fix: use user's configured name in conversation transcript summaries

[atlas-agent-omi]

Problem

Conversation summaries show "User" instead of the person's actual name when generating the transcript for the summary LLM call.

Root Cause

conversation.get_transcript() and Conversation.get_transcript() were called without user_name in process_conversation.py. The segments_as_string() method defaults to "User" when user_name is None.

Fix

• Add user_name parameter to Conversation.get_transcript() and CreateConversation.get_transcript()
• Fetch user's configured name via get_user_name(uid) in process_conversation()
• Pass user_name through _get_structured() → get_transcript() → segments_as_string()
• Also passes user_name to the app results transcript call

Note: memories.py already passes user_name correctly — this fix aligns the conversation summary path.

Fixes #5319

[greptile-apps]

Greptile Summary

This PR aims to use the user's configured name instead of "User" in conversation transcript summaries. The approach is mostly correct: adding user_name parameter through the call chain from process_conversation() → _get_structured() → get_transcript() → segments_as_string().

Changes:

• ✅ Added user_name parameter to Conversation.get_transcript() and CreateConversation.get_transcript()
• ✅ Modified _get_structured() to accept and pass user_name
• ✅ Fetched user's name via get_user_name(uid) in process_conversation()
• ❌ Critical bug: _trigger_apps() uses user_name in nested function but doesn't accept it as parameter, causing NameError

Impact:
The bug will cause conversation processing to fail when triggering apps (conversation summarization), which is a core feature. The main conversation structuring path (lines 635-638) will work correctly, but app execution (line 696) will crash.

Confidence Score: 1/5

• This PR has a critical runtime bug that will break conversation processing
• Score of 1 reflects a critical logical error that will cause NameError when apps are triggered during conversation processing. While the intent and most of the implementation are correct, the missing user_name parameter in _trigger_apps() makes this PR unsafe to merge without fixes.
• backend/utils/conversations/process_conversation.py requires immediate attention - must add user_name parameter to _trigger_apps() function and its call site

Important Files Changed

Filename	Overview
backend/models/conversation.py	Added user_name parameter to get_transcript() methods in both Conversation and CreateConversation classes - changes are correct
backend/utils/conversations/process_conversation.py	Added user_name to transcript calls but has critical bug: user_name used in _trigger_apps without being passed as parameter, will cause NameError at runtime

_{Last reviewed commit: ddb010f}

[greptile-apps]

user_name undefined here - will cause NameError

user_name is not a parameter of _trigger_apps() nor is it defined in the function scope. When execute_app runs, it will raise NameError: name 'user_name' is not defined.

Fix needed:

1. Add user_name: str = None parameter to _trigger_apps() signature (line 286)
2. Pass user_name=user_name when calling _trigger_apps() (line 696)

def _trigger_apps(
    uid: str,
    conversation: Conversation,
    is_reprocess: bool = False,
    app_id: Optional[str] = None,
    language_code: str = 'en',
    people: List[Person] = None,
    user_name: str = None,  # Add this
):

And at line 696:

_trigger_apps(
    uid, conversation, is_reprocess=is_reprocess, app_id=app_id, 
    language_code=language_code, people=people, user_name=user_name
)

[greptile-apps]

Move import to module level per coding standards

The import is inside the function, violating rule 40d2fa4f (Backend Python import rules - no in-function imports). While this file has other in-function imports, best practice is to import at the module level.

Move to top of file:

Suggested change

	from database.auth import get_user_name
	from database.auth import get_user_name

Context Used: Rule from dashboard - Backend Python import rules - no in-function imports, follow module hierarchy (source)