Best Platforms for a Shared User Research Knowledge Base (2026)

Compare AI queries, MCP, and research collection

MCP connects compatible AI tools to external research systems. Compare what a connection can read, write, and launch, alongside its plan requirements. A summary feature does not necessarily provide conversational search across the whole repository. The table distinguishes those capabilities; unconfirmed entries need vendor confirmation.

The platform profiles below link the primary product and integration sources for these distinctions. In a pilot, ask the same concrete question through the web interface and the proposed AI client. Check the returned sources, permission boundaries, and whether the agent can take an action that starts participant collection.

Compare free entry, team access, and pricing

Plan details checked October 26, 2026. Prices are USD except where marked €. A trial or sample test is different from an ongoing Free workspace. Count the people who create research separately from viewers and observers; projects, channels, and AI agents are usage allowances, not member seats.

Pricing sources: Palette, Dovetail, Condens, Marvin, Looppanel, Aurelius, Great Question, UserTesting, Stravito, and Fred. Request a quote for your actual editors, readers, collection volume, recruitment, and AI usage before comparing total costs.

Palette: run studies and build shared knowledge in one workspace

Palette fits a product team whose unanswered questions require new research as well as better access to past findings. Its public workflow combines user studies, usability tests, microstudies, connected tools, and uploaded documents. Ask Palette uses that evidence to answer questions, while insights link to their originating studies, connected tools, or documents. A PM can investigate a finding and use the same workspace to collect the missing evidence.

The developer-facing workflow is concrete: on Pro, Palette MCP brings research into Claude Code, Codex, and Cursor. The MCP documentation distinguishes Read, Write, and Commission permissions. Palette in Slack supports research questions and approved own-user study launches; paid external-panel studies are reviewed in the web app. This connects evidence retrieval with the next study, rather than requiring a separate collection product.

For ongoing learning, microstudies ask questions while users are in the product. Own-user research has no recruitment charge within the plan’s study limits; external recruitment uses credits, with the final cost confirmed before launch. Ten credits equal $1 USD. The trade-offs are explicit: collaboration, Slack, and MCP require Pro; heatmaps are still listed as coming soon. The member documentation also gives every member full workspace access. Teams needing restricted project or viewer roles should check that requirement before adopting a shared workspace.

Dovetail: connect customer evidence with cited answers

Dovetail is a strong shortlist option when calls, tickets, interviews, and documents already exist across the business. AI Chat returns answers linked to original material. Its MCP integration can retrieve evidence and create research artifacts; Enterprise also supports queries in Slack.

The purchasing trade-off is the step from a limited Free workspace to a custom Enterprise contract. Check the current user and usage-credit terms rather than translating unlimited channels or agents into unlimited people. Creating a research project through MCP does not itself recruit participants or run a study; Dovetail retired Recruit.

Condens: curate findings for a wider stakeholder audience

Condens suits teams that want qualitative analysis and a deliberately curated repository. Its AI search can work with raw evidence or validated findings. Stakeholder-facing Insights Magazines give published research an editorial structure. The MCP connection is read-only and limited to published artifacts; client options differ between Business and Enterprise. Published findings can also be queried in Slack and Teams, with plan conditions.

The important cost distinction is between solo analysis and shared stakeholder access. Lite has no viewer seats; Business includes five contributors and unlimited viewers. A participant pool stores participant information, but Condens explicitly does not provide recruitment services. Choose it for organizing and sharing research when collection is already handled elsewhere.

Marvin: combine a knowledge hub with AI-assisted research

Marvin’s Knowledge Hub centralizes customer evidence, while Ask AI returns source-cited answers. Its MCP actions include writing research artifacts, and Slack integration supports research queries and sharing. Marvin also offers AI interviews and live intercepts through a paid research add-on, so its scope extends beyond imported recordings.

Compare full users, collaborators, and viewers separately when requesting a quote. Current pricing places MCP on Pro, while older help uses different access language; confirm the contracted entitlement. The relevant trade-off is the package you need for broad sharing, collection add-ons, and security controls, not simply whether an AI feature appears on the homepage.

Looppanel: keep interviews searchable and easy to share

Looppanel fits teams whose evidence comes mainly from interviews. Its repository search connects questions with source quotes and clips. Pro provides a team workspace, global search, Ask AI, and unlimited viewers. The MCP connection is a read-only beta that must be enabled by Looppanel. Enterprise adds a stakeholder repository view and Slack queries with citations.

Check the difference between project-level search on Solo and shared retrieval on Pro. Its new Panel Management beta helps manage existing participants, screeners, and scheduling links. It is not an external recruitment marketplace, and its eventual pricing should be confirmed separately.

Aurelius: share curated research with an unlimited-user model

Aurelius organizes notes, tags, collections, clips, and linked insights into a shared research archive. Its features include AI-assisted summaries and themes. The unlimited-user pricing model can suit a team that wants many people working with a curated collection. However, read-only and project-specific permissions require Enterprise, and AI summaries should not be treated as verified conversational search across the whole repository.

Aurelius uses Zapier for workflows such as Slack sharing. Zapier also exposes limited Aurelius actions through MCP; that does not establish native, evidence-grounded repository querying from an AI client. Pilot the exact retrieval task before choosing it for a developer-facing workflow.

Great Question: connect participant operations with the repository

Great Question fits teams that want participant management, recruitment, scheduling, incentives, study execution, and a repository connected. Ask AI queries study and repository evidence with linked quotes. Its Enterprise MCP workflow adds research retrieval and study actions. Slack integration covers notifications and sharing, which differs from asking a research bot questions inside a channel.

Unlimited free observers make stakeholder distribution a useful part of the comparison. Self-serve is capped at five paid seats, while Enterprise starts at five. Recruitment and participant incentives are separate costs. If your team only needs a searchable archive, evaluate how much of the broader operations workflow you would actually use.

UserTesting Insights Hub: keep knowledge beside a testing program

UserTesting is relevant when participant access and moderated or unmoderated testing already anchor the team’s research. Evaluate the current Insights Hub offering and its Insights Discovery query feature. Answers can link to video moments or survey themes, but supported-source restrictions matter: do not assume every imported file, older study, or session type is searchable through the same AI feature.

Its MCP server supports research actions in limited early access. Current plans include Insights Hub on Ultimate and Ultimate+, with an add-on on Advanced. The platform also supports Slack sharing. Confirm the edition, commercial model, AI source coverage, and creation rights in the quote. Teams evaluating legacy EnjoyHQ data should confirm migration and feature parity rather than treating its history as proof of what a current Insights Hub contract includes.

Stravito: make an enterprise insights library discoverable

Stravito fits organizations with a substantial existing collection of market research, reports, decks, videos, and customer studies. Its AI Assistant returns evidence-grounded answers with page-level references. The published MCP workflow is read-only and respects user permissions; confirm access terms because its Connect page uses early-access language. Ask specifically which AI clients or messaging surfaces your contract supports.

The evaluation is primarily about discovering and distributing existing knowledge. Check it against the needs of a cross-company insights library, rather than assuming it replaces participant-facing study collection. Prices and member allowances need a commercial conversation.

Fred: connect study evidence with roadmap decisions

Fred, the UXR Shepherd at meet-fred.com, combines research methods with a record of why a product decision was made. Its repository connects studies, quotes, clips, and findings with decision context. That can fit teams seeking a traceable case for a roadmap choice. Team unlocks AI summaries, insights, and report generation; the presence of an AI assistant on a plan does not establish global conversational querying or native MCP.

Current pricing sends buyers to local-currency checkout amounts, so confirm the price rather than relying on older dollar figures. Fred documents sharing approved findings to selected Slack channels in its data-handling policy. Check study volume, advanced synthesis, and governance alongside the 5-, 15-, and 25-seat self-serve allowances.

Pilot the shared knowledge workflow before committing

Use one real product question, two relevant studies, and one connected document for a small pilot. Invite a PM, an engineer, and the person responsible for the original research. Give them retrieval tasks that end with a source they can inspect, then try collecting the next piece of missing evidence.

1. Ask a specific question, then open the source response, clip, or report page behind the answer.
2. Repeat it as a stakeholder or viewer, checking both useful access and restricted material.
3. Try the proposed Slack or MCP workflow. Verify the client, feature scope, and permissions actually included in the plan.
4. Create a new study or import a new interview, then check how it becomes discoverable alongside the existing evidence.
5. Price the real editors, viewers, AI usage, recruitment, and required security controls. Test an export before committing the whole archive.

Palette is worth piloting when the same team needs to ask, collect, and revisit research in its product workflow. A dedicated analysis tool may fit better when curation and access governance dominate; a broader operations platform may fit when recruitment and scheduling are the main constraint. The AI research tools comparison maps these choices to PM jobs, and the research MCP guide explains how to carry evidence into an AI-assisted product workflow.