:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
import { CONVERSATION_ID, humanMessage } from './__mocks__/chatResponse.mocks'
import { Meta, StoryObj } from '@storybook/react'
import { useActions, useValues } from 'kea'
import { HttpResponse } from 'msw'
import { useStorybookMocks } from '~/mocks/browser'
import {
AssistantMessage,
AssistantMessageType,
MultiQuestionFormQuestion,
} from '~/queries/schema/schema-assistant-messages'
import { maxGlobalLogic } from './maxGlobalLogic'
import { maxLogic } from './maxLogic'
import { Template, generateChunk, sharedMeta, useAutoSendOnce } from './maxStoriesShared'
import { maxThreadLogic } from './maxThreadLogic'
const meta: Meta = {
title: 'Scenes-App/PostHog AI/Forms',
...sharedMeta,
parameters: {
...sharedMeta.parameters,
testOptions: { waitForSelector: '[data-attr="max-multi-question-form"]' },
},
}
export default meta
type Story = StoryObj<{}>
export const ThreadWithMultiQuestionForm: Story = {
render: () => {
// Multi-question form with several questions - uses tool_calls format
const formQuestions: MultiQuestionFormQuestion[] = [
{
id: 'use_case',
title: 'Use case',
question: 'What is your primary use case for PostHog?',
options: [
{ value: 'Product Analytics', description: 'Track your product metrics and KPIs' },
{
value: 'A/B Testing',
description: 'Test different versions of your product to see which one performs better',
},
{
value: 'Session Replay',
description:
'Record and replay user sessions to understand how they interact with your product',
},
{ value: 'User Surveys', description: 'Collect feedback from your users to improve your product' },
],
allow_custom_answer: true,
},
{
id: 'team_size',
question: 'How large is your team?',
title: 'Team size',
options: [
{ value: 'Just me', description: 'I work alone on my product' },
{ value: '2-10 people', description: 'I have a small team working on my product' },
{ value: '11-50 people', description: 'I have a medium-sized team working on my product' },
{ value: '50+ people', description: 'I have a large team working on my product' },
],
allow_custom_answer: false,
},
{
id: 'experience',
title: 'Experience',
question: 'How familiar are you with analytics tools?',
options: [
{ value: 'Beginner', description: 'I have no experience with analytics tools' },
{ value: 'Intermediate', description: 'I have some experience with analytics tools' },
{ value: 'Expert', description: 'I have a lot of experience with analytics tools' },
],
},
]
const multiQuestionFormMessage: AssistantMessage = {
type: AssistantMessageType.Assistant,
content: 'To help you better, I need to understand your needs. Please answer these quick questions:',
id: 'multi-question-form-msg',
tool_calls: [
{
id: 'create-form-tool-call-1',
name: 'create_form',
args: { questions: formQuestions },
type: 'tool_call',
},
],
}
useStorybookMocks({
post: {
'/api/environments/:team_id/conversations/': () =>
new HttpResponse(
generateChunk([
'event: conversation',
`data: ${JSON.stringify({ id: CONVERSATION_ID })}`,
'event: message',
`data: ${JSON.stringify({
...humanMessage,
content: 'Help me get started with PostHog',
})}`,
'event: message',
`data: ${JSON.stringify(multiQuestionFormMessage)}`,
])
),
},
})
const { setConversationId } = useActions(maxLogic({ panelId: 'storybook' }))
const threadLogic = maxThreadLogic({
conversationId: CONVERSATION_ID,
conversation: null,
panelId: 'storybook',
})
const { askMax } = useActions(threadLogic)
const { dataProcessingAccepted } = useValues(maxGlobalLogic)
useAutoSendOnce(CONVERSATION_ID, dataProcessingAccepted, () => {
setConversationId(CONVERSATION_ID)
askMax('Help me get started with PostHog')
})
if (!dataProcessingAccepted) {
return <></>
}
return <Template />
},
parameters: {
testOptions: {
waitForLoadersToDisappear: false,
},
},
}
export const ThreadWithMultiFieldQuestion: Story = {
render: () => {
const formQuestions: MultiQuestionFormQuestion[] = [
{
id: 'experiment_config',
title: 'Config',
question: 'Configure your experiment',
fields: [
{
id: 'min_sample',
type: 'number',
label: 'Minimum sample size',
min: 100,
max: 100000,
placeholder: 'e.g. 1000',
},
{ id: 'confidence', type: 'slider', label: 'Confidence level (%)', min: 80, max: 99, step: 1 },
{ id: 'notify_on_completion', type: 'toggle', label: 'Notify me when complete' },
],
},
{
id: 'metric_type',
title: 'Metric',
question: 'What type of metric are you testing?',
options: [
{ value: 'Conversion rate', description: 'Percentage of users who complete a goal' },
{ value: 'Revenue per user', description: 'Average revenue generated per user' },
{ value: 'Engagement score', description: 'Composite metric of user activity' },
],
},
]
const multiQuestionFormMessage: AssistantMessage = {
type: AssistantMessageType.Assistant,
content: "Let's set up your experiment. Please configure the settings below:",
id: 'multi-field-form-msg',
tool_calls: [
{
id: 'create-form-multi-field',
name: 'create_form',
args: { questions: formQuestions },
type: 'tool_call',
},
],
}
useStorybookMocks({
post: {
'/api/environments/:team_id/conversations/': () =>
new HttpResponse(
generateChunk([
'event: conversation',
`data: ${JSON.stringify({ id: CONVERSATION_ID })}`,
'event: message',
`data: ${JSON.stringify({
...humanMessage,
content: 'Help me set up an A/B test',
})}`,
'event: message',
`data: ${JSON.stringify(multiQuestionFormMessage)}`,
])
),
},
})
const { setConversationId } = useActions(maxLogic({ panelId: 'storybook' }))
const threadLogic = maxThreadLogic({
conversationId: CONVERSATION_ID,
conversation: null,
panelId: 'storybook',
})
const { askMax } = useActions(threadLogic)
const { dataProcessingAccepted } = useValues(maxGlobalLogic)
useAutoSendOnce(CONVERSATION_ID, dataProcessingAccepted, () => {
setConversationId(CONVERSATION_ID)
askMax('Help me set up an A/B test')
})
if (!dataProcessingAccepted) {
return <></>
}
return <Template />
},
}
export const ThreadWithSingleQuestionForm: Story = {
render: () => {
// Single question form - uses tool_calls format
const formQuestions: MultiQuestionFormQuestion[] = [
{
id: 'data_volume',
title: 'Data volume',
question: 'What is your approximate monthly event volume?',
options: [
{ value: 'Under 1 million events' },
{ value: '1-10 million events' },
{ value: '10-100 million events' },
{ value: 'Over 100 million events' },
],
allow_custom_answer: true,
},
]
const singleQuestionFormMessage: AssistantMessage = {
type: AssistantMessageType.Assistant,
content: 'Before I proceed, I need to know:',
id: 'single-question-form-msg',
tool_calls: [
{
id: 'create-form-tool-call-2',
name: 'create_form',
args: { questions: formQuestions },
type: 'tool_call',
},
],
}
useStorybookMocks({
post: {
'/api/environments/:team_id/conversations/': () =>
new HttpResponse(
generateChunk([
'event: conversation',
`data: ${JSON.stringify({ id: CONVERSATION_ID })}`,
'event: message',
`data: ${JSON.stringify({
...humanMessage,
content: 'What pricing plan should I choose?',
})}`,
'event: message',
`data: ${JSON.stringify(singleQuestionFormMessage)}`,
])
),
},
})
const { setConversationId } = useActions(maxLogic({ panelId: 'storybook' }))
const threadLogic = maxThreadLogic({
conversationId: CONVERSATION_ID,
conversation: null,
panelId: 'storybook',
})
const { askMax } = useActions(threadLogic)
const { dataProcessingAccepted } = useValues(maxGlobalLogic)
useAutoSendOnce(CONVERSATION_ID, dataProcessingAccepted, () => {
setConversationId(CONVERSATION_ID)
askMax('What pricing plan should I choose?')
})
if (!dataProcessingAccepted) {
return <></>
}
return <Template />
},
}
export const ThreadWithMultiQuestionFormLongContent: Story = {
render: () => {
// Form with long questions and answers - agent asking for confirmation before research
const formQuestions = [
{
id: 'research_scope',
title: 'Research scope',
question:
'I found multiple potential areas to investigate regarding your user retention issues. Which aspect should I prioritize in my analysis?',
options: [
{
value: 'Onboarding',
description:
'Onboarding flow completion rates and drop-off points across different user segments',
},
{
value: 'Feature adoption',
description: 'Feature adoption patterns and correlation with long-term retention metrics',
},
{
value: 'User engagement',
description:
'User engagement frequency analysis including session duration and return visit patterns',
},
{
value: 'Cohort',
description: 'Cohort-based comparison of retained vs churned users over the past 6 months',
},
],
allow_custom_answer: true,
},
{
id: 'data_timeframe',
title: 'Data timeframe',
question:
'What time period should I focus on for this analysis? Longer periods provide more data but may include outdated patterns, while shorter periods give more recent insights but with less statistical significance.',
options: [
{ value: 'Last 30 days', description: 'Most recent data, best for identifying current issues' },
{ value: 'Last 90 days', description: 'Good balance of recency and data volume' },
{ value: 'Last 6 months', description: 'Comprehensive view including seasonal variations' },
{ value: 'Last 12 months', description: 'Full year analysis for long-term trend identification' },
],
allow_custom_answer: false,
},
{
id: 'user_segment',
title: 'User segment',
question:
'Should I focus on a specific user segment, or analyze all users? Focusing on a segment can provide more actionable insights for that group, while analyzing all users gives a broader picture.',
options: [
{ value: 'All users', description: 'Comprehensive analysis across the entire user base' },
{ value: 'New users', description: 'Focus on early retention' },
{ value: 'Power users', description: 'Understand what keeps engaged users' },
{ value: 'At-risk users', description: 'Identify churn prevention opportunities' },
],
allow_custom_answer: true,
},
{
id: 'output_format',
title: 'Output format',
question:
'How would you like me to present the findings? I can create different types of deliverables depending on your needs and who will be reviewing the results.',
options: [
{
value: 'Executive summary',
description:
'Executive summary with key findings and recommended actions (best for stakeholder presentations)',
},
{
value: 'Detailed analytical report',
description:
'Detailed analytical report with methodology, data tables, and statistical analysis',
},
{
value: 'Interactive dashboard',
description: 'Interactive dashboard with visualizations that you can explore and filter',
},
{
value: 'Prioritized list of action items',
description:
'Prioritized list of action items with expected impact and implementation complexity',
},
],
allow_custom_answer: true,
},
]
const longContentFormMessage: AssistantMessage = {
type: AssistantMessageType.Assistant,
content:
"I've analyzed your request and identified several areas that need investigation. Before I proceed with the research, I need to confirm a few things to ensure I focus on what matters most to you.",
id: 'long-content-form-msg',
tool_calls: [
{
id: 'create-form-tool-call-3',
name: 'create_form',
args: { questions: formQuestions },
type: 'tool_call',
},
],
}
useStorybookMocks({
post: {
'/api/environments/:team_id/conversations/': () =>
new HttpResponse(
generateChunk([
'event: conversation',
`data: ${JSON.stringify({ id: CONVERSATION_ID })}`,
'event: message',
`data: ${JSON.stringify({
...humanMessage,
content:
'Can you help me understand why our user retention has been declining? I need a comprehensive analysis.',
})}`,
'event: message',
`data: ${JSON.stringify(longContentFormMessage)}`,
])
),
},
})
const { setConversationId } = useActions(maxLogic({ panelId: 'storybook' }))
const threadLogic = maxThreadLogic({
conversationId: CONVERSATION_ID,
conversation: null,
panelId: 'storybook',
})
const { askMax } = useActions(threadLogic)
const { dataProcessingAccepted } = useValues(maxGlobalLogic)
useAutoSendOnce(CONVERSATION_ID, dataProcessingAccepted, () => {
setConversationId(CONVERSATION_ID)
askMax(
'Can you help me understand why our user retention has been declining? I need a comprehensive analysis.'
)
})
if (!dataProcessingAccepted) {
return <></>
}
return <Template />
},
parameters: {
testOptions: {
waitForLoadersToDisappear: false,
},
},
}
export const ThreadWithMultiQuestionFormNoCustomAnswer: Story = {
render: () => {
// Form without custom answer options - uses tool_calls format
const formQuestions: MultiQuestionFormQuestion[] = [
{
id: 'priority',
question: 'What is your top priority right now?',
title: 'Priority',
options: [
{ value: 'Growing user base' },
{ value: 'Improving retention' },
{ value: 'Increasing conversion' },
{ value: 'Boosting engagement' },
],
allow_custom_answer: false,
},
{
id: 'focus_areas',
question: 'Which areas should I cover?',
title: 'Focus',
type: 'multi_select',
options: [{ value: 'Acquisition' }, { value: 'Engagement' }, { value: 'Retention' }],
allow_custom_answer: false,
},
{
id: 'timeline',
question: 'What is your timeline?',
title: 'Timeline',
options: [{ value: 'This week' }, { value: 'This month' }, { value: 'This quarter' }],
allow_custom_answer: false,
},
]
const formMessage: AssistantMessage = {
type: AssistantMessageType.Assistant,
content: 'Please select from the following options:',
id: 'no-custom-form-msg',
tool_calls: [
{
id: 'create-form-tool-call-4',
name: 'create_form',
args: { questions: formQuestions },
type: 'tool_call',
},
],
}
useStorybookMocks({
post: {
'/api/environments/:team_id/conversations/': () =>
new HttpResponse(
generateChunk([
'event: conversation',
`data: ${JSON.stringify({ id: CONVERSATION_ID })}`,
'event: message',
`data: ${JSON.stringify({
...humanMessage,
content: 'Help me prioritize my analytics work',
})}`,
'event: message',
`data: ${JSON.stringify(formMessage)}`,
])
),
},
})
const { setConversationId } = useActions(maxLogic({ panelId: 'storybook' }))
const threadLogic = maxThreadLogic({
conversationId: CONVERSATION_ID,
conversation: null,
panelId: 'storybook',
})
const { askMax } = useActions(threadLogic)
const { dataProcessingAccepted } = useValues(maxGlobalLogic)
useAutoSendOnce(CONVERSATION_ID, dataProcessingAccepted, () => {
setConversationId(CONVERSATION_ID)
askMax('Help me prioritize my analytics work')
})
if (!dataProcessingAccepted) {
return <></>
}
return <Template />
},
parameters: {
testOptions: {
waitForLoadersToDisappear: false,
},
},
}
// Multi-question form stories with new field types
export const ThreadWithMixedFieldTypeForm: Story = {
render: () => {
const formQuestions: MultiQuestionFormQuestion[] = [
{
id: 'goal',
title: 'Goal',
question: 'What is your primary goal for this analysis?',
type: 'select',
options: [
{ value: 'Understand user behavior', description: 'See how users interact with your product' },
{ value: 'Measure conversion', description: 'Track how users move through a funnel' },
{ value: 'Compare segments', description: 'Analyze differences between user groups' },
],
},
{
id: 'features',
title: 'Features',
type: 'multi_select',
question: 'Which analytics features are you interested in?',
options: [
{ value: 'Funnels', description: 'Track conversion through steps' },
{ value: 'Retention', description: 'Measure user stickiness over time' },
{ value: 'Paths', description: 'Visualize user navigation flows' },
{ value: 'Trends', description: 'Monitor metrics over time' },
],
},
{
id: 'config',
title: 'Config',
type: 'multi_field',
question: 'A few more details',
fields: [
{
id: 'team_size',
type: 'slider',
label: 'How many people are on your team?',
min: 1,
max: 100,
step: 1,
},
{ id: 'notify', type: 'toggle', label: 'Would you like weekly email reports?' },
],
},
]
const multiQuestionFormMessage: AssistantMessage = {
type: AssistantMessageType.Assistant,
content: 'Let me tailor your experience. Please answer these quick questions:',
id: 'mixed-form-msg',
tool_calls: [
{
id: 'create-form-mixed-1',
name: 'create_form',
args: { questions: formQuestions },
type: 'tool_call',
},
],
}
useStorybookMocks({
post: {
'/api/environments/:team_id/conversations/': () =>
new HttpResponse(
generateChunk([
'event: conversation',
`data: ${JSON.stringify({ id: CONVERSATION_ID })}`,
'event: message',
`data: ${JSON.stringify({
...humanMessage,
content: 'Help me set up analytics for my team',
})}`,
'event: message',
`data: ${JSON.stringify(multiQuestionFormMessage)}`,
])
),
},
})
const { setConversationId } = useActions(maxLogic({ panelId: 'storybook' }))
const threadLogic = maxThreadLogic({
conversationId: CONVERSATION_ID,
conversation: null,
panelId: 'storybook',
})
const { askMax } = useActions(threadLogic)
const { dataProcessingAccepted } = useValues(maxGlobalLogic)
useAutoSendOnce(CONVERSATION_ID, dataProcessingAccepted, () => {
setConversationId(CONVERSATION_ID)
askMax('Help me set up analytics for my team')
})
if (!dataProcessingAccepted) {
return <></>
}
return <Template />
},
parameters: {
testOptions: {
waitForLoadersToDisappear: false,
},
},
}
export const ThreadWithTextAndNumberForm: Story = {
render: () => {
const formQuestions: MultiQuestionFormQuestion[] = [
{
id: 'project_details',
title: 'Details',
type: 'multi_field',
question: 'Tell me about your project',
fields: [
{ id: 'project_name', type: 'text', label: 'Project name', placeholder: 'e.g. My SaaS App' },
{
id: 'monthly_events',
type: 'number',
label: 'Expected monthly events',
placeholder: 'e.g. 500000',
min: 0,
max: 100000000,
step: 1000,
},
{
id: 'data_source',
type: 'dropdown',
label: 'Primary data source',
options: [
{ value: 'Web app', description: 'JavaScript/React/Vue application' },
{ value: 'Mobile app', description: 'iOS or Android application' },
{ value: 'Backend', description: 'Server-side events via API' },
{ value: 'Third-party', description: 'Import from another analytics tool' },
],
},
],
},
]
const multiQuestionFormMessage: AssistantMessage = {
type: AssistantMessageType.Assistant,
content: "I'd like to learn more about your project to give you the best recommendations:",
id: 'text-number-form-msg',
tool_calls: [
{
id: 'create-form-text-number-1',
name: 'create_form',
args: { questions: formQuestions },
type: 'tool_call',
},
],
}
useStorybookMocks({
post: {
'/api/environments/:team_id/conversations/': () =>
new HttpResponse(
generateChunk([
'event: conversation',
`data: ${JSON.stringify({ id: CONVERSATION_ID })}`,
'event: message',
`data: ${JSON.stringify({
...humanMessage,
content: 'I want to set up event tracking for my project',
})}`,
'event: message',
`data: ${JSON.stringify(multiQuestionFormMessage)}`,
])
),
},
})
const { setConversationId } = useActions(maxLogic({ panelId: 'storybook' }))
const threadLogic = maxThreadLogic({
conversationId: CONVERSATION_ID,
conversation: null,
panelId: 'storybook',
})
const { askMax } = useActions(threadLogic)
const { dataProcessingAccepted } = useValues(maxGlobalLogic)
useAutoSendOnce(CONVERSATION_ID, dataProcessingAccepted, () => {
setConversationId(CONVERSATION_ID)
askMax('I want to set up event tracking for my project')
})
if (!dataProcessingAccepted) {
return <></>
}
return <Template />
},
parameters: {
testOptions: {
waitForLoadersToDisappear: false,
},
},
}
export const ThreadWithSliderForm: Story = {
render: () => {
const formQuestions: MultiQuestionFormQuestion[] = [
{
id: 'experiment_config',
title: 'Config',
type: 'multi_field',
question: 'Configure your experiment parameters',
fields: [
{ id: 'confidence', type: 'slider', label: 'Confidence level (%)', min: 80, max: 99, step: 1 },
{ id: 'duration', type: 'slider', label: 'Duration (days)', min: 7, max: 90, step: 7 },
{ id: 'enable_holdout', type: 'toggle', label: 'Create a holdout group' },
],
},
]
const multiQuestionFormMessage: AssistantMessage = {
type: AssistantMessageType.Assistant,
content: 'Let me configure your experiment parameters:',
id: 'slider-form-msg',
tool_calls: [
{
id: 'create-form-slider-1',
name: 'create_form',
args: { questions: formQuestions },
type: 'tool_call',
},
],
}
useStorybookMocks({
post: {
'/api/environments/:team_id/conversations/': () =>
new HttpResponse(
generateChunk([
'event: conversation',
`data: ${JSON.stringify({ id: CONVERSATION_ID })}`,
'event: message',
`data: ${JSON.stringify({
...humanMessage,
content: 'Help me set up an A/B test experiment',
})}`,
'event: message',
`data: ${JSON.stringify(multiQuestionFormMessage)}`,
])
),
},
})
const { setConversationId } = useActions(maxLogic({ panelId: 'storybook' }))
const threadLogic = maxThreadLogic({
conversationId: CONVERSATION_ID,
conversation: null,
panelId: 'storybook',
})
const { askMax } = useActions(threadLogic)
const { dataProcessingAccepted } = useValues(maxGlobalLogic)
useAutoSendOnce(CONVERSATION_ID, dataProcessingAccepted, () => {
setConversationId(CONVERSATION_ID)
askMax('Help me set up an A/B test experiment')
})
if (!dataProcessingAccepted) {
return <></>
}
return <Template />
},
parameters: {
testOptions: {
waitForLoadersToDisappear: false,
},
},
}