An advisory board that disagrees with you.
A panel of grounded personas: the room you’d want for the call you’re making alone, but can’t afford. It argues your hard call in the open, then turns the table and grills you before the moment that counts.
Facing the moment that counts? Step into the Grill Room →
The questionI'm 38. I do alright, but I dread Mondays. Chase the promotion, or leave and build the thing I actually care about?
They agree the dread is real and the promotion won't fix it, but they split on the first move: two want you to write the sentence, two want you to find the person who sees you clearly.
Specificity-first vs depth-first · agreement 0.35The Hiring ManagerTell me about a time you killed something you believed in, with real advocates behind it, so the right two or three could win.
Your answer · retakeI cut an in-app community feature I'd championed. The data said activation was the bottleneck. I showed the eng leads the funnel, redirected the squad, and activation jumped 14%, time-to-value −30%.
One panel. Two ways to use it.
One side counsels you. The other side comes for you.
Bring the call you’ve been turning over alone.
It answers in opposing stances, names the disagreement out loud, and remembers the call you made.
It comes for you.
The Grill Room inverts the panel: the advisors ask, then critique every answer. Come back and your score climbs.
Upload a JD, resume, book, PDF, or notes once. Point it at any panel, session, or drill. Reusing a file never costs a second slot.
How sharp is your question?
Paste a real call you’re turning over. The panel scores how ready it is for a room of advisors, and offers a sharper version if it can. No account, no sign-up: your question is scored, never stored.
How the panel actually works.
Five moves, in the open, from a half-formed question to a call you can defend, with the disagreement kept on the record.
01 / SHARPEN The panel sharpens the ask
Before anyone answers, the panel reflects a sharper version of your ask: what’s vague, what’s compound, what you left out. Pick your wording, the sharper one, or a blend; over time you stop needing the nudge.
02 / CHARTER The right room only
Each panel has a scope. A money panel doesn’t pretend to know your marriage; an off-charter question gets flagged or held, so you’re never taking a confident answer from the wrong room.
03 / DIVERGE Each voice answers
Each voice answers independently, in character, grounded in its own corpus. They don’t see each other’s replies. No groupthink, no chorus: distinct reads from distinct stances.
04 / CONVERGE Synthesis names the tension
A synthesis names the live disagreement out loud and flags where the agreement looks suspiciously easy. You leave with the real tension surfaced, not a consensus mush.
05 / SIGNAL You teach the panel
Mark each reply Useful, Off-base, or Not-them; the panel tunes to the calls you keep bringing. Every reply carries a 0–100 signal score with cite chips back to the source. Math you can see, not a vibe.
How the panel actually works.
Five moves, in the open, from a half-formed question to a call you can defend, with the disagreement kept on the record.
The panel sharpens the ask
Before anyone answers, the panel reflects a sharper version of your ask: what’s vague, what’s compound, what you left out. Pick your wording, the sharper one, or a blend; over time you stop needing the nudge.
The right room only
Each panel has a scope. A money panel doesn’t pretend to know your marriage; an off-charter question gets flagged or held, so you’re never taking a confident answer from the wrong room.
Each voice answers
Each voice answers independently, in character, grounded in its own corpus. They don’t see each other’s replies. No groupthink, no chorus: distinct reads from distinct stances.
Synthesis names the tension
A synthesis names the live disagreement out loud and flags where the agreement looks suspiciously easy. You leave with the real tension surfaced, not a consensus mush.
You teach the panel
Mark each reply Useful, Off-base, or Not-them; the panel tunes to the calls you keep bringing. Every reply carries a 0–100 signal score with cite chips back to the source. Math you can see, not a vibe.
The operator on your panel and the contrarian won’t agree, and where they split is the answer.
The cautious voice and the take-the-swing voice both make their case, and you see exactly where they part.
The 11pm question nobody wants to say out loud. Bring it to a room that won’t flinch and won’t flatter.
Now watch it run once.
“I’m 38, earning well, but I dread Mondays. Do I leave for work I’d actually care about?” Four advisors, four stances, one synthesis. Each cites its source and shows how strong the read is.
The dread is real. The decision isn’t binary.
The Operator and the Investor agree: don’t quit without a thing to quit to, and don’t burn the salary before you’ve replaced what it buys. The Skeptic is testing whether the job is even the thing. Sometimes it’s the city, the boss, or burnout, mislabeled. The Mentor cuts to the felt sense: 10 years out, which version do you owe? Take 90 days. Run the experiment in evenings. Decide with evidence, not a Sunday-night gut.
Sit in the chair instead.
The same panel, inverted. Now it’s asking the questions.
Paste the job description and your resume, or the pitch you’re about to give, and the advisors take turns interrogating you, pressing every answer while it’s still safe to find your blind spots.
Coaching with the brakes on.
Answer a question and the panel scores it 0–100 on the spot, then asks why: where the answer went soft, what it assumed, the follow-up a sharp interviewer would have reached for. You can re-take the same question and watch the number move. The cost of a bad answer here is nothing.
SCORE 78· ASK WHY →The room the way it actually goes.
No scores mid-stream. The panel runs the room the way it actually goes: rotating asker, real follow-ups, no hand-holding. At the end, the whole panel debriefs you together — where you were strong, where you wobbled, the one answer to rebuild before the day that counts.
Each advisor asks in character, grounded in your JD and resume. When one finishes pressing, the next picks up.
Every session lands in your thread library, searchable, never truncated.
Why not just ask ChatGPT?
“Isn’t this just ChatGPT with a few personalities bolted on?”
One voice agrees with you. A panel argues in front of you, then turns the questions on you.
One voice. It mostly agrees.
- Agrees, then hedges to “it depends.”
- Forgets the call you made last week.
- Grounded in nothing in particular.
Many voices, in tension, and they grill you too.
- A panel that disagrees, on the record.
- It advises you, then grills you.
- Every answer scored, with visible math.
One person, one calendar.
- Irreplaceable, and booked solid.
- One worldview, not a whole room.
- Hard to repeat, hard to score.
Use lePanel alongside a coach, not instead of one: for the calls that don’t justify a calendar invite, and the reps you’d run before you book the real thing.
Two scores, visible math.
One grades the answer you got back. One grades the question you brought. Both show their work.
How strong the read coming back at you is: three parts, scored on every response and synthesis.
A number that just announces “I’m 90% sure” is a feeling in a costume. Signal is structural, built from agreement, source density, and your feedback over time, so you can argue with it.
A separate 0–100, by axis, before the panel replies.
Not the Signal score. Signal grades the panel’s read; this grades how cleanly you framed the call.
270 advisors, stances face-up.
Each archetype is grounded in its own corpus and holds a distinct view. The same voices advise you, then turn around and grill you.
“I got the offer I’ve been chasing. Why am I hesitating to say yes?”
No real-person advisors at launch. The archetypes do the same job, distinct stances and grounded reasoning, without the names. If a thinker wants to join lePanel with their own corpus and voice, we’ll work with them directly. Join the beta to hear when they do.
Free during beta. Yours to shape.
Closed beta, a small group of founders and operators. No pricing yet. We want to learn what the panel is worth before we charge for it.
Every feature on while we’re in beta.
Plus the custom personas you build yourself.
PDFs, books, transcripts, notes. The panel reads them.
Founder pricing locked when paid plans launch.
Talk to the team. Your feedback shapes v1.
Not used to train models. Export or delete anytime.
What people ask first.
How does the Grill Room work?
It’s the same panel, turned around. Paste a job description and your resume, or your pitch context, and the advisors take turns: one asks, a different one presses on your answer. The asker rotates, so you get the room, not a script. Practice scores every answer 0–100 and tells you why; Real is a clean grill, then a full-panel debrief. Either way it’s grounded in the JD and resume you gave it.
Are these real people?
No. Each advisor is a built persona grounded in its own corpus, so the panel disagrees with you in the open. It also convenes at 11pm without a calendar invite. Each answers independently, with no shared draft to converge on. Then a synthesis names the live disagreement and flags where consensus looks too easy.
There are two scores. What’s the difference?
Signal score (0–100) rates each answer the panel gives you: voice fidelity, agreement, citation density, with cite chips. “How you ask” score (0–100) rates the question you brought: how clearly it’s framed, by axis. One scores what you get back. The other scores what you bring. Independent on purpose.
Do I have to re-upload my resume every time?
No. Upload a resume, JD, PDF, EPUB, docx, or paste text. Do it once, and it lives in your file library. Reuse it across any panel, session, or interview drill. A reused file counts as one attachment slot and never re-bills. When you’re done, two ways to remove it: take it out of the session, or delete it everywhere at once.
Does the panel keep coaching my questions forever?
No, it fades. The “how you ask” score coaches the way you frame a call early on (today: 78, “sharpen specificity: trade ‘soon’ for a date”), then goes quiet as your asks get cleaner. The score is the proof: watch it climb while the coaching tapers. You stop needing the help because you’ve absorbed it.
Do I get a weekly summary?
On the day you choose, a quiet read across your sessions: what you keep returning to, which voices earned their place, where you’re ruminating instead of deciding, and, if you’ve been practising, how your interview answers are trending. Sent the morning you pick, never a feed.
Do my sessions stay around?
Forever, until you archive or delete them. lePanel is built against chat tools’ habit of forgetting: search your whole thread library, reopen any session, pick up follow-ups where you left them. Ask something close to a past call and it stops first, surfaces the earlier thread, and lets you decide: reopen it, or run it fresh.
Can I add my own advisors?
Yes, that’s the point. Books, podcasts, YouTube channels, Substacks, threads, your own notes. Anything text or audio. The agent chunks, embeds, and grounds the persona in it. New material updates the voice automatically.
Why archetypes instead of real people?
Naming real public figures as commercial advisors raises right-of-publicity and endorsement issues that we won’t shortcut. Archetypes give you the same usefulness, distinct stances, grounded reasoning, without the legal grey area. Where individual thinkers want to join lePanel as named advisors, we’ll partner with them directly and credit accordingly.
Will my data be used to train models?
No. Your sources, sessions, and personas are yours. We process API calls with our model providers under terms that exclude training. Export or delete everything any time.
How is this different from a coach?
It isn’t a replacement; a great coach is irreplaceable. lePanel is for the calls that don’t justify a calendar invite: the 11pm decision, the question you’re turning over alone. And it remembers. The Grill Room adds the other half: a panel you can practice against on your own schedule, that scores your climb.
When does it ship?
Closed beta is running now. Public launch summer 2026. Waitlist invites going out in batches. Earliest signups get first access and founder pricing when paid plans launch.
Does it cost anything?
Not during beta. Paid plans will launch alongside the public release, but everyone in the beta gets it free for the duration and locks in founder pricing afterwards.
Get an advisory board that disagrees with you.
Rolling out in batches, decision-makers first. Free during beta. Founder pricing locked for the first 500.