Overview
What if you had an AI-powered coaching assistant trained on your unique methodologies, available 24/7 to support your clients and prospective leads? Delenta Agent Studio allows you to build customized AI coaching companions. These agents emulate your tone, adhere to your guidelines, draw from your personal materials, and guide clients when you aren't immediately available.
Whether assisting prospective clients on your public profile or supporting active clients inside their portal, your AI agent helps streamline communication, answer questions, and prompt session bookings.
Before You Begin
Ensure you have an active Delenta account with access to Delenta AI features.
Gather your core coaching materials (PDFs, text files, or Markdown documents containing your methodology, FAQs, or program details) to train your agent's knowledge base.
Step-by-Step Instructions
Building and publishing an AI agent in Delenta Agent Studio is a 4-step wizard process: Define, Knowledge, Publish, and Deploy.
Step 1: Access Agent Studio & Define Your Agent
Log in to Delenta and select Delenta AI from the left-hand navigation menu, then click Agent Studio.
Click the + Create Your First Agent (or + Create Agent) button.
Fill out the Define parameters:
Photo: Upload a photo for your agent (optional).
Name: Give your agent a name (e.g., Leadership Coach Bot or your AI persona's name).
Audience: Specify who this agent supports (e.g., Mid-career professionals, Executive leaders).
Purpose: Outline what the agent helps clients do (e.g., Guide clients through leadership principles and answer booking questions).
Voice & Response Length: Choose the preferred tone (e.g., Warm, Encouraging, Direct, Concise) and set response length (Short, Medium, or Detailed).
Persona: Provide clear instructions explaining how the agent should sound and behave (e.g., "You are a calm, direct executive coaching companion. Ask one focused question at a time...").
Guardrails: Explicitly define boundaries and what the agent must not do (e.g., "Do not give legal, medical, or financial advice. Do not negotiate pricing.").
Escalation: Choose when the agent hands off conversations to you (e.g., Coach When Needed, On Client Request).
Click Save & Continue.
Step 2: Add Knowledge Sources
Train your agent using your own coaching resources so it speaks accurately on your behalf.
Under Document Type, select the appropriate category (e.g., Methodology, Program Details).
Upload your source files by dragging and dropping them into the upload area.
Supported File Formats: PDF, Text (.txt), or Markdown (.md) files up to 5MB each.
View your attached files under the Knowledge Sources list.
Click Continue.
Step 3: Review & Publish
Review the summary of your agent's setup, including its Name, Audience, Purpose, Persona, Guardrails, and attached Knowledge Sources.
If everything looks correct, click Publish. Once published, your agent is ready to be deployed.
Step 4: Deploy Your Agent
Decide where your AI agent will interact with users:
Select Deployment Locations:
Public Profile: Interacts with anonymous visitors on your Delenta landing page to answer discovery questions and prompt booking discovery calls.
Client Portal: Assists logged-in active clients inside their portal with program navigation and course support.
Configure Public Profile Settings (If applicable):
Welcome Message: Customize the initial greeting visitors see when opening the chat.
Starter Prompts: Add clickable conversation starter buttons (e.g., "Book a discovery conversation", "Unpack my pipeline pressure").
Channels: Enable Text Chat (Voice chat functionality coming soon).
External Website Embedding (Optional):
If you want to embed the agent widget on your own external website, enable Allow Embedding on My Website.
Add your website domain to Allowed Websites.
Copy the generated Embed Code or Hosted Link and paste it into your website's code.
Click Save Public Settings or Done to finalize.
View Logs & Monitor Performance & Costs
Once your agent is deployed, you can track its activity, performance metrics, actual conversation transcripts, and precise AI resource consumption costs.
Accessing Agent Logs
Go to Agent Studio from your dashboard.
Locate the specific agent in your agents list.
Click the View Logs icon (bar graph icon) in the top-right corner of the agent card.
Understanding Statistics & Usage Costs
At the top of the Agent Logs page, four key metric cards provide real-time visibility into your agent's usage and financial impact:
Runs: The total number of conversation executions completed by the agent. This helps you measure client engagement and activity volume.
Failures: The number of uncompleted or errored runs. Monitoring this helps identify potential issues with agent prompts or knowledge retrieval.
Total Tokens: The aggregate amount of language model processing units consumed by your agent. Tokens reflect the volume of text processed in both visitor prompts and agent responses.
Est. Cost: The estimated dollar amount ($) accrued based on token consumption. This provides clear, granular tracking of how much it costs to run your AI companion, down to fractions of a cent per conversation (e.g., $0.004 per interaction).
Reviewing Conversation Logs & Individual Costs
Below the summary statistics, you can filter and audit individual interactions:
Filter Options: Filter conversations by Status (e.g., All, Success, or Failures).
Conversation List: Displays each interaction thread with visitor type (Website Visitor, Logged-in Client), deployment channel (Public Profile, Website Embed), timestamp, status, and thread ID.
Transcript & Cost Breakdown: Click Details next to any conversation row to open a side-drawer panel showing:
Per-Conversation Cost & Tokens: Exact token count and estimated cost for that specific chat session.
Transcript History: Complete dialogue between the visitor and AI agent, complete with timestamps.
Visitor Feedback: Feedback ratings left by users (e.g., Thumbs Up or Thumbs Down).
Privacy Note: Transcripts are automatically retained for a limited duration (e.g., 7 days) unless deleted earlier by the visitor. Drafts prepared by the agent remain pending until confirmed by the visitor.
Helpful Tips
Start Specific: Keep your persona instructions clear and concise. Define the agent's exact role so it provides direct, structured answers.
Test Thoroughly: Before making your agent public, open your Public Profile link or preview window and test common client questions to ensure responses match your expectations.
Use Quality Source Docs: Clean Markdown (.md) or PDF files formatted with clear headings produce the best and most accurate responses from your agent.
To-Do's & Things to Watch Out For
DO clearly state medical, legal, and financial disclaimers in the Guardrails section.
DO verify that your external website domain is added under Allowed Websites if you are using website widgets.
DON'T upload files containing sensitive private personal data or confidential account login information into the Knowledge base.
DON'T forget that file uploads are limited to PDF, TXT, and MD formats up to 5MB per file. Audio or video files cannot be uploaded directly as knowledge sources.
FAQs & Troubleshooting
Q: Can my AI agent automatically book sessions on my calendar?
A: The agent guides prospective clients through your availability and prompts them with direct links to book discovery calls on your public profile.
Q: What happens if a client asks something outside the knowledge base?
A: According to your defined Guardrails and Escalation settings, the agent will politely inform the user that it cannot answer and suggest contacting the human coach directly.
Q: Can I edit an agent after it has been published?
A: Yes! You can return to Agent Studio at any time to update persona details, upload new knowledge documents, or change deployment channels.