How Talklet works — three ways to use the product: Talk to AI, Train with AI, and Sunday Live human group meetings.
Three ways to use Talklet. Pick one, walk through how it works, sign up when something here makes you want to try it. No demo, no waitlist — five minutes free across all characters.
Pick a character. Talk for as long as it takes.
Seven AI characters, each one a different conversation. Voice in, voice out. They listen, they ask, they push back when it matters. After every call you get a short reflection note in your account — your own thinking, captured.
Each character has a voice, a temperament, and one thing they want to know. See the full lineup on the home page →
How a Talk session goes
Five things to know before you start.
- 1Five minutes free across all characters. No card, no friction. Use them all if you want.
- 2Voice-first. You talk, the character talks back. No typing, no slow back-and-forth. It feels like a phone call that nobody else is on.
- 3They push back. Most AI chatbots agree with whatever you say. Talklet's characters disagree on purpose when the answer you need is the one you didn't want to hear.
- 4You hang up when you're done. No timer pressure during the trial. When the five minutes are up, the call ends cleanly and you can buy more time or come back tomorrow.
- 5A short reflection note is written to your account — what was on your mind, what you worked out, what's still open. Yours to keep.
What the reflection note looks like
You came in tired and left with one question.
You noticed something about your week before you said it out loud — that the tiredness isn't the work itself, it's the feeling of never being done. You named the difference between resting and not-working. That distinction matters.
The harder question is still open: what does resting actually look like for you, in practice, this week?
Worth coming back to
- What’s one thing you’d do this week that isn’t "useful"?
- Who in your life rests well, and what do they do differently?
Sample · written automatically after the call · no audio kept
A small thing that matters
Your character remembers your last conversation.
Not the world's. Not other users'. Just yours. When you come back to Mira next week, she knows what was on your mind last time and where you left it. Most AI chatbots forget you the second you close the tab — Talklet's characters don't.
Rehearse the conversations that actually matter.
Work or home. The salary review you've been postponing, the boundary with a parent, the feedback you keep softening, the “no” you owe someone. The opponent doesn't fold. You get a scorecard at the end.
Workplace · 5 scenarios
Salary review — Marcus
Calm, evidence-driven manager. Won't be flattered into a yes.
Giving hard feedback — Jordan
Defensive teammate. Takes it personally first.
Saying no to your boss — Priya
Pressuring senior leader. Will reframe and re-ask.
Resigning — Devin
Surprised, then strategic. Tries every retention move.
Speaking up to authority — Pia
Senior leader. Dismisses until you push with evidence.
Personal & family · 5 scenarios
Telling an aging parent they need help — Anne
Warm, hurt, then in denial. Plays the autonomy card.
Asking a close friend for real help — Sara
Distracted, minimising. Helps only if you push past the deflection.
The recurring money fight — Erik
Defensive partner. Brings up the past. Trigger-prone topic.
Telling family a decision they won't approve — Lena
Loving and judgmental. Plays the identity card.
Reconnecting after a long silence — Tom
Guarded. Tests whether you mean it before opening up.
How a Practice session goes
Five things to expect.
- 1Pick a scenario. Four work scenarios today; more coming.
- 2Set the intensity. Calm · Firm · Hyper-defensive. How hard do you want them to push?
- 3Add context (optional). Two or three sentences about your specific situation. The opponent stays in character but adapts.
- 4Have the conversation. They won't fold. They won't agree just because you sound certain. They'll reframe, push, and try every angle their character would.
- 5Get the scorecard. Clarity, evasion, escalation. Plus the lines worth trying differently next time.
What the scorecard looks like
Clarity
7 / 10
You named the number early and didn't soften it.
Evasion
3 / 10
Skipped one direct question about timing.
Escalation
Low
Held the tone. No defensiveness on either side.
Lines worth trying differently
Sample · written automatically after the rehearsal · no audio kept
Two small groups. One question each.
Every Sunday.
The AI is for the everyday. Sundays are for the room. Two parallel meetings each week — one for reflecting on what's on your mind, one for the hard conversations you've been rehearsing. Up to six people each, thirty-five minutes, just click and join.
Reflection · This Sunday · 20:00 CEST
“How do you feel about the increasing use of AI in daily life?”
35 minutes · up to 6 people · fully booked
Join the meeting →Debrief · This Sunday · 20:00 CEST
“What's one thing you'd say differently next time?”
35 minutes · up to 6 people · 1 of 6 seats taken
Join the meeting →How a Sunday goes
No host. No script. No pressure to perform.
- 1Monday morning we email you this Sunday's question. That's the only email you'll get from us.
- 2Reserve a seat if you know you're in. Or just show up at 20:00 — seats stay open until they fill.
- 3Click join. No download, no host, no waiting room. The room opens, you're in.
- 4Talk for 35 minutes. Up to six people. Whoever wants to speak speaks. Nobody runs the meeting.
- 5The reflection writes itself. Same as the 1:1 sessions. Your words, captured. No audio kept.
Why two tracks
Reflection rooms think out loud. Debrief rooms compare notes on practice.
Reflection is for people who came in carrying something — a decision, a feeling, a question that didn't resolve. Talking to other humans about it does what AI can't.
Debrief is for people who rehearsed something hard this week and want to compare notes. Did you have the conversation? What happened? What surprised you?
Both run at the same time. Pick whichever fits the week you've had.
Nothing leaves the room.
Most platforms make money from what you say. Talklet is built on the opposite trade. You speak freely because nothing exists that can leak, surface in a feed, or embarrass you a year from now. Here's the full picture.
What we do keep
Your reflection notes & scorecards
Written summaries after each call, in your account. Yours to read, export, or delete any time.
What each character remembers about you
So Mira knows what was on your mind last time. Stored only against your account. Never aggregated, never used to train models, never seen by other users.
Meeting metadata
When you joined, how long, which character. Used to bill correctly and operate the service.
Your account & billing details
Email, plan status, payment records via Stripe. EU-processed. Exportable. Deletable.
What we don't
No audio kept. Ever.
Voice runs through the model in real time and is gone. There is no replay button because there is nothing to replay.
No transcripts after the reflection writes
Live transcripts exist only long enough to generate your reflection note. Then they're deleted. Nothing raw stays.
No feed, no algorithm, no recommendations
Talklet doesn't have a wall, a timeline, or a “for you” tab. Nothing about you is shown to anyone else.
No training on your data
What you say to Mira never trains another user's Mira. Your reflections are not in the model.
No notifications you didn't ask for
One email on Monday with Sunday's question. That's it.
The most useful conversations are the ones you got to practice first.
Talk with a character tonight. Rehearse the hard one this week. Bring the rest to a real room on Sunday.
Start free →By joining you agree to our Privacy Policy. Unsubscribe anytime.