For L&D and HR teams
Your managers learned the framework. Now let them practice it.
Realistic voice practice for the hard conversations in your program, scored against your own criteria. You see who’s ready before the real conversations start.
On a 30-minute call we build one of your program’s conversations live, and you take the first round.
- Start with
- One program and a cohort of 10 to 30 people
- Runs for
- Six weeks, measured against goals you set in week one
- Built from
- Your framework, your playbook, your standards
| Criterion | First round | Latest round |
|---|---|---|
| Describes the issue specifically | 58% | 92% |
| Explains the impact on others | 46% | 83% |
| Asks for their side before solving | 21% | 67% |
| Agrees on a next step together | 38% | 88% |
Asking for the other person’s side is still the most common miss. Aim the next workshop there.
The gap
Workshops stop where the hard part starts.
Role-play doesn’t scale
A facilitated role-play day gives each person one or two attempts, once. The real conversation comes weeks later.
Homework doesn’t happen
Nobody logs into an LMS to practice telling someone their work isn’t good enough. So they practice on a real person.
Completions prove nothing
Leadership wants to know whether behavior changed. Attendance and quiz scores can’t answer that.
How it works
Your program, with far more practice in it.
- 01
Bring your material
Upload your framework, playbook, or rubric. Talkthrough drafts the scenario, the person on the other side, and the criteria in your words. You edit them.
- 02
Assign it to a cohort
People get a link, read a short briefing, and start talking. A round fits in a coffee break, on a phone. Due dates and reminders do the chasing.
- 03
They practice and retry
The other person has goals, a mood, and things they won’t volunteer. After each round, people see what they met and missed, and retry the moment that went wrong.
- 04
You see who’s ready
Every round is scored against your criteria, with the words that earned it. Spot the skill each cohort misses most, before it matters.
See it with one of your own conversations, on a 30-minute call.
Request a demoWhat a learner sees
Feedback on the moment, and a way back in.
Every criterion points to what the person actually said. When something missed, they replay that moment and try it again, instead of starting the whole conversation over.
- Describe what happened
It came in two days after the deadline we agreed on.
- Explain the impact
Priya had to present without the numbers.
- Ask for their side first
At 02:41 you suggested weekly check-ins before asking what happened. Daniel went quiet after that.
Retry from 02:41 - Agree on a next step
Let’s go through the timeline together on Monday.
The AI isn’t the coach. Your people are.
Talkthrough is a practice environment, not an AI coach. Your facilitators and managers stay in charge of what is practiced and what good looks like. Talkthrough gives people the repetitions a live session can’t fit.
- You set the standardCriteria come from your rubric, in your language. The other person never sees them, so it can’t coach anyone toward a score.
- Coaches review the workRead or listen to any assigned round, add a note, and change a score when you disagree.
- Private by roleLearners see all of their own rounds, and so does their coach. Admins see progress and scores, not what was said.
- Keep your live role-playTalkthrough adds practice before and after it, right before the real conversation.
Six weeks, one program, your data.
The pilot ends in a readout against the goals you agree on in week one, so you walk into the budget conversation with your own results.
- Week 1
Agree on the goals
Pick the program and the conversation that matters most, and decide what success looks like.
- Weeks 1–2
Build with us
We build the first scenarios and criteria with you from your existing material.
- Weeks 2–6
Run the cohort
Assign practice to 10 to 30 people and watch who practices and what changes.
- Week 6
Read out
Completion, and the change from first to last round on each skill.
The first step is a 30-minute call. No preparation needed.
Request a demoWhat L&D teams ask first.
Request a demoWill people actually do it?
There’s no LMS to log in to. People get a link, read a short briefing, and start talking. A round takes about seven minutes and works on a phone, even on a commute.
Can we trust the scoring?
Scores are against your criteria, and every one points to the words that met or missed it. Coaches can change a score and say why, and learners see both.
Will it sound robotic?
Try it on the call. The other person has their own goals and mood, and reacts to what the learner actually says. You choose how tough they are.
Where does our data go?
OpenAI runs the live voice conversation and scores rounds. Anthropic’s Claude writes the feedback and builds scenarios. Both are used through business APIs, and your material and transcripts are not used to train models.
What about our security review?
Bring your security team in early. We’ll walk them through where data goes and answer your questionnaire, and tell you plainly what’s in place today and what’s on the roadmap.
How is it priced?
Per learner, per year, with a fixed-fee pilot for your first program. We’ll go through the numbers for your cohort on the call.
Request a demo
Bring the program you’re running next.
On a 30-minute call we build one of its conversations into a practice scenario with you, you try it yourself, and we sketch what a six-week pilot would look like.
- 5 minYour programWho it’s for, and the conversation that matters most.
- 12 minYour scenarioWe build the other person and the criteria from your material.
- 8 minYour first roundYou practice, see the feedback, and retry a moment.
- 5 minYour pilotWhat six weeks on this program would look like, and who’d take part.
Thanks.
We’ll be in touch within one business day with times for your demo.
If there’s a conversation you’d like us to build, reply to our email with it. It’s the fastest way to a useful first scenario.
Take the product tour