For software engineers who just became managers

See how you actually lead.
While the mistakes are still free.

LeadSim is an AI simulation of the team you will actually run next: five software engineers and a fleet of coding agents. A conflict, a slipped deadline, a senior drifting toward the door, an agent that ships faster than your junior learns. Nothing follows a script; consequences compound from your decisions and land when they would in real life. What you keep is a report on the patterns you fell into, evidenced by what you actually did.

Join the September waitlist See what you walk away with
Try it 01

One scene. Your own words.
See what it tracks.

Consequences are not scripted — they compound. Every decision changes the state of your team, and that state decides what hits you next. One scene below, played for real.

leadsim · sample scene · week 6 of 12
Thursday · 18:40

The demo is Monday morning. The build has been red since lunch: the payments integration Julia owns keeps failing under load, and she has said "done by end of day" twice this week. Marek, your senior, could fix it — he starts two weeks of leave tomorrow and has stopped offering. You know the code well enough to fix it yourself tonight, in about two hours.

A real decision takes more than a few words. Write it the way you would say it.
Your words stay on this page. Nothing is sent — to me or anyone.
Locked. In the full run, three things happen now.
01 · Record
Your exact words go on the record.
Not a paraphrase. If a pattern shows up over the run, this decision is quoted back to you in the report as evidence — the sentence you just wrote, verbatim.
02 · Debt
A debt may have just opened.
The most common answer here is some version of "I'll fix it myself tonight." It feels like protecting the team. The library calls it Hero Reflex — and you will not feel it land today.
03 · Consequence
The cost arrives week 9, not tonight.
Julia's estimates stop shrinking, because her scope did. The next red build reaches you on demo day — she routed it around you, to Marek. The report connects that morning to this Thursday.
Join the September waitlist See the sample report
Problem 02

The one thing nobody gives a manager is
a mirror.

60%
Gartner research, via Wharton (2024)
of new managers fail within their first two years
The same share got no training when they stepped into the role. You learn the job on live people, and the feedback arrives late: a resignation you did not see coming.
Nobody
reviews how you manage, only what you ship
Two managers ship the same quarter. One built the team up getting there, the other spent it down. On paper the quarters are identical, so nobody asks.
Invisible
blind spots that feel reasonable from the inside
Avoiding a hard conversation feels like protecting the team's focus. Micromanagement feels like being helpful. You cannot find these by asking yourself.
The next team 03

Half your next team
won't be human.

Agents are taking the well-specified, verifiable work — the exact tasks juniors used to learn on. What is left for you is deciding, verifying, and owning the outcome. That is a different job, and nobody is training anyone for it.

67%
Stanford Digital Economy Lab · ADP payroll data
fewer junior engineers hired than in 2022
The junior who used to take the routine tickets is not being hired. What replaced them does not need a career path. Your remaining people still do.
Delegate
to the agent — or to the human who needs the rep
The agent is faster tonight, every night. Every task you route to it is a rep Julia never gets. Growing people used to be free; now it costs you velocity, on purpose.
Sign-off
you own work that nobody on the team wrote
The fleet ships, the tests are green, and the name on the release is yours. Responsibility did not get automated. It concentrated — on you.

The September run plays this out. In week 7 of the simulation, an agent fleet joins your team — faster than your junior, cheaper than your senior, loyal to the spec instead of to you. The dilemmas above stop being hypothetical, and your report shows how you actually handled them.

The report 04

The report does not tell you who you are.
It shows you what you did.

Several pages, written from your record: the patterns you fell into and the specific decisions that evidence each one, quoted back to you. Below — an excerpt from a real report. It is the founder's own test run, which makes him the only person on earth who could publish it.

leadsim · private report · excerpt from a real run
Quarter report · Project Phoenix · 12 weeks Reader: the player. That is the entire list.
"You moved fast when the problem was external and went quiet when it was sitting across from you."

Every claim below is computed from your decisions and what followed. No personality reads.

83%
scope delivered · 17 points behind plan
1
departure · senior frontend developer
8 / 11
moments left unanswered
3 / 4
promises made, left unpriced
One of the three most expensive decisions · week 5
"Please take the time off and focus on going back to convalesce, this is the most important right now. We will handle the release."
Kasia had an accident. This was the kind call — and the right one. She went offline without guilt, and the seniors self-organized around her tickets. But being carried through a hard moment left a junior who overworks to prove herself feeling she owed the team. She came back early; that night her commits landed at 10:47 PM and 11:20 PM, and she picked up two extra tickets without flagging them. The bill: her overcommit pattern ran uninterrupted into the launch crunch, with seniors absorbing the spillover each time.
From signal to action · the wrap
External signals — a competitor launch, a CEO call, a stakeholder reframe — drew a response within the same day. Internal signals — Marek's slipping quality in week 4, his withdrawal in week 11, Anna's DM asking if he was okay — received no response at all.
Nobody else receives a copy. No dashboards, no aggregates, no admin view.

Above: a fragment of a real report — verdict, the run's numbers, one of its three most expensive decisions. Decisions are quoted back verbatim: what the player actually wrote in the moment, not a summary of it. That is what "evidence" means here.

The report cannot flatter you. Every pattern it can name is one of 35 in a fixed library written before you ever played: Hero Reflex, Shadow Micromanagement, Ostrich Silence. The model selects from that list and quotes your own decisions as evidence — and when one playthrough did not see enough to be sure, it says so.

Join the September waitlist

Privacy 05

Private
by design.

 
Workshop / coaching
LeadSim
Who sees the result
Your manager and HR, at minimum
You. That is the entire list.
When you do it
A date in someone else's calendar
Whenever you sit down. Stop and come back.
Can you tell if it worked
No. You feel more confident, which is a different thing.
Play it again in six months and compare.
What you keep
A certificate and a book recommendation
A report on decisions you actually made

The report goes to you and stops there: no dashboards, no aggregates, no admin view. If the person who decides your promotion could read it, you would play the version of you that promotes well. Playthroughs run through the Anthropic API, which does not train on your data.

Behind it 06

Built by someone who has watched this fail dozens of times.

Founder · LeadSim
Michał Orzechowski
20 years in software engineering · Mentor to dozens of IC-to-manager transitions

Twenty years inside software engineering organizations, watching the same scene: a strong individual contributor gets promoted, everyone crosses their fingers, and six months later the team is fragile — and nobody can point to what went wrong. LeadSim makes the cause visible before it costs you a person, a quarter, or both.

"LeadSim is the tool I wished I had when I was promoting people who deserved the opportunity but weren't ready for the reality."
Connect on LinkedIn
Waitlist 07

Ten places in September.
The waitlist is open.

The pilot is a trade, stated plainly. You get the full run, your private report, and my direct line while the product is still being shaped. I get your honest feedback: at each of the run's three milestones we meet and talk through what happened and how it read. Those three conversations are the price of the place.

Access opens September 2026. Ten places, and they go by fit, not by queue: it is built for people running a real team right now. Join the waitlist below. I read every entry myself and reply personally the week of September 1 — if the fit is there, we talk for fifteen minutes and the place is yours.

The pilot is free; the three milestone conversations are the deal. I reply to every entry personally the week of September 1 — places go by fit, not by queue, so joining early loses you nothing and being thoughtful in the form gains you a lot.
You're on the list.
I read every entry myself and I will come back to you personally the week of September 1. If we are a fit, the next step is a fifteen minute call and one of the ten places.