Working beta

An agent that
interviews your
whole company.

Bell calls every employee, runs a 10-minute conversation, and turns it into a structured map of how your company really uses AI.

The gap

Surveys are dead. Nobody fills them in — and they never told you the truth anyway.

Self-reported forms get skipped, rushed, or gamed. People round up, tick the safe box, and skip the part where they admit they have no idea what’s sanctioned. A real conversation goes where a form can’t — it asks the follow-up, hears the hesitation, and surfaces the actual picture of how work gets done.

How it works

01

Invite

Drop in your team — Bell emails each person a private link.

02

Interview

A voice agent runs the same structured 10-minute conversation, 24/7, no scheduling.

03

Insight

Every call is extracted into a maturity score, tools, blockers, and sentiment.

What you get

Every conversation, rolled up into one honest dashboard.

Maturity by department

/ 100
Engineering
78
Sales
64
Marketing
71
Operations
42
Finance
35

Overall

58/100

Emerging adoption. Strong pockets, uneven across the org.

Sentiment

Optimistic61%
Neutral24%
Skeptical15%

Top blockers

No clear policy on what's allowed41%
Don't know which tools exist33%
Worried about data privacy26%

Notable quote

“I use it every day for drafts, but I’d never put a client name in it — nobody told me if that’s allowed.”

Account Manager, Sales

LiveKit · Deepgram STT · Claude · Deepgram TTS · Inngest · Postgres (RLS)

Working beta — end-to-end

The thinking

The most honest research happens in conversation, not in a form field.

The signal a company needs lives in the things people only say out loud — the workaround, the quiet doubt, the tool nobody approved but everyone uses. So I taught an agent to have that conversation, and to have it at the scale of a whole company — every employee, same depth, in days, not quarters.