Unresolved Intents — How to Find and Fix Gaps in Your Agent's Knowledge
Learn what unresolved intents are, how to read the unresolved intents table, and how to use this tab to continuously improve your AI Agent's accuracy.
What are Unresolved Intents?
Every time a customer asks a question that your AI Agent cannot answer with confidence, it gets logged as an unresolved intent. These are the gaps in your agent's knowledge — questions that fell through because the answer was not available in your FAQs, websites, or instructions.
Think of this tab as your agent's feedback report. It tells you exactly what your customers are asking that your agent cannot handle yet.
How to Read the Unresolved Intents Table
Query
The customer's original question shown in bold, with the agent's interpretation of the intent shown below in grey
Occurrences
How many times customers have asked this same question — higher number = more customers hitting this gap
Date Created
When this unresolved intent was first recorded
Actions
A Resolved button (green text) to mark the intent as handled
How to Evaluate an Unresolved Intent
Before taking action, ask yourself one question: is this query relevant to my business?
Not every unresolved question needs an answer. Customers sometimes send random messages, spam, or questions completely unrelated to your business. These should be ignored. Questions that are relevant — things your customers genuinely need help with — should either become an FAQ or an intent.
The Three Actions You Can Take
When you click the green Resolve button on any unresolved intent, you get three options:
Ignore
The question is irrelevant, spam, or not related to your business
Create FAQ
The question is relevant and can be answered with a fixed, static answer
Create Intent Filter
The question is relevant but needs a structured interaction — like collecting data or triggering a journey
Option 1: How to Ignore an Irrelevant Question
Use this when the question has nothing to do with your business or is spam.
Click the green Resolve button on the intent row.
Click Ignore.
The question is dismissed and removed from the list.
Option 2: How to Create a FAQ from an Unresolved Intent
Use this when the question is relevant and can be answered with a simple, direct answer that will be the same every time.
Click the green Resolve button on the intent row.
Click Create FAQ.
The Question field is automatically filled with the customer's original question.
Type the answer in the Answer field.
Click Create FAQ.
The FAQ is added to your agent's knowledge base immediately.
The next time a customer asks the same question, the agent will answer it automatically using the FAQ you just created.
Examples of questions to turn into FAQs: "What are your working hours?", "Do you offer free delivery?", "How do I reset my password?", "What is your return policy?"
Option 3: How to Create an Intent from an Unresolved Intent
Use this when the question is relevant but cannot be answered with a simple static reply. Instead, it requires a structured interaction — like collecting information from the customer, looking up data, or triggering a journey.
For example, if customers frequently ask "I want to book a site visit", a static FAQ answer is not enough. You need the agent to collect the property name, preferred date, and time, and then assign the conversation to a team member. This is where creating an intent makes sense.
Click the green Resolve button on the intent row.
Click Create Intent Filter.
Fill in the following fields:
Intent Name — give it a clear name, for example: Book Site Visit.
Description — describe when this intent should trigger and what it does.
Parameters (optional) — add any information the agent needs to collect from the customer before triggering the journey.
Click Create Intent.
Once created, the intent is converted into a tool and moves to the Tools tab. From there, you can build a journey for it in Bot Studio — just like you would for any Bot Tool (covered in Article 2.3.1).
What Happens After Creating an Intent
The intent appears as a new tool in the Tools tab.
Click the Journey button on that tool's row to open Bot Studio.
Build the journey — add action blocks like Send Message, Assign Agent, Google Sheets, Call API, etc.
Save & Enable the journey.
From this point on, whenever a customer sends a similar message, the AI Agent will recognise the intent, collect the required parameters, and trigger the journey automatically.
When to Create a FAQ vs When to Create an Intent
The answer is the same every time
The response depends on information the customer provides
No data needs to be collected
The agent needs to collect details like name, date, phone number, etc.
It is a simple question with a direct answer
The interaction requires multiple steps — like assigning an agent, saving data, or calling an API
Examples: pricing, policies, working hours, contact details
Examples: booking appointments, registering leads, requesting callbacks, placing orders
If you are unsure, start with a FAQ. You can always convert it to an intent later if you realise a structured interaction is needed.
How Often Should You Check This Tab?
At least once a week. In the first few weeks after going live, you may see many unresolved intents — this is normal. As you keep resolving them by adding FAQs and intents, the number will drop over time.
Why This Tab Matters
Your customers are telling you exactly what they need help with. Every unresolved intent is a real question from a real customer. By resolving them — either as FAQs or intents — you ensure the next customer who asks the same thing gets an instant, accurate response without waiting for a human agent.
This is the single most effective way to improve your agent's accuracy over time.
Summary
In this article, you learned what unresolved intents are, how to evaluate whether a question is relevant or irrelevant, and the three actions you can take: ignoring irrelevant questions, creating FAQs for simple answers, and creating intents for structured interactions that require data collection and automated journeys. Make it a habit to check this tab weekly — it is your most valuable tool for continuous improvement. In the next article, we will cover the Settings tab — how to configure your agent's model, reasoning, fallback behaviour, and voice settings.
Last updated