Understand every AI conversation. Measure how customers felt about it.
Your AI agent has conversations every day. Robofy helps you learn from every one. Analyze what customers needed, whether the conversation reached a resolution, where your agent struggled, and how visitors rated the experience, all in one place.
Move beyond chat volume and transcripts. Robofy combines AI-powered conversation intelligence with direct customer feedback, so your team can see the pattern, investigate the session behind it, and make the next conversation better.
AI can explain what happened. Your customer can tell you how it felt.
There are two valuable ways to learn from an AI conversation. The first is what the conversation reveals: the customer’s intent, sentiment, outcome, topics, failure reasons, and recurring patterns. The second is what the customer chooses to say about the experience: a rating and, when they provide it, written feedback.
Neither signal replaces the other. Intelligence helps your team interpret interactions at scale. Feedback brings the customer’s own voice into the review. Together, they make it easier to decide what deserves attention.
Conversation Intelligence
Understand what happened in the session and what keeps happening across your customer conversations.
- What did the customer need?
- Did the conversation appear resolved?
- What question, topic, or failure reason appeared?
- Is the same issue showing up again?
Customer Feedback
Give web-chat visitors a simple way to rate a completed experience and add an optional comment.
- What score did the visitor give the conversation?
- Did they leave written feedback?
- Which session should the team review next?
Intelligence finds the pattern. Customer feedback helps you validate where to look closer.
What is conversation intelligence for AI agents?
Conversation intelligence for AI agents is the use of AI to analyze customer conversations and turn them into structured, reviewable insights. Instead of treating every chat as an isolated transcript, Robofy helps your team understand why customers contacted you, how the conversation appeared to progress, and what your agent may need to improve.
Robofy can generate session-level information such as Primary Intent, Sentiment, Resolution Status, AI Summary, Failure Reason, and Improvement Suggestion. It can also surface insight-level details such as Category, Severity, Customer Question, Insight Text, and Topic. Across conversations, your team can explore recurring themes and patterns.
Conversational AI talks to customers. Conversation Intelligence helps your team learn from those conversations.
A customer may ask about pricing and the AI agent may answer. Conversation Intelligence identifies pricing as the primary intent, helps your team review the sentiment and outcome, and connects that session with similar conversations. Customer feedback adds the visitor's own rating of that experience.
Your AI agent is talking to customers. Can your team see what those conversations are telling you?
When an AI agent handles dozens or thousands of conversations, reading transcripts one by one does not scale. Message counts can tell you that customers are engaging, but they cannot reliably tell you why customers came, whether their needs were met, which questions repeat, or how the experience was received.
- A conversation ends, but you do not know whether the customer got what they needed.
- The same product or support question appears again and again, but the pattern stays hidden in transcripts.
- A customer leaves a poor rating, but the team cannot understand the conversation without finding the session manually.
- Your agent misses a business detail, but there is no clear path from the missed answer to the knowledge you should add.
Robofy connects what happened in the conversation with what the customer chose to say about the experience.
See the signals that matter after the chat ends.
Every completed interaction holds valuable clues. Robofy automatically surfaces key signals so your team can focus on what matters most.
Understand customer intent
See the main reason customers start conversations, whether that is pricing, product information, technical support, complaints, booking, or the business-specific intents your team defines.
Measure conversation outcomes
Review sentiment and resolution status to understand how conversations appear to perform, rather than relying on volume alone.
Hear direct customer feedback
Collect a 1–5 conversation rating in web chat and give visitors the option to leave a written comment when the interaction is ready to close.
Discover knowledge gaps
Find questions and topics where the agent may need better information, clearer instructions, or a more useful workflow.
Identify recurring patterns
Move beyond a single session by grouping similar insights into broader customer themes your team can investigate.
Review the evidence
Open the rated or analyzed conversation when you need to understand the exact exchange behind a dashboard signal.
From AI conversation to customer insight to a better next response.
Robofy turns scattered conversation activity into a repeatable learning loop. Your team can set the context, review what happened, compare direct feedback with intelligence, and improve the agent where the evidence points.
Set what matters
Configure the context that helps Robofy interpret future conversations for a particular AI agent. Define the business-specific Primary Intents and Categories your team cares about, and customize the display labels your team uses for Sentiment, Severity, and Resolution.
Have the conversation
Customers ask questions through your AI agent. The agent handles the interaction while Robofy records the session your team can later review.
Generate intelligence
Robofy structures the available conversation intelligence: what the customer wanted, how the interaction appeared to go, what topics and categories appeared, and where the agent may need improvement.
Collect web-chat feedback
When feedback collection is enabled for web chat, the AI agent can present a rating card after the conversation has naturally reached a useful close. The visitor can choose a 1–5 score and may add an optional comment.
Review and improve
Use the dashboard to identify a trend, a poor rating, an unresolved outcome, or a recurring theme. Open the underlying session, inspect the conversation, and use the evidence to improve your knowledge, instructions, or operational process.
See how Robofy turns a conversation into intelligence your team can use.
A transcript tells you what was said. Structured conversation intelligence helps your team understand what it means. Robofy helps organize the relevant signals from a conversation so you can review the reason, outcome, and next improvement without losing the original customer context.
Intent
The customer asks whether Enterprise supports SSO. Robofy identifies the primary intent as Enterprise Features.
“Does Enterprise support SSO?”
“I don’t have enough information to confirm that.”
Outcome
The agent gives an incomplete answer. The session is available for review as an unresolved outcome.
“Does Enterprise support SSO?”
“I don’t have enough information to confirm that.”
Knowledge gap
The conversation indicates that the agent may need the correct SSO information in its knowledge or instructions.
“Does Enterprise support SSO?”
“I don’t have enough information to confirm that.”
Recurring theme
Related Enterprise SSO questions can become a pattern your team investigates across conversations.
“We need SSO for our Enterprise team.”
“I don’t have enough information to confirm that.”
Customer feedback
The visitor gives the conversation a 1-star rating. This is direct customer input, not a model classification.
“I still need the SSO answer.”
★ ☆ ☆ ☆ ☆ · 1 out of 5
Unresolved Enterprise SSO question · Poor rating · Add SSO information
Give customers a simple way to rate the conversation, then give your team the context to act on it.
Available in web chat when feedback collection is enabled and the AI agent presents the card. WhatsApp conversations do not receive the rating card.
See the customer signal. Understand the conversation behind it.
The Robofy dashboard brings Conversation Rating and Conversational Intelligence into the same workspace. Satisfaction data tells you how visitors rated the experience. Intelligence helps your team understand the reasons, outcomes, questions, and patterns around the conversations worth reviewing.
Conversation Intelligence
Primary Intent Distribution
See which reasons for contacting your AI agent appear most frequently.
Category Distribution
Understand which business-defined categories appear across your customer conversations.
Failure Reasons
Why conversations miss the desired outcome.
Customer Topics
Subjects appearing across analyzed interactions.
Recent Insights
Intent, outcome, sentiment and improvement suggestions.
Recurring Themes
Patterns grouped from similar insights.
A one-star rating is an important signal, but it is not a diagnosis. Open the session, read the exchange, and use its available intelligence to determine what your team should do next.
Find what your AI agent does not know yet.
An AI agent cannot answer every question perfectly on day one. The important question is whether your team can recognize the gaps that matter and improve them before the same issue frustrates more customers.
Robofy helps you investigate unresolved outcomes, failure reasons, customer questions, and recurring topics. When direct feedback is available, a poor rating or a written comment can add urgency and customer context to the pattern your team sees.
Customers repeatedly ask whether Enterprise supports SSO. Several sessions show an unresolved outcome and a missing-knowledge failure reason. One visitor rates the conversation poorly and comments that they still need the answer. Your team now has a clear improvement: add accurate SSO information to the agent’s knowledge and review the updated response.
Turn missing answers into better AI knowledge.
- 1Identify the unanswered or weakly handled question.
- 2Verify the right business information with the responsible team.
- 3Improve the agent’s source knowledge, instructions, or connected workflow, then monitor future conversations.
Understand the conversation. Then understand the pattern.
Session-level review
Open one conversation when you need to see the exact customer question, agent response, available intelligence, and rating or feedback tied to that session.
Cross-conversation learning
Look across many sessions to see which intents, categories, topics, failure reasons, ratings, and themes deserve attention at a team level.
One conversation can tell you what happened. A pattern tells you where to invest.
Configure intelligence and feedback around the job each AI agent does.
A sales agent, support agent, and booking agent should not be measured by the same vocabulary or the same definition of success. Robofy lets you configure Intelligence Settings per AI agent, so your team can provide business context, define preferred Primary Intents and Categories, and choose the display labels it uses for Sentiment, Severity, and Resolution.
For web-chat feedback collection, Rating Settings lets your team enable or disable the rating experience and provide the instruction that tells the agent when it should ask for feedback.
Business Context
Give intelligence the context it needs to interpret conversations for this agent.
Primary Intents and Categories
Guide the classifications your team uses for customer reasons and insight types.
Display Labels
Customize the workspace labels for sentiment, severity, and resolution without changing the underlying classification.
Feedback Collection
Enable or pause the web-chat rating card and guide the agent to ask at a natural close.
- Business context
- Enterprise sales and product questions
- Primary intents
- Pricing · Enterprise Features · Support
- Categories
- Knowledge Gap · Product Interest
- Display labels
- Not Resolved → Open Need
- Feedback collection
- Enabled
- When to ask
- Invite feedback when the conversation has naturally reached a useful close.
- Visitor response
- 1–5 stars and an optional comment
Configure each agent for the conversations it is responsible for.
One conversation dataset. Different questions for every team.
For CX Leaders
Understand contact reasons, satisfaction signals, recurring friction, and experience patterns across the customer journey.
For Customer Support Teams
Find unresolved conversations, negative feedback, failure reasons, and knowledge gaps without reviewing every chat.
For Product Teams
Discover recurring requests, objections, questions, and feedback that reveal unmet needs or unclear product information.
For AI Agent Owners
See where the agent performs well, where customers struggle, and what the next improvement should address.
For Sales Teams
Identify pricing questions, product interest, enterprise requirements, and conversation feedback that affect buying intent.
Explore the details behind every conversation signal.
Aggregate metrics tell you what is happening. Individual session intelligence helps you understand why.
Explore AI analysis and direct customer feedback:
Then open an individual conversation to see intelligence such as Primary Intent, Sentiment, Outcome, AI Summary, Improvement Suggestion, Category, Severity, Customer Question and Extracted Insight.
Hover or focus to pause. Swipe to explore on mobile.
Watch one conversation become a clear next step.
Robofy Agent AI Agent"I don't currently have enough information to confirm whether SSO is available."
AI Agent"I understand SSO is important for your compliance needs. Unfortunately, I don't have the specific Enterprise plan details to confirm this. Would you like me to connect you with our sales team?"
AI Agent"I don't have specific details about Enterprise API access either. I recommend reaching out to our sales team for complete Enterprise plan details including SSO and API capabilities."
Conversation Intelligence
- Primary Intent
- Enterprise Features
- Sentiment
- Neutral
- Outcome
- Not Resolved
- Category
- Knowledge Gap
- Severity
- Medium
- Topic
- Enterprise SSO
- Summary
- Customer asked whether SSO is available for Enterprise accounts, but the AI agent did not have enough knowledge to confirm it.
- Improvement Suggestion
- Add Enterprise authentication and SSO information to the knowledge base.
- Team review
- Review SSO answer: unresolved session with negative feedback.
Now imagine similar questions appearing repeatedly. Robofy's theme intelligence can group related insights so the team sees that Enterprise SSO is not just one customer's question, but a recurring topic worth investigating.
Illustrative conversation, AI analysis, and direct visitor feedback.
2,847 conversations
You know conversations happened. Understanding what they mean requires reading them manually.
- Pricing is one of the most common customer intents
- Knowledge Gap appears repeatedly in product questions
- Unresolved sessions can be filtered immediately
- Similar customer issues become recurring themes
- Improvement suggestions highlight what the AI agent may need next
Different Businesses Need Different Conversation Taxonomies
SaaS
Manufacturing
Hospitality
Financial Services
The intelligence model should reflect the conversations your customers actually have. That is why Robofy allows the configuration to be customized per AI agent rather than forcing a universal list of business intents.
Understand eligible AI conversations across website chat and WhatsApp.
Website Chat
Analyze the conversations your AI agent handles on your website.
Apply the same intelligence approach to customer interactions on WhatsApp.
Robofy supports AI-agent conversations on website chat and WhatsApp through the product’s session-processing pathway. Conversation Intelligence helps your team understand sessions available for analysis across those channels.
Customer ratings are collected in web chat only. The rating card is not shown in WhatsApp conversations.
Frequently Asked Questions
Conversation intelligence and customer feedback, explained.
What is conversation intelligence for AI agents?
Conversation intelligence analyzes interactions between customers and an AI agent and turns them into structured insights. In Robofy, those insights can include primary intent, sentiment, resolution status, summaries, failure reasons, categories, topics, recurring themes, and improvement suggestions.
How is customer feedback different from Conversation Intelligence?
Customer feedback is the visitor’s explicit 1–5 rating and optional written comment. Conversation Intelligence is Robofy’s analysis of the session. They are complementary signals: feedback captures the customer’s direct view, while intelligence helps your team understand the interaction and patterns around it.
What does the Ratings dashboard show?
It shows Average CSAT, total ratings, the number of ratings with and without written feedback, a 1–5 score distribution, and recent ratings for the selected period.
Can I open the conversation behind a rating?
Yes. The Ratings workspace lists recent rated sessions and lets your team open the relevant session to review its conversation.
Where are ratings collected?
The current rating card is collected in web chat when feedback collection is enabled and the AI agent calls the feedback-collection tool. The current Rating Settings surface states that WhatsApp conversations do not receive the card.
Can I control when the agent asks for a rating?
Yes. Rating Settings lets your team enable or disable feedback collection and set the instruction that tells the AI agent when to show the rating card.
What does Robofy analyze in a conversation?
Robofy can generate Primary Intent, Sentiment, Resolution Status, AI Summary, Failure Reason, and Improvement Suggestion. Individual insights can also include Category, Severity, Customer Question, Insight Text, and Topic.
Can I create my own customer intents?
Yes. Each AI agent can have a preferred list of Primary Intents. Robofy uses these values while classifying sessions and can suggest a concise new intent when a conversation does not fit the configured list.
What is the difference between Primary Intent and Category?
Primary Intent is the main reason for the overall conversation. Categories apply to individual insights extracted within the conversation. A session therefore has one primary intent and may contain several insights with different categories.
Can I customize sentiment, severity, and resolution labels?
Yes. Robofy uses underlying classifications while Intelligence Settings lets your team customize the display labels used in the workspace.
Does changing Intelligence Settings update old conversations?
No. Configuration changes apply to future intelligence processing. Conversations that have already been analyzed retain their existing results, and there is currently no manual historical re-analysis function.
Does Conversation Intelligence automatically trigger Human Handoff?
No. Conversation Intelligence and Human Handoff are separate systems. Intelligence can identify signals such as sentiment, severity, and outcome, but those classifications do not automatically initiate a handoff.
Does Conversation Intelligence work with WhatsApp?
Website Chat and WhatsApp sessions currently use the underlying session-processing pathway used for Conversation Intelligence. The current web-chat rating card is separate and is not shown in WhatsApp conversations.
Does Conversation Intelligence consume AI message credits?
The current intelligence-processing pipeline does not deduct Robofy message credits. Credit deduction applies to AI chat messages rather than Conversation Intelligence processing itself.
Build a stronger AI agent operation.
Turn every AI conversation into a better next conversation.
Your customers are already telling you what they need, where the experience falls short, and what your AI agent should learn next. Robofy gives your team the intelligence and direct feedback to listen closely, with the context to improve with confidence.
Start with the conversations you already have. Build the feedback loop your AI agent needs next.