> For the complete documentation index, see [llms.txt](https://connectyai.gitbook.io/documentation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://connectyai.gitbook.io/documentation/usage-guides-enterprise-plan/chat-features/controlled-follow-up.md).

# Controlled Follow-up

Precision editing of steps within the scope of a given question

### What is Controlled Follow-up?

**Controlled Follow-up (CFU)** is a precision-editing feature within **Connecty AI** that lets users modify a specific part of the answer or its answer preparations steps e.g. metric definition. It does NOT change the question's main intent. It is designed for both technical and non-technical users to make changes *within an existing context or intent*, saving time and increasing trust in the modeling process.

CFU is available in two environments:

* **In Chat:** Adjust a specific step in the question-to-answer pipeline, such as the grammar logic or filters.
* **In Metricverse:** Edit a specific entity (e.g. metric, measure, dimension, subject) and see where the change will propagate.

{% hint style="info" %}
The key difference between CFU and FU is that in CFU you can only change part of the **answer**, whereas in FU you can also modify or add additional information to the **question**
{% endhint %}

***

### Why Use CFU?

* 🚀 **Fast Iteration:** Make surgical changes without resetting the full logic pipeline.
* 🧠 **Context-Preserving:** Keeps previous steps intact unless context expansion is explicitly needed.
* 👥 **Accessible for All Users:** Works with natural language or technical input.
* 📊 **Impact-Aware:** Highlights cascading effects on downstream metrics and questions.

***

### How It Works

### **1. In Chat**

* **Step Selection:**\
  Choose a step from the answer preparation flow (e.g. *context inference*, *grammar logic*, *final query*) or click into a specific entity inside the grammar.
* **Make Adjustments:**\
  Use natural language or precise expressions to describe the desired change. For example:
  * *"Replace revenue with profit"*
  * *"Group by product line"*
  * *"Use 'discounted\_price' table instead"*
* **Context Limitation Reminder:**\
  The AI regenerates *only the selected step*, using the context established in earlier steps. If your change requires new tables or dimensions not previously included, you’ll need to use a [**standard follow-up**](https://connectyai.gitbook.io/documentation/usage-guides/chat-features/follow-up-questions) (e.g. click “Reply” or switch to the AI SQL Editor).

### **2. In Metricverse**

* **Entity Selection:**\
  Choose the entity you want to update: metric, measure, dimension, subject, filter, or relationship.
* **Describe the Change:**\
  Input can be:
  * A precise expression (e.g. `SUM(revenue) - SUM(COGS)` for a technical user)
  * A natural language definition (e.g. *“profit should exclude all promotional and logistics costs”* for a business user)
* **Regeneration & Impact Analysis:**\
  AI updates the entity definition and performs **cascading impact analysis**, identifying:
  * Which downstream metrics will be affected
  * Which saved questions, dashboards, or charts may change
  * Whether the logic remains valid

***

### CFU vs [Follow-up](https://connectyai.gitbook.io/documentation/usage-guides/chat-features/follow-up-questions)

| Feature              | Controlled Follow-up (CFU)                        | Standard Follow-up (Reply / Editor)           |
| -------------------- | ------------------------------------------------- | --------------------------------------------- |
| **Scope**            | Edits within existing intent & context            | New or expanded context / intent              |
| **Usage**            | Step-by-step precision edits                      | Full regeneration of logic                    |
| **Risk of Breakage** | Higher if edit requires new tables not in context | Safer for broader changes                     |
| **Best For**         | Quick refinements, deep dives, surgical edits     | Starting over, branching logic, new questions |

***

### Real-World Examples

* **In Chat:**

  > "Change the filter to only include last quarter."\
  > Connecty AI adjusts the grammar step without altering your grouping or selected metrics.
* **In Metricverse:**

  > You select a metric called **Gross Margin** and enter:\
  > *"It should exclude affiliate fees and partner rebates."*\
  > Connecty AI regenerates the metric logic, highlights affected KPIs and dashboards, and explains the cascading changes.