Make customer decisions with evidence,
inside Claude and ChatGPT.

Built for small teams that talk to customers every week.

Bring the calls, emails, and surveys you already have. UpSight reads them, finds what people keep saying, and shows you the exact quote behind each answer.

Start freeBook a call
Bring in
  • audio and video files
  • transcript files
  • recorded calls
  • built-in surveys
  • Otter
See which path your customers support.

Why connect an AI assistant to your customer conversations?

Claude and ChatGPT are good at reasoning and bad at knowing. Ask either one “what are our top pricing objections from VP-level buyers?” and it will produce a confident, generic answer, because it has never heard a single one of your calls.

The answer exists. It is in a recording from three weeks ago, in a survey response from last quarter, in the second half of a discovery call nobody rewatched. UpSight already turns those into evidence: quotes tagged by speaker and timestamp, clustered into themes, linked to the person and the account. The MCP server hands that evidence to your assistant, so the answer comes back with the receipt attached.

In short: the model does the reasoning, UpSight supplies the customer truth, and every claim links to the moment it was said.

Conversations in. Answers with the quotes behind them out.WHAT YOU HAVECustomer callsInterviewsSurvey answersEmailsMeeting notesDocumentsUpSightReads it allWHAT YOU GETThe answer“Fix onboarding, not pricing.”What people keep sayingA one-page briefThe quotes behind itevery answer links to its source
Your conversations go in. Clear answers, with the quotes behind them, come out.

What can the assistant actually do?

Full read and write, not just search. One OAuth connection covers every project in your workspace.

Conversations
Search, read transcripts, list by date, participant or segment
Insights and themes
Query themes, read the evidence behind them, create new insights from patterns
Tasks
Create, update, complete and list action items across projects
People and organizations
Look up contacts, add notes, link evidence to accounts
Opportunities
Create opportunities, update stages, attach conversation evidence

What does it look like in practice?

Three prompts teams run in the first week.

Prompt 1

Find the evidence behind a business question

Across our last 30 conversations, what are the top pricing objections from VP-level buyers? Cite evidence and confidence.

Prompt 2

Generate segment-specific follow-ups

Draft 3 follow-up emails for clinic ops leaders in the West region who mentioned onboarding friction. Keep each under 120 words.

Prompt 3

Turn signals into a plan

Create a 2-week plan from our top churn-risk themes: 5 tasks, owners, expected impact, and one research validation step per task.

The pattern that works: Goal (the business decision) → Scope (segment, timeframe, source) → Output (format) → Constraint (cite evidence, give confidence, name the next action).

How is this different from pasting transcripts into a chatbot?

Coverage
Pasting: Whatever fits in the context window
UpSight: Every conversation in every project
Provenance
Pasting: The model summarizes; you trust it
UpSight: Every claim links to a timestamped moment
Write-back
Pasting: Copy answers out by hand
UpSight: Create tasks, insights and opportunities from chat
People
Pasting: Names in a transcript
UpSight: Persistent person and account records that grow with each call

Which clients work?

Any client that speaks MCP. Documented set-ups today: Claude (Desktop and claude.ai), ChatGPT (Connectors), OpenClaw, and any other MCP client. Set-up is the same three steps.

  1. Step 1

    Add an MCP server with the URL https://getupsight.com/mcp in Claude, ChatGPT, OpenClaw or any MCP client.

  2. Step 2

    Click Sign in with UpSight and approve access. All of your projects are authorized at once.

  3. Step 3

    Ask something. A good smoke test: "Summarize top churn signals this month with evidence."

OAuth handles scopes, revocation and team roll-out. If your client can’t do OAuth, a project-scoped API key works as a fallback. Full instructions, client by client, are on the set-up page.

What does it cost?

Free

$0

Read-only MCP access, one project, five AI analyses a month, unlimited recording and transcription.

Pro

$29/mo

Full read and write for your agents, unlimited AI analysis, the standard lenses (BANT, JTBD, competitive) and the AI CRM. Single user.

Team

$39/user/mo

Shared evidence, custom lenses, Slack and Teams push, SSO and audit log. Two-seat minimum. Stakeholders view for free.

Current details, including annual pricing, are on the pricing page. Analyzing calls without an assistant in the loop? See customer call analysis.

Every answer shows its source.

Click any finding to open the quote, who said it, and when. Every answer points back to the quotes behind it, so you can check it before you act on it.

Frequently asked questions

Good evidence gets argued away, lost between meetings, or buried under the next task.

Teams were doing the hard part. They talked to customers. The truth still got lost. That's why I built UpSight.

UpSight keeps the quote, the source, and the next step together, so you can act before it gets lost again.

Rick MoyFounder, UpSight

Start with the calls you already have.

Upload a few recordings and ask your first question. Free to start.

Start freeBook a call