Features

From the moment a user says it to the task your agent picks up.

Sense is one pipeline with a lot of doors into it. A call, a screen recording, a widget report, a survey — each lands as a timestamped transcript, gets read against the goals you set, and comes out as work you can prioritize with the quotes still attached. These are the parts, in the order they run.

01capture

Every way a conversation reaches you.

A call you sat in on, a tester's screen, a user hitting friction inside your product, or a question you send out async. Each one needs a different door, and none of them ask the person on the other end to make an account.

Meeting bot

Drop in a meeting link and the bot joins the call, records, and brings the conversation back into your workspace. Offline conversations can be recorded as audio straight from Sense.

Read more
Chrome extension

Record your screen and voice from the browser and get a shareable link. Grab snapshots mid-recording and circle exactly what you mean.

Read more
Embeddable feedback widget

A launcher on your site that starts recording a user's voice and screen in one click. No account, no form. Snapshots can be cropped, drawn on, and commented. Nothing enters your workspace until you approve it.

Read more
Public intake links

Collect feedback from people who have no account and no login.

Read more
Surveys

Async capture at scale when a call is not worth scheduling.

02the evidence chain

Nothing is a summary.

Each layer is built out of the one below it, and the bottom layer is the words themselves. That is what makes a claim checkable instead of something you have to take on faith.

Timestamped transcripts

Every line is tied to its exact moment in the recording. Click a sentence and the video jumps there.

Anchors

Sense extracts the verbatim things users said, each one pinned to its spot in the source. These are the ground truth everything above rests on.

Insights

Findings synthesized from clusters of anchors, so no summary floats free of the words behind it.

Themes

Insights bundle into higher-level patterns, ranked by how much evidence sits behind them.

Goals

Feedback is evaluated against the research questions you set, so extraction is scoped to what you are trying to learn.

03asking across it

The pattern lives in the workspace, not in someone's head.

Once the material is in, you can interrogate it — one source or every conversation in the project at once.

Discuss

Ask a question against a source or an entire project folder and get an answer pulled from the actual material.

Folder-level grounding

Sense synthesizes across every conversation in a project, so the pattern lives in the workspace rather than in one person's head.

04the backlog

Findings you can act on, not a library you have to read.

The end of the chain is work, already prioritizable, with everything behind it still attached.

Auto-drafted backlog

Insights become prioritizable action items. Sense produces the tasks rather than a repository of findings someone still has to read and translate.

Evidence on every task

Each action item stays linked to the quotes and recordings it came from.

Reporter attribution

Every ticket points back to the person who raised it, so you know who to tell when it ships.

05the hand-off

Out to whoever builds it — human or agent.

A task is only useful once it reaches the thing that does the work. Sense hands over the evidence with it.

Copy prompt

Hands a ready, evidence-linked task to Cursor, Claude Code, or Lovable in one click.

MCP server

Query your whole research graph from your coding agent: signals, anchors, insights, themes, and the action-item backlog.

Shareable reports

For the humans who need the narrative, not just the agents writing the code.

All 18 of them, in one workspace.

Create a workspace, turn on the widget or drop a meeting link in, and the chain runs from your first capture. Free to start.