Inject a real customer scenario into a live coding session, record everything the candidate does, and let Claude score how they actually used AI to get the job done. I'm opening this up to a small group of teams before a wider launch — request access below.
2.4×
more productive hires
AI-fluent vs. non-fluent engineers
$287k
avg cost of a mis-hire
salary + recruiting + replacement
5
rubric categories scored
per session, by Claude
How it works
No setup required from the candidate. Just a link and a real problem to solve.
Choose from a library of real customer-context scenarios — or author your own. Set the time limit and send an invite link. No installations, no accounts required for candidates.
Customer-context briefing, starter code, sandbox environment — all in the browser.
rrweb captures every keystroke, AI prompt, terminal run, and iteration. We record how candidates use their tools, not just the final result.
Full DOM replay + code snapshots every 30s + sandboxed code execution logs.
Claude Opus analyzes the full recording against a structured 5-category rubric: prompt engineering, solution quality, iteration patterns, debugging approach, and creativity.
Detailed per-category scoring with evidence from the candidate's actual session.
The problem
Leetcode and system design interviews were built for a world before AI tooling. Today, a candidate who can prompt Claude effectively, iterate rapidly, and chain tools together will outperform a "brilliant" engineer who can't — by 2–3×. You have no way to measure that today.
Sample Maite Score Report
84
/ 100 — AI Tooling Score
"Strong prompt iteration patterns — rewrote the initial Claude prompt 3 times with increasing specificity. Effectively used the sandbox to validate edge cases. Missed error handling on malformed input records."
Return on investment
We modeled the ROI using published bad-hire cost data, GitHub's productivity research, and McKinsey's 2024 AI impact study. The assumptions are conservative.
Model assumptions
$185k/yr
Avg engineer total comp
Levels.fyi 2025
46%
Industry bad hire rate
LinkedIn Talent Trends 2024
+35%
Maite hire quality lift
Conservative estimate
2.4×
AI-fluent productivity gain
McKinsey / GitHub 2024
Cost per mis-hire: $185k salary + $37k recruiting (20% TC) + $28k onboarding + $37k replacement recruiting = $287k per bad hire. Productivity value of additional high performers: (2.4 − 1) × $185k = $259k/engineer/year incremental value delivered.
Startup
8 hires/year
Starter pack ($99/assessment)
Annual net value
$1.1M
91,819× ROI
$99/assessment · 12 assessments/yr
Scale-up
30 hires/year
Growth pack ($79/assessment)
Annual net value
$3.0M
83,979× ROI
$79/assessment · 45 assessments/yr
Enterprise
120 hires/year
Scale pack ($59/assessment)
Annual net value
$11.9M
112,216× ROI
$59/assessment · 180 assessments/yr
Assumes 1.5 candidates assessed per hire (light pre-screening before Maite). ROI figures are illustrative projections based on published research. Actual results vary.
Features
rrweb DOM replay captures every keypress, AI prompt, and iteration — not just the final code.
Claude Opus analyzes the full session against your rubric and returns evidence-backed per-category scores.
E2B-powered sandboxes let candidates run and iterate on real code. No install required.
Pre-built scenarios modeled on real customer problems across data engineering, APIs, and systems.
RLS-enforced org isolation. Each team sees only their candidates and results.
WorkOS-powered SAML/OIDC integration with Okta, Azure AD, and Google Workspace.
Candidates receive a magic link. Results are emailed automatically when scoring completes.
Weight categories by what matters for your role — FDE, IC, staff, or specialized tracks.
Webhooks and API to push scores directly into Greenhouse, Lever, or Ashby.
Beta pricing
You only pay when you actually assess a candidate. These are the rates I'm planning to launch at — beta users lock them in.
One Maite assessment costs as little as $59. One bad engineering hire costs $287,000. That's a 4,864× downside if you skip it.
| Pack | Assessments | Per assessment | Pack total | Savings vs. on-demand | |
|---|---|---|---|---|---|
| On demand | 1+ | $149 | — | — | Request Beta Access |
| Starter | 20 | $99 | $1,980 | Save $1,000 | Request Beta Access |
| Growth Popular | 50 | $79 | $3,950 | Save $3,550 | Request Beta Access |
| Scale | 100 | $59 | $5,900 | Save $9,100 | Request Beta Access |
| Enterprise | 500+ | Custom | — | Max discount | Request Beta Access |
All packs include: session recording, Claude analysis, email delivery, scenario library, admin dashboard, and API access. Enterprise adds SSO, audit logs, SLA, and custom rubrics.
Every engineering hire you make without Maite costs you a coin flip on AI fluency. At $287k per mis-hire, that's an expensive habit.
I read every request myself — you'll hear back within a day.
$3.0M
net value for a 30-hire team
83,979×
return on investment at $79/assessment
$59
minimum cost per assessment (Scale pack)