Last known major update
October 7, 2026
Frank AI added interview control settings.
Official announcementAbout the tool
Frank AI is an AI researcher that interviews your customers. You share your website and goals, chat with its research agent for about two minutes to build an interview outline, and send a link. Frank then talks to people by voice, video or WhatsApp chat, asks follow-up questions as the conversation goes, and can run hundreds of interviews at the same time in more than 30 languages. Typical uses are discovery interviews, feedback after a feature release, churn and exit interviews, win and loss analysis and market research. After the interviews you get sentiment and thematic analysis across all conversations, transcripts and recordings, short video reels of the main insights and a chat assistant that answers questions about the results. A widget embeds interviews in your website or app, and a webhook sends results to your own systems. A public REST API and an MCP server let tools such as Claude and Codex create studies and read transcripts; studies created this way stay drafts until you approve publishing them. The Free plan gives 60 voice minutes a month with interviews of up to 10 minutes. Starter costs $59 a month, or $49 a month billed yearly, for 120 minutes; Growth costs $199, or $166 billed yearly, for 450 minutes; Business costs $375, or $312 billed yearly, for 950 minutes. Paid plans start with a 14-day free trial, and students get 50% off. Frank is made by Prelaunch Group, Inc., the company behind Prelaunch.com.
Key use cases
Customer Feedback Analysis
Sort user feedback into insights that support, product, marketing and customer success teams can act on.
Journey Mapping
Organize personal, customer or project experiences into clear stages, patterns and progress over time.
API & developer readiness
Public REST API at https://api.hifrank.ai with a secret key sent as a bearer token. Endpoints create products (POST /v1/products) and research studies (POST /v1/researches), attach a product, publish a study, add or remove participants, return embed codes and list interviews and transcripts (GET /v1/researches/{id}/interviews, GET /v1/interviews/{id}). A webhook sends interview results to your systems. The MCP server at https://api.hifrank.ai/mcp works with Claude Code, Claude Desktop, Codex and other MCP clients through browser sign-in, with permission scopes such as research:read, interviews:read and research:publish; publishing a study needs a separate approval.


