Examples:
    • Free shipping over $50
    • 30-day return policy
    • Gift wrapping available
  3. Claudia will include these disclaimers when appropriate to the conversation
  4. Ensures important policies are communicated to customers
Custom Brand Guidelines
  1. Enter any additional brand-specific instructions
  2. This can include:
    • Brand values and messaging
    • Preferred language and terminology
    • Product positioning guidelines
    • Customer service standards
  3. These guidelines are injected into all prompts
  4. Use this for brand-specific requirements not covered by other fields
Saving Brand Voice Settings
  1. After configuring all fields, click Save Brand Voice
  2. Settings are validated and saved immediately
  3. All future Claudia responses will follow these rules
  4. Changes take effect immediately – no restart required
Testing Brand Voice
  1. Click Test Brand Voice to validate your settings
  2. The system will check for conflicts or issues
  3. Review any warnings or suggestions
  4. Adjust settings as needed
Understanding Brand Voice Impact on Claudia

Brand Voice has the highest priority in prompt generation:

  • Priority 1 (Highest) – Brand Voice compliance rules (absolute, cannot be overridden)
  • Priority 2 – Base AI prompt (Claudia’s identity and core rules)
  • Priority 3 – Sentiment instructions (unless Brand Voice sets fixed tone)
  • Priority 4 – Phase prompts (from AI Playbooks or default)

This means:

  • Brand Voice rules always override other prompts
  • If Brand Voice sets a fixed tone, sentiment-based tone adjustments are ignored
  • Forbidden words are blocked even if they appear in playbook prompts
  • Competitor restrictions apply to all recommendations
  • Legal restrictions are enforced regardless of context
Integration Between AI Playbooks and Brand Voice

AI Playbooks and Brand Voice work together:

  • Brand Voice sets the foundation (absolute rules that cannot be violated)
  • AI Playbooks customize behavior within Brand Voice constraints
  • When saving custom playbooks, the system validates them against Brand Voice rules
  • If a playbook violates Brand Voice, you’ll be prompted to adjust it before saving
  • This ensures all customizations maintain brand compliance
Best Practices
For AI Playbooks
  • Start with default playbooks and customize gradually
  • Test each customization with real customer queries
  • Use AI Suggestions to discover optimization opportunities
  • Focus on phases where you see the most customer interactions
  • Keep playbooks concise but specific
For Brand Voice
  • Set Brand Voice rules before customizing playbooks
  • Be specific with forbidden words (include variations if needed)
  • Regularly review legal restrictions to ensure compliance
  • Test Brand Voice settings after major updates
  • Document custom guidelines for team reference
Troubleshooting
Playbooks Not Taking Effect
  • Verify Brand Voice rules aren’t conflicting with playbook content
  • Check that playbooks were saved successfully
  • Clear any caching if enabled
  • Test with different phase/sentiment combinations
Brand Voice Too Restrictive
  • Review forbidden words list – may be too comprehensive
  • Check if voice tone is preventing necessary tone adjustments
  • Consider adjusting custom guidelines to allow more flexibility
  • Test with sample queries to find the right balance
Validation Errors
  • If playbook validation fails, review the specific violations
  • Check Brand Voice settings to understand restrictions
  • Adjust playbook content to comply with Brand Voice rules
  • Contact support if you need exceptions for specific cases