Practice.coach changes every week. Here's what actually shipped — no marketing fluff, written in terms of what you can now do as a coach.
When a voice session ends you land right back on your dashboard, History immediately shows the session as being evaluated, and a notification with a link finds you wherever you are once the score is ready. No more staring at a screen that does nothing.
LorelAI now opens a training call by handing you the floor ("You lead the session — open with your first question") instead of asking whether you're ready. Exactly how it should feel for a coach.
Your training plan, the recommended session and the mode picker are now together on a single page — and Start drops you straight into the recommended session instead of another picker.
On the dashboard and in the training plan, each level now lists its skills — checkmarks on what you've demonstrated, a preview on what's locked. Levels beyond your current one used to say nothing at all.
Practice clients got new portraits that load reliably — and if an image ever fails, you see neat initials instead of a broken icon.
In text practice you can deliberately hold silence with one button instead of typing — the client responds to it, and your use of silence counts toward the Silence as a Tool skill.
Before a session you can set the client's situation ("frustrated with admin work…") and they'll live it — without reciting it back as the goal. Discovering the session goal is still your job as the coach.
A session that's still being scored shows "Evaluating…" in History and on the dashboard instead of an empty score, and flips over by itself the moment it's done.
Simulated clients hold their role better: they don't spill the whole story on your first question, don't nod along to nonsense advice, and don't speak in stage directions. Practice feels more like a real session.
The Start button reacts instantly even on a slow connection, the subscription page loads faster, and several screens stopped re-downloading data they already had.
We added another check that a voice-session transcript can only ever be saved to the call it actually belongs to.
Once you finish the welcome call, the app reliably registers it — before, it could keep prompting you to do it again even after you already had.
When an evaluation took too long, the score could simply never arrive and the page would keep saying "Evaluating". Evaluations now get the time they need to finish — and if one genuinely fails, it tells you so and offers a retry instead of leaving you watching a spinner.
Starting a practice session is now a clear choice between a full session and a short drill on one skill — it used to be blurry. What used to be called "Skill Gym" is now consistently named "Skill drill" everywhere.
When the app recommends a skill from outside your current level, it now says so plainly — instead of implying it belongs to your current phase.
The first time you open a Foundations skill, you get a short why-it-matters explainer with a strong/weak example before you start. Skip it if you want, or reopen it anytime.
In text chat with the AI client, the app now bolds a couple of key words in each reply, so scanning the transcript afterward is faster.
In rare cases, the app couldn't finish preparing a personal opening line for a session — full session, Skill drill, or the paired exercise — and swapped in a generic one instead. It now gets the room it needs to finish properly.
Session detail now shows how you did on the opening, the middle and the close — separate from your overall score. If a session skipped a real opening agreement, you get honest feedback instead of a made-up number.
If a session stalled for hours or days and then picked back up, you'd just get a plain low score with no context. The app now recognizes that and flags it — in the summary email and in the session detail.
An all-day Google Calendar entry (a birthday, a holiday, an out-of-office block) no longer gets matched to a client by first name alone — before, you could get a brief for a session that was never actually scheduled.
In three more places where LorelAI reads real-client transcripts, it now checks whether an idea actually originated with the client — instead of counting it against you as unsolicited advice.
As soon as you hang up a LorelAI voice practice call, the app starts scoring it right away — and takes you straight to the session once the score is ready. Before, you'd wait for the overnight batch.
While you're still in Foundations, the voice-practice button is visible but locked, with an explanation of when it opens up. The welcome call stays available from day one.
The same skill or ICF competency now carries the same name across the dashboard, practice, history and sessions — the naming used to drift between screens.
If you had any practice history, the Skills page would fail to load. It now renders reliably.
Set the session length before you start. A live pacing cue shows where you are — so you actually get to the close, not just the opening.
The persona stopped interviewing the coach and talks like a real person. It holds on to what it already told you — name, numbers, situation.
Describe your typical client once and keep several profiles on file. Pick who you want to work with at the start of any session.
The level-up suggestion is saved with the session. Come back to it a week later in the session detail, not just mid-practice.
Mark the spots on the transcript where you felt stuck. Only then does the AI reveal what it saw — and you compare. It trains your own judgment instead of obedience to a score.
Skill Gym and Blind Spot Training now get the same wide ICF-competency read as a full session.
No more endless spinner. The evaluation retries on its own — and when it genuinely can't finish, it says so instead of going quiet.
Score, ICF-competency evaluation and the ICF Session Observation form in one place — ready to submit.
Grant consent for a single session, your mentor sees it and writes you feedback. Revoke that consent any time.
The name now matches what actually happens there — and the ICF vocabulary.
On your first voice call, LorelAI shows you where everything lives instead of leaving you to hunt for it.
One email a week on what moved. One click to unsubscribe, from every email we send.
Fixed a handful of scoring edge cases — a session that ended early no longer wrongly says you didn't write anything when you did, and short one-question drills are no longer judged as harshly as a full session.
Previously unreachable once you were logged in. Now it's at the bottom of the left sidebar.
A new "Train by voice" button sends you from your plan directly into a LorelAI call focused on exactly what you're working on.
It could previously end before you'd answered the first question. Now it waits.
Columns now stack instead of squeezing side by side.
Print or save the ICF Session Observation form and your mentor's feedback in one click — in a form you can submit as is.
Start a focused voice practice session and LorelAI steps into the client's role for the whole call — instead of coaching you through it as a mentor.
You speak, she answers out loud. No phone, no scheduling — open it and talk.
One number for how often your clients actually shift. It moves when you get better — not when you log more hours.
Where we don't have enough data yet, we say so. No placeholder percentage, no progress bar the app invented.
A portfolio of the recurring patterns in your coaching, each with a recommended drill — and a timeline of how each one moves.
Practice with clients who match your actual book — industry, role, typical agendas — instead of generic stand-ins.
Upload a transcript of a real session and get the same 0–100 score as your practice. For the first time, drills and live work sit on one scale.
See when you're next with a client — and what you're walking into.
Just starting? You get today's drill. Coaching for real? You get your pattern to work on.
Landing, pricing, about, contact and the legal pages in both languages, with a switch in the footer.
Join now and that's your price for good.
Every screen, not half of them. Mobile included.
Your session transcripts never leave your account to train models. We also shipped a round of hardening around login, uploads and sharing.
The field on a client now clearly asks about the client's coaching experience, not yours. Session analysis also factors in your own level, so growth suggestions land where you actually are.
Alongside the challenge question, there's now a second, optional one: what the client hopes to get out of coaching. Fill in whichever fits.
When a session closes with the coach supplying the client's takeaways — without asking first — the analysis now calls it out. It never affects your score; it's just something to reflect on.
The Pattern Review and Certification Path cards no longer flash "unlock in Solo plan" or a zero you don't actually have while they're loading — they wait a beat, then show your real numbers.
There's a new "Refer a coach" item in your account menu. Enter their email, tweak the note we've already drafted, and we send it as a personal recommendation from you. Anyone who joins through your link gets 14 days of Solo; you get a free month once they subscribe. You can see who you've written to and how it went.
Older changes (April–May 2026) aren't listed — the product was in a closed phase back then. Curious what we're building now? Get in touch.