Product teams already using Amplitude
Connect qualitative themes to the users, behaviors, funnels, retention, releases, cohorts, and replay evidence already in the analytics platform.
Independent tool overview
Amplitude AI Feedback (formerly Kraftful) at a glance
Kraftful was acquired by Amplitude in July 2025, and its voice-of-customer technology now powers Amplitude AI Feedback. The current product ingests support tickets, app reviews, calls, social conversations, surveys, files, pasted text, and API-fed feedback; groups it into bugs, complaints, requests, praise, and other themes; and connects what customers say with Amplitude behavior, cohorts, session replay, experiments, guides, and surveys. AI Feedback is available across Amplitude plans, including a limited free allowance, with larger analysis volumes sold for Growth and Enterprise. It can accelerate discovery, but generated themes, severity, counts, user identity, and proposed actions still need source-level and human validation.
Visit the official Amplitude AI Feedback (formerly Kraftful) site ↗
- Current product
- Amplitude AI Feedback
- Former product
- Kraftful, acquired by Amplitude on July 10, 2025
- Current launch
- Fully integrated AI Feedback released November 12, 2025
- Product type
- AI voice-of-customer analysis integrated with product analytics
- Inputs
- Support, sales, reviews, social, community, surveys, files, pasted text, Amplitude chats, and HTTP API
- Insight types
- Requests, complaints, loved features, bugs, pain points, brand and feature mentions, and takeaways
- Behavior connection
- Feedback Events can be analyzed with Amplitude charts, cohorts, session replay, experiments, guides, and surveys
- Refresh
- Sources generally ingest daily; some can refresh as often as every 15 minutes
- LLM provider
- OpenAI for AI Feedback inference, with contractual no-training controls documented by Amplitude
- Residency
- US and EU regional data-plane options; verify account provisioning and connector flow
- Availability
- Included across Amplitude plans with volume and capability limits; larger analysis volumes available as an add-on
- Reviewed
- August 31, 2026 from Kraftful and Amplitude acquisition, product, documentation, pricing, security, API, and release sources
Overview
What Amplitude AI Feedback (formerly Kraftful) is
Kraftful is no longer best evaluated as a separate startup product. Amplitude acquired the company on July 10, 2025 and launched the fully integrated AI Feedback successor on November 12, 2025. The old Kraftful site remains online as a legacy product and acquisition record, while the maintained destination for new teams is Amplitude.
AI Feedback centralizes qualitative signals from customer support, sales calls, Slack, app stores, review sites, social channels, files, pasted text, and an HTTP ingestion API. Each comment, reply, review, or call utterance becomes an Amplitude Feedback Event, allowing teams to chart qualitative themes alongside product usage.
A hybrid pipeline first clusters feedback with classical machine learning, then uses an LLM to produce categorized insights. Current categories include feature requests, complaints, loved features, bugs, pain points, brands, feature mentions, and takeaways. Teams can define product areas and filter by source, date, severity, specificity, sentiment, and other event properties.
New Insights are regenerated as the source data changes and are intentionally ephemeral. Teams can save an important finding as a persistent trend, track semantically related mentions over time, view underlying source context, and create alerts for spikes such as critical bugs or complaints.
The integration with Amplitude is the main advantage over legacy Kraftful. Product teams can relate feedback to cohorts and behavior, inspect session replay, ask the Global Agent for evidence-backed analysis or a draft PRD, create guides and surveys, and send approved findings toward Jira, Linear, or a coding agent.
Amplitude documents that AI Feedback uses OpenAI as a third-party LLM for inference, that OpenAI does not use the data to train foundation models, and that requests are handled transiently with encryption in the account's regional data plane. Teams still control which sources are connected and remain responsible for lawful collection, access, retention, deletion, sensitive-data handling, and every decision made from an AI-generated interpretation.
Use cases
Who Amplitude AI Feedback (formerly Kraftful) is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
Multi-channel feedback programs
Centralize support, calls, stores, reviews, communities, social sources, surveys, and custom pipelines without manually reading every record.
Bug and complaint monitoring
Track recurring pain, severity, specificity, product area, volume, trend movement, and the original customer context.
Roadmap discovery
Quantify feature requests and customer language, then compare stated demand with observed behavior before prioritizing work.
Customer-success and product collaboration
Give support, research, product, growth, and engineering a shared evidence layer with source links and controlled project access.
Custom feedback ingestion
Send standalone comments, threaded conversations, or call transcripts through the current HTTP API when no native connector exists.
Closed-loop experimentation
Move from feedback signal to behavior analysis, a tested product change, a targeted guide or survey, and post-release trend measurement.
Capabilities
Core Amplitude AI Feedback (formerly Kraftful) features
Native feedback connectors
Connects services including Zendesk, Intercom, Salesforce Service, Freshdesk, HubSpot, Gong, Slack, app stores, G2, Trustpilot, Reddit, Discord, X, and Steam.
Files, text, and API ingestion
Accepts CSV and DOCX uploads, pasted feedback, and an HTTP API for comments, threads, and call-style conversations.
Feedback Events
Turns each ingested feedback unit into an Amplitude event with source, timestamp, rating, author, text, and assigned insight properties.
Hybrid analysis pipeline
Combines classical clustering with LLM interpretation rather than sending an undifferentiated corpus to a single prompt.
Automatic categories
Groups input into feature requests, complaints, loved features, bugs, pain points, brand mentions, feature mentions, and high-level takeaways.
Product areas
Assigns insights to team-defined parts of a product using the area name and description, enabling focused filtering and ownership.
Severity and specificity
Scores how much pain a theme represents and whether the underlying feedback is concrete enough to act on.
Source-grounded drill-down
Keeps mention counts, source information, dates, links, and verbatim context available beneath the AI summary.
Saved trends
Persists important themes and attaches new semantically related mentions so teams can measure movement over time.
Alerts
Builds Amplitude charts and alerts from category, source, severity, sentiment, product area, and other feedback-event properties.
Behavioral analysis
Lets teams segment feedback with analytics, merge supported identities, build cohorts, and compare qualitative signal with actual usage.
Global Agent analysis
Answers questions across feedback, retrieves customer quotes, identifies major issues, and can draft a PRD from selected evidence.
Action paths
Can create guides or surveys and exposes early-access Jira, Linear, and coding-agent actions from an insight.
Customer Feedback Agent
Runs interactive or scheduled analysis for themes, sentiment movement, emerging issues, and prioritized recommendations.
Process
How the Amplitude AI Feedback (formerly Kraftful) workflow works
Step 1
Confirm the product transition
Treat Amplitude AI Feedback as the maintained successor; do not base a new procurement or migration on legacy Kraftful pricing, login, or feature pages.
Step 2
Define the decision
Specify which product area, customer population, time range, and decision the analysis should inform, plus what evidence would change the team's conclusion.
Step 3
Inventory feedback rights
Map each source's owner, consent, contract, purpose, personal data, sensitive content, retention, region, and whether feedback may be combined with behavior.
Step 4
Minimize before connection
Exclude unnecessary call content, secrets, health or financial details, employee conversations, minors' data, and privileged or regulated information.
Step 5
Use a dedicated project and roles
Apply existing Amplitude project permissions, restrict connectors and exports, require MFA, and avoid placing unrelated clients or business units in one data boundary.
Step 6
Connect representative sources
Start with a small balanced set across support, sales, reviews, surveys, and community rather than letting one high-volume channel dominate.
Step 7
Normalize and deduplicate
Assign stable feedback IDs, distinguish replies from separate customers, map timestamps and ratings correctly, and prevent the same issue from multiple pipelines being counted twice.
Step 8
Define product areas clearly
Use specific names and descriptions, test ambiguous overlaps, and assign owners so category and routing quality can be audited.
Step 9
Calibrate against a human sample
Review a stratified set of raw mentions and measure category, severity, specificity, sentiment, product-area, identity, and quote accuracy before trusting aggregate output.
Step 10
Save only validated trends
New Insights change as data refreshes; preserve an important theme only after confirming the underlying mentions and creating an explicit tracking definition.
Step 11
Triangulate words with actions
Compare feedback volume with affected users, account value, behavior, funnel loss, replay, experiment results, and support burden rather than prioritizing by mention count alone.
Step 12
Review every action
Treat generated PRDs, recommendations, Jira or Linear issues, coding-agent prompts, guides, surveys, and customer messages as drafts requiring an accountable human.
Step 13
Monitor and correct
Track source health, refresh lag, duplicates, false grouping, missing segments, drift, alerts, downstream decisions, and whether shipped work actually reduces the trend.
Step 14
Delete from the source first
For data-subject deletion, remove the record upstream before Amplitude so the connector cannot re-import it; separately handle unmatched or aggregated feedback.
Cost
Amplitude AI Feedback (formerly Kraftful) pricing and free plan
Amplitude states that AI Feedback is available on every current plan. The Free plan publishes an allowance of 2,000 AI Feedback records, while larger volumes can be purchased as an add-on on Growth and Enterprise. Amplitude's broader billing is event-based: Plus can begin at $0 for the first two million monthly events and scales with usage; Growth and Enterprise are custom. Confirm the unit, refresh period, overage behavior, source limits, and AI Feedback add-on quote for the intended account.
Free
$0
A permanent no-card plan for evaluation and smaller product programs.
- 2 million events per month
- 2,000 published AI Feedback records
- Unlimited seats
- 10,000 monthly session replays
- AI Agents, MCP, analytics, limited experiments, guides and surveys
- Confirm feedback-record reset, connectors, history, and export limits in the workspace
Plus
Starts at $0; usage-based
For small teams that need to scale event volume and analytics beyond Free.
- First 2 million monthly events are free
- Credit card required
- Scales up to 70 million events
- Custom events, formulas, cohorts, alerts, heatmaps, and two-year retention
- AI Feedback is included, but confirm its plan-specific record allowance and any overage path
Growth
Custom event-based price
For businesses needing advanced analysis, permissions, monitoring, and larger feedback volume.
- AI Feedback included
- Larger feedback-analysis volumes available as an add-on
- SSO and project permissions
- Advanced behavioral exploration and monitoring
- Request exact record, source, retention, replay, region, support, and add-on terms
Enterprise
Custom event-based price
For large organizations needing portfolio scale, stronger access controls, and contracted terms.
- AI Feedback included with higher-volume add-on available
- Advanced user management and RBAC
- Data Access Controls and unlimited portfolio projects
- Custom service, security, support, residency, retention, and procurement terms
- Request a DPA, subprocessor review, deletion test, audit material, and connector-specific data-flow map
Startup Scholarship
One free year of Growth if eligible
A published program for qualifying early-stage startups.
- Fewer than 20 employees
- Less than $10 million in funding
- Application required
- Confirm post-scholarship pricing and AI Feedback allowance before onboarding sensitive or high-volume sources
Pricing checked . Check current pricing at the source ↗
Assessment
Amplitude AI Feedback (formerly Kraftful) strengths and limitations
Where it stands out
- Kraftful's core voice-of-customer capability now has a maintained successor inside a large product-analytics platform
- Combines qualitative feedback with observed behavior, cohorts, replay, experiments, surveys, and guides
- Supports a broad set of service, sales, review, social, file, and custom-API sources
- Feedback Events make AI-derived themes available to Amplitude's standard charting, segmentation, and alerting tools
- Source links and raw mentions make summaries auditable instead of presenting only a generated conclusion
- Saved Trends distinguish persistent monitoring from the changing New Insights feed
- Product areas, severity, and specificity add useful prioritization dimensions beyond sentiment and frequency
- The HTTP API supports comments, threaded conversations, and call transcripts when a native connector is missing
- AI Feedback is present on every Amplitude plan and has a published free allowance
- Customer Feedback Agent can provide scheduled and interactive synthesis without changing source configuration
- Amplitude documents the third-party LLM, no-training arrangement, encryption, regional processing, and access inheritance
- US and EU data-plane options plus published security and trust material support enterprise review
What to consider
- Legacy Kraftful marketing and pricing pages can make the standalone product appear current even though Amplitude is the maintained successor
- Moving into Amplitude changes the buying decision from a focused feedback tool to a broader event-based analytics platform
- The public pricing page does not fully spell out AI Feedback record limits, reset cadence, source limits, history, or add-on price for every plan
- Most connected sources refresh daily, and even the faster paths are not a universal real-time alerting system
- New Insights are regenerated and can change; an unsaved observation may disappear or be regrouped after the next analysis
- AI summaries can merge distinct issues, split one issue into several themes, misread sarcasm, lose qualifiers, or attach the wrong product area
- Severity and specificity are model-assigned judgments, not objective business impact or engineering priority
- Mention counts can be inflated by duplicate pipelines, long support threads, repeat reporters, bots, campaigns, or one highly vocal segment
- Feedback is selected and non-representative; silence does not mean satisfaction, and stated requests often differ from actual behavior
- AI Feedback summaries are generated in English, which can flatten nuance in multilingual sources even when raw comments retain their language
- The first 1,000 characters of feedback text and 200 characters per insight are stored in documented event properties, so long context may be truncated
- Identity merging by email can be incomplete or incorrect and can turn anonymous qualitative data into a more sensitive behavioral profile
- Connectors can ingest customer names, email, calls, support details, employee messages, reviews, and social data that require purpose and consent review
- Deleting in Amplitude before the upstream system can allow a connected source to restore the record on the next sync
- The User Privacy API cannot automatically match unmerged feedback identities, and aggregated or general feedback may require manual deletion
- AI Feedback uses a third-party OpenAI model and Snowflake is listed as an AI Feedback subprocessor; contracts and regional flow should be reviewed
- No-training and transient-inference commitments do not eliminate exposure through connectors, internal permissions, logs, exports, or downstream actions
- Generated PRDs and recommended actions can overfit loud feedback, omit technical constraints, and create false confidence without product and engineering review
- Early-access Jira, Linear, and coding-agent actions can move a mistaken interpretation into execution quickly
- Vendor accuracy and hallucination-safeguard claims do not replace a benchmark against the team's own languages, products, channels, and edge cases
Compare
Amplitude AI Feedback (formerly Kraftful) alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Data Analysis
Survicate AI Surveys
A survey-first customer-insight platform for teams that prioritize active feedback collection over deep linkage with product-event analytics.
Explore Survicate AI Surveys →Marketing
SurveyMonkey AI
A mature survey platform with AI-assisted creation and analysis for structured research programs and broad respondent workflows.
Explore SurveyMonkey AI →Business Operations
ChatGPT
A flexible assistant for manually analyzing de-identified feedback samples when a connected, continuously refreshed voice-of-customer system is unnecessary.
Explore ChatGPT →Project Management
Claude
A document-oriented assistant for controlled qualitative synthesis and research drafts when teams want to keep source selection and interpretation manual.
Explore Claude →Questions
Amplitude AI Feedback (formerly Kraftful) FAQs
Does Kraftful still exist?
Kraftful's site and historical materials remain online, but Amplitude acquired the company in July 2025. The maintained product for new teams is Amplitude AI Feedback, launched as a native platform capability in November 2025.
Is Amplitude AI Feedback the same as Kraftful?
It is the product successor built from Kraftful's voice-of-customer technology, not merely a renamed standalone account. It now uses Amplitude events, roles, analytics, cohorts, replay, experiments, guides, surveys, agents, and billing.
What does AI Feedback analyze?
It ingests support tickets, sales and call transcripts, app and software reviews, Slack and community discussions, social sources, surveys, Amplitude Assistant chats, CSV or DOCX files, pasted text, and HTTP API records.
How does AI Feedback organize feedback?
A hybrid clustering and LLM pipeline creates themes such as requests, bugs, complaints, praise, pain points, brands, and takeaways. It also assigns product areas, severity, and specificity and preserves links to supporting mentions.
Is Amplitude AI Feedback free?
It is included on the Free plan, which currently publishes a 2,000-record AI Feedback allowance. Plus, Growth, and Enterprise also include the feature, with larger analysis volumes available as an add-on on Growth and Enterprise.
How often does AI Feedback update?
After initial import, most sources pull new feedback once daily; some support refreshes as often as every 15 minutes. New Insights regenerate with each analysis, while Saved Trends attach relevant mentions on a continuing cadence.
Can AI Feedback connect comments to user behavior?
Yes. Feedback becomes Amplitude events and can be mapped to existing users when supported identity fields such as email align. That linkage is powerful but raises accuracy, consent, access, deletion, and profiling responsibilities.
Does Amplitude use feedback to train AI models?
Amplitude says AI Feedback processes selected customer data for inference only and that OpenAI does not use it to train foundation models. Review the current DPA, AI terms, subprocessor list, region, retention, and connector data flows for your organization.
Can AI Feedback write a PRD or create an issue?
The Global Agent can draft a PRD from feedback, and current docs list early-access Jira, Linear, and coding-agent actions. Treat all of these as proposals and require a product, research, security, and engineering review before execution.
Are AI Feedback insights accurate?
They are useful hypotheses, not ground truth. Inspect the underlying comments, test category and identity accuracy on a representative sample, remove duplicates, compare affected behavior, and involve researchers and domain owners.
How do I delete feedback data?
Delete it in the original connected source first, then in Amplitude, or the next sync may restore it. User Privacy API deletion works for feedback tied to merged Amplitude identities; other feedback and aggregates may need manual deletion.
Who should choose AI Feedback?
It is most compelling for teams already using or considering Amplitude and wanting one loop from customer language to product behavior and action. A dedicated research repository or survey platform may fit teams that do not need event analytics.
Bottom line
Our Amplitude AI Feedback (formerly Kraftful) verdict
Kraftful's acquisition produced a stronger active successor: Amplitude AI Feedback connects qualitative customer language with the behavioral evidence needed to judge scale and impact. That integration is the reason to choose it. It is less attractive if the team wants a standalone research repository without event-based analytics. Start with the free allowance, connect a deliberately small source set, benchmark themes against human coding, validate identity and deletion flows, and require source-level review before any AI-generated roadmap item, alert, customer intervention, or code task becomes real work.
Visit Amplitude AI Feedback (formerly Kraftful) website ↗Similar Project Management Tools
View all Project Management AI tools →
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.