Opportunities › Product analytics for AI chat and voice agents › Proposed app
Murmur
Find where users get frustrated with your chat or voice agent
- Score
- 45
- confidence 63
- Evidence
- 16
- items on the page
- Rivals
- 3
- named in sources
- First version
- 45 d
- estimate
- Type
- Desktop app
- PWA: runs in the browser, installs on Mac and Windows
- Market since
- 2023
- Emerging market · AI estimate
ScreensConcept
Desktop app: Product engineers review frustration clusters, analyze conversation replays, and act on user feedback with analytical dashboards at a desk. Sample screens on a computer, in the proposed brand. Their numbers are sample data.
Overview
Frustrated sessions, up 0.8 pts7.4%
About this appAI proposal
Murmur reads the production conversations of your chat or voice agent and shows where users struggle. It flags rageprompting, repeated rephrasing, corrections to the agent and silent drop-offs, then groups them into clusters you can act on.
It also finds what users keep asking for in different words. Many phrasings of one missing feature become a single cluster with a count, example messages and a trend, so you can decide what to build or fix next.
Built for product engineers who have traces and latency numbers but still guess whether users got what they wanted. Murmur does not replace observability or evals. It covers what they do not: what users meant and where they gave up.
- Tagline
- Hear what your users meant, not just what they typed.
- Voice
- PlainCuriousDirect
- Why this name
- A murmur is the low sound of many people talking. The name suggests listening to the crowd of conversations behind your agent.
In the first version
Members say which features they would use. It tells PWAS what people want; it changes no score.
Frustration signals
Detects cursing, repeated requests, corrections and abandoned sessions in chat and voice transcripts, without asking users for ratings.
No member yet
Request clusters
Groups differently worded asks, across languages, into one feature-request cluster with counts and sample messages.
No member yet
Failure clusters
Groups conversations where the agent was corrected or ignored, so you fix patterns instead of single chats.
No member yet
Conversation replay
Open any cluster and read the full conversations behind it, with the frustration moments highlighted.
No member yet
Brand identityAI proposal
Check the name against trademark registers before using it.
Palette
#1F5F5BDeep Teal · primary#F6F1E7Paper · surface#1F2A2EInk · text#E0673BEmber · accent, frustration#D9A441Straw · data
Typography
Aa
Murmur
ABCDEFGHIJKLMNOPQRSTUVWXYZ 0123456789
Hear what your users meant, not just what they typed.
PricingAI proposal
The free Starter and $499/month Pro come from a public claim in the Hacker News post, not a verified page. PostHog's verified pricing page lists prices on request. Treat these tiers as a starting point to test.
Starter
Free
- Low message volume
- Frustration and request clusters
- Short retention
Pro
$499 a month
- Higher message volume
- Longer retention
- Weekly digest and release check
- Privacy masking
Enterprise
Contact us
- Highest volume
- Custom retention
- Security review and SSO
Would you pay Pro · $499 a month?
Your answer tells PWAS which prices people accept. It changes no score.
No answers yet.
What it could be worthEstimate
What a software business making this much sells for today, from real deals. Pick a monthly revenue.
If it makes
Worth about
$194,000
Sale range $132,000 to $263,000
$120,000 a year in revenue
From 87 software businesses listed and sold on Empire Flippers in the last year, revenue vetted by the broker: they sell for about 19× their monthly revenue, most between 13× and 26×.
Measured on profit, the middle deal is 32× monthly net profit: about 2.7 years of profit.
An estimate, not an offer or financial advice. A real price depends on profit, growth, churn, how much of the work is the owner's, and the buyer.
How it would growAI proposal
How the product would reach new customers through its own use, and how the first ones would be found. A plan to test, not a result.
- Built-in loop
- No loop in the product. Murmur sits inside a team's own tooling. Digest emails go to teammates who already use it, but it does not publish output to outsiders. Growth comes from content, community posts and direct sales.
- Do they become customers?
- Test whether showing real cluster findings in public write-ups (with permission) brings in teams building agents who then sign up for the free tier and upgrade.
- First customers
- Product engineers at seed to Series A startups whose main product is a chat or voice agent and who handle at least 100,000 messages a month.
- Where to reach them
- Hacker News and dev forums: Post worked examples of clusters found in anonymized agent logs, and offer free analysis of a sample.
- Agent builder communities: Join Discord and Slack groups for agent builders, and help people find frustration patterns in their data.
- Direct outreach: Email founders of public chat and voice agent products with one cluster found from their public demo.
- First milestone
- Five teams send production traffic for four weeks and two of them pay for Pro.
Why people already pay for this
- VerifiedHacker News post received 85 points and 52 comments
- Public claimChat and voice products lack the same analytics framework as web apps; clicks and funnels become less useful when the interface is language.
- Public claimUsers rarely give explicit feedback on agents; instead they curse, ask again, correct the agent, or leave silently.
- Public claimProduct engineers have technical visibility from latency, errors, and traces but must guess whether users got what they wanted.
Tech stack
Every proposal builds on the same PWA base, installable from the browser on any phone or computer. Dashed boxes are what this app adds.
- AppNext.js and TypeScript
- PWAManifest and service worker
- UITailwind CSS
- Data and authSupabase Postgres
- PaymentsStripe
- EmailResend
- HostingVercel
- This app · Language analysisHosted LLM and embedding modelsNeeded to label frustration signals and cluster differently worded requests across languages.
- This app · Event ingestionQueue-backed intake with OpenTelemetry endpointMessage volume can be large and bursty, so intake must not block on analysis.
RisksAI inference
- AIFrustration and intent labels may be wrong or inconsistent across languages and voice transcripts, and teams will stop trusting clusters they cannot verify.
- CommercialObservability and eval vendors could add similar conversation analytics. The $499/month Pro price is only a public claim and may not hold.
- DataConversations hold personal data. Customers will need masking, retention limits and deletion before they send production traffic.
- TechnicalAnalyzing every message with a language model is costly at high volume, which may squeeze margins on low-priced tiers.
Everything marked AI proposal or AI inference was written by a model from the evidence on the opportunity page. The proposed name must be checked against trademark registers before it is used.