feat: surveys on Max AI #33111
feat: surveys on Max AI #33111
Conversation
|
Size Change: 0 B Total Size: 2.57 MB ℹ️ View Unchanged
|
<
8000
a href="/PostHog/posthog/pull/33111/commits/61e9f5374db3357d3cfe5685ef9e9e3bd09ad3dd" class="Link--secondary">61e9f53
📸 UI snapshots have been updated2 snapshot changes in total. 0 added, 2 modified, 0 deleted:
Triggered by this commit. |
📸 UI snapshots have been updated1 snapshot changes in total. 0 added, 1 modified, 0 deleted:
Triggered by this commit. |
📸 UI snapshots have been updated1 snapshot changes in total. 0 added, 1 modified, 0 deleted:
Triggered by this commit. |
There was a problem hiding this comment.
PR Summary
Implements AI-powered survey creation through Max AI, enabling users to create and launch surveys using natural language chat interactions within PostHog.
- Added
SurveyCreatorToolinproducts/surveys/backend/max_tools.pywith GPT-4 integration for converting natural language to structured survey configurations - Defined comprehensive Pydantic schemas in
products/surveys/backend/survey_schema.pyfor validating LLM output and enforcing survey configuration rules - Added
DEFAULT_SURVEY_APPEARANCEinposthog/constants.pyto maintain consistent styling hierarchy (frontend defaults -> team overrides -> LLM overrides) - Wrapped
Surveyscomponent withSurveysWithMaxToolinfrontend/src/scenes/surveys/Surveys.tsxto provide AI chat interface while preserving existing functionality - Implemented extensive test suite in
products/surveys/backend/test_max_tools.pycovering survey creation, validation, and edge cases
10 files reviewed, 9 comments
Edit PR Review Bot Settings | Greptile
| @@ -319,3 +319,32 @@ class FlagRequestType(StrEnum): | |||
| PERMITTED_FORUM_DOMAINS = ["localhost", "posthog.com"] | |||
|
|
|||
| INVITE_DAYS_VALIDITY = 3 # number of days for which team invites are valid | |||
|
|
|||
| # Sync with frontend/src/scenes/surveys/constants.tsx | |||
| DEFAULT_SURVEY_APPEARANCE = { | |||
There was a problem hiding this comment.
style: Consider alphabetizing the properties in DEFAULT_SURVEY_APPEARANCE for better maintainability
| thankYouMessageHeader: Optional[str] = Field(default=DEFAULT_SURVEY_APPEARANCE["thankYouMessageHeader"]) | ||
| thankYouMessageDescription: Optional[str] = Field(default="We appreciate your feedback.") |
There was a problem hiding this comment.
logic: Default thankYouMessageDescription hardcoded to 'We appreciate your feedback' - should use the DEFAULT_SURVEY_APPEARANCE constant like other fields
| thankYouMessageHeader: Optional[str] = Field(default=DEFAULT_SURVEY_APPEARANCE["thankYouMessageHeader"]) | |
| thankYouMessageDescription: Optional[str] = Field(default="We appreciate your feedback.") | |
| thankYouMessageHeader: Optional[str] = Field(default=DEFAULT_SURVEY_APPEARANCE["thankYouMessageHeader"]) | |
| thankYouMessageDescription: Optional[str] = Field(default=DEFAULT_SURVEY_APPEARANCE["thankYouMessageDescription"]) |
| widgetType: Optional[str] = Field(default=DEFAULT_SURVEY_APPEARANCE["widgetType"]) | ||
| widgetLabel: Optional[str] = Field(default=DEFAULT_SURVEY_APPEARANCE["widgetLabel"]) | ||
| widgetColor: Optional[str] = Field(default=DEFAULT_SURVEY_APPEARANCE["widgetColor"]) | ||
| zIndex: Optional[str] = Field(default=DEFAULT_SURVEY_APPEARANCE["zIndex"]) |
There was a problem hiding this comment.
style: zIndex should be typed as number instead of str to maintain type consistency and prevent invalid values
| ## Current Context | ||
| Team survey configuration: {team_survey_config} | ||
| Existing surveys: {existing_surveys} |
There was a problem hiding this comment.
logic: Inconsistent format in template variables - using both {team_survey_config} and {existing_surveys} at line 45-46, but then only {team_survey_config} again at line 73. Ensure consistent context injection.
| - **PMF**: "How would you feel if you could no longer use [product]?" (Very disappointed/Somewhat disappointed/Not disappointed) | ||
| - **Feedback**: "What could we improve about [feature]?" (open text, optional) | ||
| Current team survey settings: {team_survey_config}""" |
There was a problem hiding this comment.
style: Redundant injection of team_survey_config at end of prompt - already included in 'Current Context' section above. Remove line 73 to avoid confusion.
| # Use the proper serializer for validation and creation | ||
| from posthog.api.survey import SurveySerializerCreateUpdateOnly | ||
| from datetime import datetime |
There was a problem hiding this comment.
style: Move these imports to the top of the file with other imports
| # Use the proper serializer for validation and creation | |
| from posthog.api.survey import SurveySerializerCreateUpdateOnly | |
| from datetime import datetime | |
| from posthog.api.survey import SurveySerializerCreateUpdateOnly | |
| from datetime import datetime |
| if result.should_launch: | ||
| survey_data["start_date"] = datetime.now() |
There was a problem hiding this comment.
logic: Using datetime.now() without timezone awareness could cause issues. Consider using django.utils.timezone.now()
| except Exception: | ||
| return "Unable to load existing surveys" |
There was a problem hiding this comment.
style: Bare except clause is too broad - specify which exceptions you expect here for better error handling
| You are helping create surveys for this PostHog team. | ||
|
|
||
| Current context: | ||
| - Total surveys: {{{total_surveys_count}}} |
There was a problem hiding this comment.
logic: total_surveys_count is referenced in template but never passed to chain.invoke()
Problem
Closes #32818
Changes
Adds AI survey creation to PostHog's Max tool - users can now create and launch surveys through natural language chat.
What's New
create_surveyMaxTool that converts natural language into configured surveysKey Features
Testing
Navigate to Surveys page and try:
Implementation
products/surveys/backend/Part of the Max AI initiative to streamline PostHog workflows through natural language.
Did you write or update any docs for this change?
How did you test this code?
Unit/integration tests for the max too. @kappa90 i'm not that familiar with evals though, maybe we can pair on it and you can show me around? Are they a good fit for this case?