Dust agents look it up when you ask. Evermuse has already counted it.
Dust is a shared workspace where teams build AI agents on company knowledge and tools. But “how many customers asked for this?” should not depend on what one agent retrieves in one reply. It needs every conversation processed as it arrives, repeat asks merged, and a quote behind each number. That is the customer evidence layer Evermuse adds, and Dust agents can call it over MCP.
The short answer
Dust is a platform for building, sharing and running AI agents: teams connect company knowledge and tools, then create agents that retrieve what was written down and act on it when someone asks. Evermuse is the customer evidence layer for product teams: it continuously reads every customer call, ticket and chat, merges repeat requests, and tells you how many customers asked for something, with the quotes to prove it. They work well together, because Dust agents can query Evermuse over MCP.
A workshop for agents. A record of what customers said.
Dust was built so every team can build the AI it needs. Evermuse was built so product decisions rest on what customers actually said, not on whoever spoke last.
Evermuse
The customer evidence layer that counts
Built for the moment someone asks “how many customers actually want this?” It answers with a number you can defend and every quote behind it.
- Processes every call, ticket and chat as it lands
- Merges repeat requests, counts each customer once
- Ranks needs by segment, with quotes behind each
- Pushes cited evidence into specs, PRs and agents
Dust
Multiplayer AI agent platform
A shared workspace where people and agents use the same company context, tools and models to get work done. Excellent for building agents for many workflows. Not designed to keep a counted record of every customer conversation for you.
- Build and share custom agents
- Connectors to Slack, Notion, Drive and more
- Schedules, triggers and multi-agent workflows
- Spaces, permissions and audit logs
Bluntly: Dust is the better way to build agents across your company. Evermuse is the better way to know what your customers want.
Looking it up vs. keeping count
Same question, asked at different moments. One does the work when you ask. The other did it as each conversation came in.
Process everything, keep the count
A call, ticket or chat arrives → Evermuse reads it in full and pulls out each request → each request is matched to the same need from other customers → each account is counted once → the count links to every quote. When you ask, the work is already done, so the number is the same tomorrow.
More on this in How many customers asked for this feature?
Ask, then retrieve
You message an agent → it searches the knowledge and tools it was given → what it finds goes to the model → the model writes an answer from that. It is a strong pattern for drafting, triage and research. For a count, the answer covers only what that one reply retrieved, and how well it works depends on how the agent was built.
Right for “help me with this.” Shaky for “how many?”
What Evermuse adds to an agent platform
Not another place to build agents. A maintained record of what customers said, counted and cited, that every agent and every teammate can draw on.
Done before anyone asks
An agent works when someone sends it a message: it looks up what it can reach, then writes an answer. Evermuse works the other way round. Every conversation is read as it lands and folded into the record, so when a PM asks "how many customers want SSO?" the count already covers all of them.
One need, however it was phrased
"SAML login," "Okta support," and "our IT team needs single sign-on" are the same request. Evermuse merges them into one signal and counts each account once, so a customer who raised it on five calls counts as one customer, not five.
A quote under every number
Each signal links to the exact quote, who said it, which account, and when. When someone says "38 customers asked for this," anyone can open all 38 and read them. That is what lets the number survive a roadmap review.
A maintained pipeline, not a side project
A homegrown insight agent is only as good as the person who built it, and it needs someone to keep it working. Evermuse is maintained for you, writes cited evidence into specs and PR checks, and serves it over MCP to Dust, Claude Code, Cursor and Codex alike.
Side by side, weighted for product work
An honest scorecard. Dust wins several rows cleanly, and we'll say so.
| Capability | ||
|---|---|---|
| Processes every customer call, ticket and chat as it arrives, before anyone asks | ||
| Keeps a running count of how many customers raised each need, out of the box | ||
| Merges the same request said in different words, and counts each customer once | ||
| Every number backed by the exact quote, speaker and timestamp | ||
| Customer evidence written into specs and checked in PR reviews | ||
| Works without anyone on the team building or maintaining an agent | ||
| Connects to call recorders, support tools and CRM | ||
| Works with agents over MCP | ||
| Answers cite their sources | ||
| SOC 2 Type II certified | ||
| Build and share custom agents for any team or workflow | ||
| Agents that take actions in connected apps on schedules and triggers | ||
| Connectors to internal knowledge: Slack, Notion, Drive, GitHub and more | ||
| One AI workspace for every department, with spaces and permissions |
Processes every customer call, ticket and chat as it arrives, before anyone asks
Keeps a running count of how many customers raised each need, out of the box
Merges the same request said in different words, and counts each customer once
Every number backed by the exact quote, speaker and timestamp
Customer evidence written into specs and checked in PR reviews
Works without anyone on the team building or maintaining an agent
Connects to call recorders, support tools and CRM
Works with agents over MCP
Answers cite their sources
SOC 2 Type II certified
Build and share custom agents for any team or workflow
Agents that take actions in connected apps on schedules and triggers
Connectors to internal knowledge: Slack, Notion, Drive, GitHub and more
One AI workspace for every department, with spaces and permissions
Based on publicly available information as of November 2026. A partial mark for Dust means a team can build it with agents, but it is not provided out of the box. Dust is a trademark of its owner; this page is not affiliated with or endorsed by Dust.
Where Dust is the stronger choice
If these are your main problems, start with Dust. We'd rather tell you now.
Building your own agents
Dust lets the people closest to the work build agents with their own instructions, knowledge and tools, then share them with the team. Evermuse is not an agent builder, and does not try to be one.
Workflows that act
Dust agents can run on schedules and triggers, coordinate with other agents and take actions in connected apps, from ticket triage to CRM updates. Evermuse is a source of customer evidence that agents read from, not a general automation tool.
One AI workspace for every team
Engineering, support, sales, marketing and data teams can all work in Dust, with spaces to separate data and permissions by team, plus audit logs and usage analytics. Evermuse is built for the teams that decide what to build and the people who talk to customers.
Reach across internal knowledge
Dust connects to Slack, Notion, Google Drive, GitHub, Salesforce, Zendesk and many more, and lets teams pick the model that suits each agent. If your need is AI over the whole of your internal knowledge, that breadth matters.
Five questions to ask both tools
Run them on your own data during a trial. Be hardest on us.
- 1How many customers asked for this feature in the last 90 days? Do you get the same number if you ask again next week?
- 2Was that count built from every call and ticket, or from what the agent pulled in for this one message?
- 3If one account raised it on five calls, is that one customer or five?
- 4Can I click from the number to every quote behind it, with speaker, account and date?
- 5Who built the agent that produced this answer, and who keeps it working when the volume of calls doubles?
Which one, or both?
It comes down to whether you need a place to build agents, or a record of what customers want.
Choose Evermuse if…
Roadmap calls come down to who argued loudest, you can't say how many customers asked for something without a week of digging, or your coding agents build without knowing what customers said. You want a number you can defend and the quotes behind it, kept current without anyone on the team maintaining an agent to produce it.
Choose Dust if…
You want every team to build and share its own agents on company knowledge, run workflows on schedules and triggers, and take actions in the tools they already use, all in one governed workspace with the model of their choice for each job.
Running both
Dust stays where your teams build and run agents. Evermuse is the customer evidence layer those agents draw on. A Dust admin adds the Evermuse MCP server as a remote MCP server, and any Dust agent can ask what customers need and get counted, cited evidence back, so no team has to rebuild its own customer-insight agent. The same server works with Claude, Cursor and ChatGPT. For how the pieces fit, see What's inside a company brain?
Evermuse vs. Dust: FAQ
Is Evermuse a Dust alternative?
Only for one job. Dust is a platform where teams build, share and run AI agents on top of company knowledge and tools like Slack, Notion, Google Drive, GitHub and Salesforce. Evermuse is the customer evidence layer for product teams: it reads every customer call, ticket and chat as it arrives, merges repeat requests into one signal, and counts them with quotes attached. If you want agents for many teams and workflows, Dust is the better fit. If you need to know what customers want and how many asked, Evermuse is.
Can a Dust agent tell me how many customers asked for a feature?
Some teams already use Dust agents over their customer conversations, and an agent can find and summarize the calls and tickets it reaches when you ask. A count that holds up is a different job: every conversation has to be read, the same request merged across different wording, and each customer counted once, and someone has to keep that working as volume grows. Evermuse does that work continuously, ahead of the question, so the number is the same when you ask again tomorrow.
Can Dust agents use Evermuse?
Yes. Dust lets workspace admins add remote MCP servers by URL, with OAuth supported, and Evermuse has an MCP server. Once it is added, any Dust agent your team builds can ask Evermuse what customers said about a topic and get counted, cited evidence back, instead of each team rebuilding its own customer-insight agent. Dust lists 5 remote MCP servers on its Business plan and unlimited on Enterprise.
We built a customer-insight agent in Dust. Why add Evermuse?
Ask two questions about it: who maintains it, and what it does when there are more calls than fit in one request. Evermuse is a maintained pipeline rather than a side project one team owns. It processes every conversation, keeps the counts current, and links each number to the exact quotes behind it. Your Dust agent can keep doing the work it is good at and call Evermuse for the customer evidence.
Who is each tool built for?
Dust is built for teams across the company, from engineering and support to sales, marketing and data, with the people closest to the work building their own agents. Evermuse is built for product, research, design and engineering teams, and also serves sales and CS. A product team can start on Evermuse without anyone building or tuning an agent first.
How much does Dust cost compared with Evermuse?
Dust publishes its pricing: a Business plan for teams of up to 100 people with Free, Pro and Max seats, each carrying a monthly credit allowance (Pro is listed at €30 per seat per month, or €24 billed yearly), and a custom-priced Enterprise plan. Credit use depends on the model and tools an agent uses. Evermuse has self-serve plans priced by monthly credits, each with a 14-day money-back guarantee, and an Enterprise plan for larger companies.
Is Evermuse secure enough to connect to Dust?
Evermuse is SOC 2 Type II certified and GDPR-ready, encrypts data with AES-256, offers SSO on Business and Enterprise plans, and offers EU or US data residency on Enterprise. Agents connect through OAuth 2.1 and scoped tokens, and every answer links back to the source conversation it came from.
Teams that stopped guessing what customers want
“I'm reviewing the insights your product provided – my mind is blown! This is such a game-changer.”

Shira Dassa
Product @ Yotpo
$436M Raised · 600+ Employees
“Last month alone, we'd save 8.5 hours per team member using Evermuse.”

Min Zhou
Design Lead @ OpenSea
$427M Raised · 700+ Employees