Overview

ScatterSpoke was rebuilt from scratch as v3 in August 2026: guided retro boards, static surveys and AI-run interview surveys, and Ask, which answers questions across closed retros and surveys with cited evidence. An official OAuth MCP server is included on every tier, including Free.

Best for: Engineering orgs that want AI to synthesise feedback across many teams and surveys — and agent builders who need a survey backend over MCP.

How ScatterSpoke describes itself

The newer pages sell ScatterSpoke 3.0 as survey infrastructure first: AI-moderated interviews that ask the follow-up a form can't, a free survey plan pitched straight at SurveyMonkey and Typeform users, and an MCP server for agent builders.

"Interview depth, at survey scale."

The homepage and retro pages still carry the older pitch — "the AI copilot for team feedback", Jira/Slack/Teams badges, AI summaries with action items — which the v3 retro board doesn't ship. The documentation is blunter about what changed:

"v2.5 collected feedback into boards you read, while v3 collects feedback into evidence you can question."

Our take

The AI on top of the retro is the real asset. Group with AI merges the leftover cards into named groups without touching the ones people made by hand, the closed-retro summary tracks which themes repeat and grow across runs, and Ask answers questions over closed retros and surveys and cites the feedback behind them. AI-run interview surveys are a genuine option for async health checks between sprints.

The board recovered in September: the manual Group phase is back, writing stays private until the facilitator reveals it, voting is blind by default, and columns can be edited mid-session. Action items are still missing — the vendor's own walkthrough records the next step in a comment thread. There are no integrations in v3: nothing pushes to Jira or posts to Slack or Teams, and export is one CSV. The docs also describe Signals and Automations pages that the web app doesn't include as of October 2026.

Best fit: teams that want AI synthesis across retros and surveys, and don't need the retro to end in a tracked, owned action.

Features

Retrospectives

TemplatesCustom templates· Create your own retro formats in the templates gallery (shared workspace templates need Business); on a live board, columns can be renamed, added or removed but not reorderedAnonymous inputPrivate brainstorm (hide until reveal)· On by default for guided boards — others' cards show as 'Feedback hidden until reveal' until the facilitator reveals or moves to Group; Live boards and no-sign-in link boards are always visiblePrivate/public retros· Four access rungs — one team, the whole workspace, link with email verification, or link with no sign-in (forced anonymous)Custom process flow· Two run styles — Guided (Feedback, Group, Vote, Review) or Live with no phases; the phases themselves are fixed

Health checks

Health checks· Recurring Survey Series (weekly pulse, monthly team health check) plus a per-retro 1–5 temperature check; no dedicated multi-dimension health modelMood / sentiment check

Facilitation

Meeting timerPresentation mode· Review-phase discussion stage — the facilitator puts one card or group on a shared stage, working down a vote-ranked queueComments on cardsDrag-and-drop grouping· Group phase (restored September 2026) — drag one card onto another to make a named group; groups can be renamed, pinned, detached or disbandedAsync mode

Voting

Independent dot votingCustom vote countVote at any time· Any time on a Live board; Vote-phase only on a guided board

Engagement

Mobile friendlyPolling· Survey builder with nine question types plus an NPS preset; not an in-retro pollCheck-in / check-out· Anonymous 1–5 temperature check runs beside the board; no check-out roundGIFs· Pasted Giphy/Tenor and image URLs render inline; no GIF picker or search

AI

AI grouping / clustering· Pro and up — Group with AI merges the remaining cards into named groups in the Group phase, leaves hand-made groups alone, and suggests names for groups people makeAI template suggestions· AI designs an Agent Survey from a plain-English goal in a Design interview; retro formats are picked by handAI summary· Survey Results Briefs are AI-written; a closed retro's summary is statistics only, so an AI recap of a retro means a metered Ask queryAI sentiment / heatmapsAI retro / health-check insightsAI discussion prompts· 'Probe deeper' on a theme drafts a pre-briefed Agent Survey; Agent Surveys generate adaptive follow-up questionsMetered AI (credits / quota)· Ask queries and Agent Surveys draw on shared workspace pools — 250 Ask / 20 Agent Surveys a month on Pro, 2,000 / 150 on Business, 5,000 Ask on Enterprise; Free gets 50 Ask / 10 Agent Survey credits once; top-up packs of 250 / 50

Reporting

Cross-team reporting· Ask can read every team's closed retros and surveys at once, and the MCP server groups retro totals and themes by team; no cross-team dashboard in the appExport formats· CSV

Enterprise

SSO / SAML· SAML 2.0 (Okta, Entra ID, ADFS) — paid add-on on Business, included on Enterprise; self-serve metadata upload and SSO enforcementSCIM provisioning· Enterprise tier only per the pricing page; the v3 docs cover SAML just-in-time provisioning, not SCIMSOC 2 Type 2· Vendor states SOC 2 Type II from the start with annual penetration testing; no public trust portal or report pageGDPR compliantAudit logs· Enterprise tier only — Free, Pro and Business see a blurred sample behind an upgrade promptFacilitator-only roleGuest / observer rolesMeeting historyEmail domain restrictions· Verified-email guest access mode and SSO domain verification; no workspace-wide email-domain allowlistUptime SLA / dedicated support· Priority support on Business, dedicated CSM on Enterprise; no uptime figure publishedPublic API· Business and Enterprise only; scoped bearer keys, but no public endpoint-level reference published yetMCP server

ScatterSpoke

Overview

How ScatterSpoke describes itself

Our take

Score breakdown

Pros

Cons