A Frank pilot, seen from the inside.
These are the four weeks of a pilot, step by step: what you hand over, what gets measured, who sees what, and what gets signed before the first session. It is exactly what will happen in your company, written down before it starts and with no hidden steps.
The four weeks of a pilot.
A small group, scenarios co-designed with you, and one measurement at the start and one at the end, with the same definition at both moments. This is what happens, in the order it happens.
The pilot runs on one of the existing programs or on one built from your company’s own documents.
- Week 0
Documents and scenario approval
You hand over what your company already has: culture handbook, sales playbook, pricing policy, CRM notes. Frank reads that context and proposes scenarios with your products, your objections and your numbers inside. You correct them and approve the ones that go in. None is used without your approval.
An approved program with its stories, and the group that will take part.
- Week 1
Baseline
Each person does their first voice session. Frank learns how that person speaks before comparing them with anything: that is their starting point, and every later measurement is a change relative to it. An accent, a disfluency or an unusual timbre are not noise. They are the baseline.
Each participant’s individual baseline. Private.
- Weeks 2 and 3
Practice on the stories
Sessions of five to ten minutes on the program’s stories. The counterpart arrives at each session carrying what happened in the last one, and the story branches depending on how each person held the conversation. There is no coach and no advice during the session. The report comes at the end, in words, and only the person who practiced sees it.
A private report per session for each person.
- Week 4
Reassessment and aggregate report
Each person does a session on a comparable scenario nobody saw during the pilot: same nature, same length, new situation. It is measured with the same definition as in week 1 and compared against that same person’s baseline. The organization receives an aggregate report on the group, never anyone’s individual report.
The aggregate report for the organization and the final report for each person.
No step gets skipped. Without week 0 the scenarios aren’t yours; without week 1 there is nothing to compare against; without week 4 there is a single snapshot, not a before and after.
What is measured before and after.
The same definition in week 1 and in week 4. Each person is compared with themselves, not with an average or an external ideal.
What is measured
Observable conversational behavior, and descriptive acoustic signals compared against that same person’s baseline, with signal quality and uncertainty stated.
- Assertiveness
- vocal energy and firmness
- Engagement
- prosodic variability
- Conflict handling
- pace control under pressure
- Authority
- resonance and spectral balance
- Active listening
- turns, pauses and overlap
What stays out of the report
- No diagnosis of emotions, personality or internal mental states.
- No prediction of job performance.
- No selection, ranking or automated employment decision about a person.
- Individual reports are private. Organization views use aggregated data.
The reason for measuring against a person’s own baseline instead of an ideal is in why Frank.
Who sees what.
Three parties take part in a pilot and none sees the other two’s data by default. This is written into the contract before the first session.
The person practicing
Their full report, told in words: what they held, where their voice tightened, what to practice next time. They can delete their sessions before the retention period ends.
The organization
How the team and each person are moving relative to their own starting point, not against an ideal.
The Frank team
What it takes to run the pilot: that sessions run, that the audio quality is sufficient, and that the aggregate report goes out on time.
Audio is stored with the person’s consent for a maximum of 90 days. Then it is deleted. The report is kept without the audio.
What gets signed before the first session.
A confidentiality agreement. It puts in writing what would otherwise sit in an email, and it is the first thing any legal team will ask for.
- 01
What is stored, for how long, and who can see it.
- 02
That no data from the pilot is published without your company’s written authorization. Not by name, not anonymized.
- 03
That individual reports feed no employment decision: not hiring, not promotion, not dismissal.
- 04
That the group, the scenarios and the time window are set before the pilot starts, not after seeing the results.
If your legal team needs to adjust a clause, it gets adjusted before the start, not during.
The standard a result meets before it is published.
A result is published only if it is auditable: who authorized it, with what definition it was measured, over how many people and between which dates. It is the same bar worth demanding of any number in this market, ours included.
A publishable result meets these three conditions, no exceptions.
- 01
Written authorization from the client
With the company’s name and the date. An anonymous case can’t be audited, and what can’t be audited doesn’t get published.
- 02
Before and after measured with the same definition
The same dimensions, computed the same way, in week 1 and in week 4. If the definition changes between the two moments, the comparison is void.
- 03
Declared sample size and time window
How many people, how many sessions, and between which dates, next to the figure and not in a footnote.
Every figure arrives with the company that stands behind it and the date, right next to it. The method that produces it is the one you just read, step by step.
Start these four weeks with your team.
A demo takes 30 minutes, needs nothing installed, and ends with a scenario already written for your company.