How much leverage
is your team leaving on the table?
what AI can add, and how to spend the gain.
Find your leverage
Your results update as you go and stay on this page. Nothing is sent anywhere unless you ask us to email your report.
Every person on a team has a working relationship with every other person, and each one needs upkeep — questions, reviews, hand-offs, context. Across the team those relationships grow as n(n−1)/2, so each new hire adds less focused capacity than the last. Frederick Brooks described this in The Mythical Man-Month (1975). Ceremonies, status reporting, and approval chains sit on top of it.
Working relationships by team size — your current size highlighted
Per person. Check a typical week's calendar.
27% of the team's week goes to coordination — about 2.1 people's worth of time, or $378,400 a year. Some coordination is necessary. The question is how much your structure requires.
At 8 people, adding one more person adds about 0.75 of a person in focused capacity, holding your meeting load constant. This is a conservative floor: it leaves out onboarding time and the extra meetings bigger teams tend to add.
Streamlining how your current team works recovers about 0.9 FTE of focused capacity — same people, no new hires. The Leverage Multiplier below shows what AI-native workflows add on top.
Streamlined = your current team: ceremonies capped at 2 h/week, status at 1 h/week, outside decision waits halved, same team and upkeep.
A meaningful share of every week goes to keeping the team aligned rather than building. Most of this is recoverable through smaller cells, direct ownership, and fewer approval hops — without asking anyone to work harder.
For some activities, researchers have measured how much less time AI-assisted developers need in controlled conditions. This tool applies those measured savings to the share of focused time your team spends in each activity, and labels the activities where no measurement exists.
The formula is: multiplier = 1 / (1 − capacity released). If AI frees 30% of capacity, your team can achieve 43% more output — the same principle as the productivity gain in the underlying studies. Most measurements come from short, well-defined tasks, so treat the result as an upper bound and compare it with the field evidence shown next to your result.
Calculated for your team of 8, from the Coordination Tax above.
Set each slider to the share of that activity handled with AI assistance — your current level or a target. Time savings are from the cited studies at full adoption.
Process gains only — the baseline.
New features, scaffolding, boilerplate. Measured on a single greenfield task, so treat it as an upper bound for everyday feature work.
Rewriting, restructuring, and improving existing code. Completed in nearly two-thirds the usual time.
Writing technical docs, comments, and changelogs. Completed in about half the usual time.
Unit, integration, and regression tests. A conservative estimate in line with the measured code-writing results.
Reviewing others' code for correctness, style, and security. AI can pre-screen common issues; human review remains essential for context and design.
Investigating unknowns, comparing options, and scoping solutions. No study isolates this work; the field-trial average (+26% tasks completed, about 21% less time per task) stands in for it.
Your team's focused-time output relative to the same work without AI assistance — from published measurements where they exist, and labelled estimates where they don't.
- +39%merged pull requests after companies made a coding agent their default UChicago / Cursor, 2025 (small comparison group)
- +26%tasks completed across 4,867 developers at three companies Cui, Demirer et al., 2025
- +24%merged pull requests among Microsoft engineers adopting command-line coding agents Microsoft, 2026
METR's 2026 follow-up measured 4–18% faster and noted its numbers likely understate the gain: many of the developers who benefit most declined to work without AI for the study.
Every one of these measures AI added to existing teams and processes. None measures a delivery model designed around it.
Across 8 engineers × 12 working weeks × 0% released
- · Speedup rates are from full-adoption conditions in the cited studies
- · Partial adoption scales linearly (50% AI-assisted = 50% of the speedup)
- · Total capacity released is capped at 70% — irreducible non-automatable work
- · Lab findings reflect well-defined tasks; complex or unfamiliar work sees much lower gains (McKinsey 2023: <10% on high-complexity tasks, and some junior developers were slower)
- · Applies to focused time only — meetings and coordination are not accelerated
Leverage is the gain. What you do with it is the strategy.
Your 8-person team goes from 5.8 to 6.8 output-equivalent FTEs with a streamlined process. Set AI adoption in the Leverage Multiplier to add the AI gain.
Build faster, build better, or build cheaper. Most organisations choose a mix.
Compared with your team today working without AI assistance; if you already use AI, part of this gain is already yours. AI figures come from the published studies in the Leverage Multiplier; process assumptions are listed in the Coordination Tax.
Your 8-person team, today spending 27% of its week on coordination, goes from 5.8 to 6.8 output-equivalent FTEs.
The link carries your inputs only — share it with your team or leadership.
Email me my report
Your numbers, the three ways to spend the gain, and a link back to these results.
Leverage is the gain. What you do with it is the strategy. Build faster, build better, or build cheaper.
The first tool shows how much of your team's week is left for building; the second shows how much faster that building time can go. Together they give your leverage. Some organisations spend it on speed, keeping the team and shipping sooner. Some spend it on quality, putting the freed capacity into testing, security, and paying down debt. Some spend it on cost, delivering the same output with a smaller team. We help you choose the mix that fits your goals, then change the process, the tooling, and the habits to get there.
Talk to us about your goals→Short reads from
real delivery work
Two-to-four-minute notes on what actually changes when teams restructure around AI — concrete before-and-afters, not theory.