CatalystLTI is a private, voice-first simulator where engineers rehearse high-stakes leadership conversations against an AI actor and receive deterministic behavioral telemetry — the Leadership Telemetry Index — plus a targeted drill.

Is CatalystLTI therapy or a personality test?

No. CatalystLTI is leadership communication practice. It is not therapy, not clinical, and not a personality test.

What does the Leadership Telemetry Index measure?

Two v1 indices: the Presence Index (speech rate, pause discipline, filler density, steadiness) and the Decisiveness Score (commitment language, ownership, and whether you close with a clear next step). Scores are computed deterministically in code; an LLM only writes the narrative.

LTI · eng

Station IC → VP · private lab

Rehearse the conversation before it ships to the room.

CatalystLTI is a voice simulator for high-stakes engineering leadership — scope pushback, headcount defense, the two-minute exec update. One AI actor. Deterministic telemetry. One drill on the signal that slipped.

Start a rep Spec: how it works

Not therapy
|
Not a personality test
|
No employer view

ch-1 · live channelrec ●

pressure onset · t+00:42

Presence

— /100

scored after call

Decide

—

no live scoreboard

Mode

PRIVATE

1 human · 1 actor

Fig. 1 — Channel live; indices computed post-session (deterministic)

Failure modes in the field

Leadership is the part of engineering no one lets you dry-run.

P-01

You found out you were bad at the exec update — in the exec update.

P-02

Your hardest conversations get one take, with your career on the line.

P-03

“Be more strategic” is feedback you can’t put on a test bench.

P-04

Presence and decisiveness are judged constantly and coached almost never.

Control loop · rev A

Five stations. Then close the loop until it’s muscle memory.

S01
01
Load a scenario
A high-stakes leadership moment matched to your stage and weak signals — not a random quiz.
S02
02
Run the channel
One live voice call with a named character. Different voice and agenda each scene. Hands-free optional.
S03
03
Extract telemetry
Behavioral signals from finalized speech. Deterministic rules — no model invents your score.
S04
04
Read the report
Every claim ties to a transcript moment. Presence dimensions + decisiveness, with evidence.
S05
05
One targeted drill
A short rep on the signal that slipped. Then loop. Muscle memory, not a lecture.

Why it holds under load

Close enough to the real room that your nerves show up.

A different counterparty every time

Not the same generic assistant on a loop. Each scenario is a named character with their own voice and agenda — the principal defending his system, the CFO who wants a number, the customer who's done being patient. You learn to read the room because the room keeps changing.

Hands-free channel

Turn on hands-free and CatalystLTI reads the brief, waits for “start,” and runs the rep without a tap — walk, commute, ten minutes before the real one. Best on Chrome or Android; passenger seat, not the driver's.

Instrumentation

Observable behavior. Scored in code.

No LLM invents your number. Signals come from finalized speech and deterministic rules — every claim in the report points at a transcript moment.

Presence Index

How grounded and credible you stay under pressure — six dimensions, so you see which one cost you the room, not a single opaque score.

ComposureClarityAuthorityEngagementResponsivenessConviction

Decisiveness Score

Whether you made a recommendation and committed — strong modals, clear asks, and less death-by-qualification.

Reference frame

Your baseline, not the average.

“Sound more confident” mostly rewards a deeper voice and a better mic. We score how steady you stay when the conversation gets hard — relative to your own calm baseline.

Baseline-relative

Calm opening minutes set your normal pace and hedging. Absolute pitch never touches the score — body size and mic bias stay out.

Recovery time

Everyone flinches when challenged. We measure how fast you return to baseline after each pressure moment — the trainable part.

Six-dimension heatmap

Composure through conviction — each scored, with the focus dimension and the timestamp where it slipped.

CPD ledger

Your reps count toward CPD

Substantive, scored sessions feed a versioned certificate of completion — practice hours you can self-report to your engineering body. Serials re-download; more hours issue a new revision.

Start practising

Isolation · single operator

Built for one person practising in private.

One human, one AI actor

No multi-speaker recording of real meetings. Ever.

No employer dashboard

Your scores are yours. There is no manager-facing view.

Delete cascades

Remove a session and its audio, transcript, telemetry, and report go with it.

Next hard conversation

The next hard conversation is coming. Take the first one here.