theyasked.me

Compare · Facts about Featurebase checked 2026-10-22

Why AI product teams pick TheyAsked over Featurebase

Featurebase has grown into a support suite: inbox, AI agent, help center, and feedback as one module. TheyAsked does one job. It tells you, and the agents you build with, what customers asked for, who asked and what they pay.

Start free · Move from Featurebase

Short answer

Pick Featurebase if you want one tool for support and feedback: a shared inbox, an AI agent that answers tickets, a help center, and feedback that flows in from those conversations and your call recorder. Pick TheyAsked if your team already has support covered and wants a focused demand layer that Claude, Cursor or your own agents can query and act on: 33 MCP tools, ranking by revenue and momentum in one call, interactive views inside Claude, and a flat price per workspace instead of a price per seat.

Why AI product managers and engineers pick TheyAsked

  1. Answers ranked for you, not raw records

    TheyAsked tools return requests ranked by unique voters, MRR of the voting companies, momentum (last 14 days against the 14 before) and opportunity. Ask "what should we build next for Enterprise" and the answer comes back already sorted. Featurebase ranks by requesters and revenue in its app, but its MCP documentation describes typed tools over the general API and no ranking or analytics tools.

  2. Who asked, in one call

    `who_asked_for` returns the plan mix, the companies with their MRR, the platforms and the voter notes behind a request in a single call. Your agent does not have to list voters, fetch each user and add it up. That is the question an AI PM gets in every roadmap review.

  3. Views that render inside Claude

    In claude.ai and Claude Desktop, TheyAsked tools draw interactive views through MCP Apps: a ranked ledger with sparklines, KPI cards, segment breakdowns, request and account cards. You can mark a request planned from the view, with a confirm step. Five MCP prompts (weekly digest, triage, roadmap review prep, release notes, account brief) package the common workflows.

  4. Guardrails built for agents that write

    Ten granular scopes, a read-only endpoint at /mcp/readonly, OAuth 2.1 with dynamic client registration for claude.ai and ChatGPT, and API keys for Claude Code and CI. Customer text is fenced as untrusted data, destructive tools need an explicit confirm, and every AI action lands in an audit log. Featurebase also takes this seriously, with separate read and write tools and some actions kept to API keys only.

  5. One price for the workspace, not per seat

    Featurebase charges per seat: $29, $59 or $99 per seat per month billed yearly, plus $0.49 per AI resolution. TheyAsked Pro is $29 a month for the workspace with up to 10 admins and MCP included, and Business is $99 with unlimited admins. Voters are never counted on any plan.

  6. Built for product engineers who ship the loop

    REST API, webhooks, a web widget, iOS and Android WebView embeds with signed JWT SSO, and a portal on your own domain. When a request ships, the changelog notifies everyone who asked. Your agents can log feedback from a call, vote on behalf of a customer and draft the release notes through the same MCP server.

TheyAsked vs Featurebase

Line by lineTheyAskedFeaturebase
What it isFocused feedback and demand tool: boards, voting, roadmap, changelogSupport suite: inbox, AI agent, help center, feedback, surveys, product tours
Pricing modelFlat per workspace: Free $0, Pro $29/mo, Business $99/moPer seat: Growth $29, Professional $59, Enterprise $99 per seat/mo billed yearly
Admins includedFree 3, Pro 10, Business unlimitedFree plan has 1 seat; paid plans priced per seat
AI usage costAI answers included, with a daily allowance per plan$0.49 per AI resolution on paid plans
MCP server33 tools (17 read, 16 write) from ProHosted at mcp.featurebase.app, on paid plans; tool names not published
Ranking in the MCP toolsUnique voters, MRR of voting companies, momentum, opportunityRanking by requesters, revenue or opportunity in the app; no ranking tools documented for MCP
Interactive views and prompts in ClaudeMCP Apps views plus 5 MCP promptsNone described in the MCP docs
MCP access control10 granular scopes, /mcp/readonly, OAuth 2.1 + DCR and API keysmcp:read and mcp:write scopes, OAuth sign-in, some actions API-key only
Prompt-injection defenseCustomer text fenced as untrusted, confirm on destructive tools, audit logRead and write tools separated; a write connection that cannot read customer-written text *
Mobile embed for end usersiOS and Android WebView embed with signed JWT SSOWebView of the portal with a JWT in the URL; native mobile app is for agents
Feedback from support and callsLog from calls through MCP (create_post, vote_on_behalf with notes)Captured from support conversations; Fathom, tl;dv, Grain, Gong, Fireflies, Granola
Help desk and help centerNot includedIncluded: inbox, ticketing, live chat, help center articles

From each vendor’s public pages, checked 2026-10-22. * Not confirmed on Featurebase’s own site.

When Featurebase is the better pick

  • Teams that want support and feedback in one tool: shared inbox, ticketing, live chat, help center and feedback in one customer portal.
  • Teams that want an AI agent (Fibi) answering support tickets, with feedback captured from those conversations automatically.
  • Teams that record customer calls in Fathom, tl;dv, Grain, Gong, Fireflies or Granola and want requests pulled from them without an agent in the middle.
  • Teams with Salesforce or HubSpot who want two-way CRM sync and open deal value tied to requests.

Moving from Featurebase

In Featurebase, go to Settings, Data, Imports & Exports and click Export Data. You get a CSV with each request's title, description, creation date, author name and email, category, labels, status, ETA, requester count, monthly spend and custom fields. In TheyAsked, open Settings, Import and choose the spreadsheet route (CSV or XLSX). Columns are matched by name and you can remap each one: title, body, board, status, tags, votes, author email, created date and ETA. Map category to board and labels to tags. The export carries requester counts, not individual voters, and no comments, so to bring real votes add a voterEmails column built from Featurebase's requester export before you import. Help center articles and support conversations stay in Featurebase, since TheyAsked does not do support.

How importing works

Questions

What is the best Featurebase alternative for AI product teams?
If you mainly need feedback and prioritization that your AI tools can use, TheyAsked is built for that. It has 33 MCP tools that rank requests by revenue and momentum, views that render inside Claude, and flat pricing per workspace. If you need a help desk too, Featurebase covers more ground.
Does Featurebase have an MCP server?
Yes. Featurebase hosts one at mcp.featurebase.app on its paid plans. It works with Claude, ChatGPT, Cursor, VS Code, Claude Code and Codex, and covers feedback, help desk, help center, users, reports and updates. Its docs do not list tool names, ranking tools or interactive views.
Is Featurebase priced per seat?
Yes. Featurebase lists Growth at $29, Professional at $59 and Enterprise at $99 per seat per month, billed yearly, plus $0.49 per AI resolution. TheyAsked charges per workspace: Pro is $29 a month with up to 10 admins, Business is $99 with unlimited admins, and voters are never counted.
Can I move my feedback from Featurebase to TheyAsked?
Yes. Export requests from Featurebase as CSV, then import the file in TheyAsked through the CSV or XLSX route and map the columns. Titles, descriptions, statuses, labels, ETAs and vote counts come along. Individual voters need a voter email column, and comments are not in Featurebase's export.
Which is safer for agents that write to my board?
Both take it seriously. TheyAsked offers 10 granular scopes, a read-only endpoint, fencing of customer text, a confirm step on destructive tools and an audit log. Featurebase separates read and write tools and keeps some actions to API keys only. Its earlier developer docs also described a stricter split, where a writing connection cannot read customer-written text at all. TheyAsked does not enforce that split; it fences customer text and offers a read-only endpoint instead.

Sources

Other comparisons