---
title: "How Backstage IT turned invisible AI usage into a provable readiness score across 130+ developers"
description: "How What's Next built an AI Employee Scan that gives Backstage IT provable, per-developer AI-readiness scores across 130+ engineers."
canonical: "https://www.whatsnext-ai.com/cases/backstage-it-ai-readiness-scan"
published: "2026-08-14T00:00:00.000Z"
updated: "2026-08-17T23:55:45.112Z"
client: "Backstage IT"
---

# How Backstage IT turned invisible AI usage into a provable readiness score across 130+ developers

How What's Next built an AI Employee Scan that gives Backstage IT provable, per-developer AI-readiness scores across 130+ engineers.

![Backstage IT developers at work in the Chișinău delivery office](https://fgzcpjbyiakhjifaciaj.supabase.co/storage/v1/object/public/media/backstage-it-ai-readiness-scan-hero.jpg)

## Overview

Backstage IT is a Dutch nearshoring company, headquartered in Zutphen with its delivery operation in Chișinău, one of the largest IT employers in Moldova, serving clients across Europe. With 130+ developers spread across multiple clients, they had no scalable way to see who was actually using AI, and how well. What's Next built them an AI-readiness assessment that answers it with data.

## Details

- Industry: IT nearshoring & software development
- Region: Netherlands & Moldova
- Company size: 130+ developers
- Timeline: 6 weeks

## Challenge

Backstage IT sells development teams on the quality of their people. But with 130+ developers spread across multiple clients, there was no scalable way to see who was actually using AI, and how well.

The only way to find out by hand was to interview every developer, one at a time. More than a hundred conversations. And by the time the last one was written up, the first ones were already out of date.

That left leadership unable to answer a few basic questions. Which teams actually use AI in their work, and which are only experimenting? Where are the champions who can pull everyone else along, and where are the real gaps? Can they show a client proof that their developers are AI-ready, or is it just a claim?

Without those answers, the AI rollout stayed a guess instead of a plan. And the line 'our developers are AI-ready' stayed exactly that, a sentence in a sales conversation with nothing underneath it.

## Approach

We did not start with the question 'can we measure AI usage.' We started with the harder one: how do you measure it so it holds up when a client pushes back.

Step 1, rule out the wrong measurement. A self-report survey is fast and close to worthless. People overrate their own AI level, and a score someone gives themselves is not evidence to a client. The measurement had to be about what a person actually does, not what they say they can do.

Step 2, anchor the scoring in existing frameworks. Instead of an in-house opinion, we built a defensible 0 to 5 maturity ladder (Unaware, Aware, Pockets, Structured, AI-Native), with the production threshold set deliberately at Pockets. That ladder lines up with five independent models (CMMI, MIT CISR, Gartner, Microsoft, Google Cloud) that all land on the same shape. The rule we held ourselves to: no vibes, every score traceable to a source.

Step 3, score by scenario, not self-report. Each employee is scored on five dimensions: AI competency, sentiment and readiness (on the Prosci ADKAR backbone), current usage, automation potential, and data readiness. Competency is tested with a concrete scenario, not a self-rating. The gap between competency and potential is exactly the ROI signal.

Step 4, design the roll-up. One scorecard per person, backed by direct quotes from the interview, rolls up into a department view (a composite plus the weakest ADKAR letter as the cheapest lever) and into a company report with a cross-department heat map and a prioritised programme of work.

The methodology was not optional. It is the reason every score in the final report traces back to a source instead of an opinion, and that is the difference between a number a client believes and a number they wave away.

## Solution

We built the AI Employee Scan: a platform that assesses an entire organisation without anyone having to run a hundred interviews.

A personal agent for every employee: each developer gets their own personalised agent that walks them through a set of AI questions live and scores the answers on the spot. It is not a static form and not a survey. The agent adapts the questions and scores by scenario, which is what makes scenario-based scoring possible across 130+ people. It runs in any language, so each developer is interviewed in their own native tongue, where their answers are sharpest and the scoring reads what they actually mean instead of their second-language phrasing.

Five dimensions, anchored in evidence: every interview produces scores on five axes. Data readiness in particular is not judged against a soft definition but a concrete checklist. Every flag is classified against the EU AI Act's four risk tiers and tagged with a GDPR concern, so a data flag cites the regulation and the remediation, not a hand-wave. The scan holds itself to that same standard: a built-in PII check runs over the interview data, so personal information is spotted and handled deliberately, never passed through unchecked.

From person to department to company: the same five dimensions roll up across three tiers. Each tier writes once and feeds the next, so the company report is grounded all the way down to individual quote-level evidence.

Delivering that coverage by hand takes weeks and is stale by the time it is finished. The scan does the same coverage differently:

AspectManual, one at a timeAI Employee ScanHow each developer is assessedA separate half-hour interview plus write-up, one person at a timeTheir own agent-led session, scored live, all 130+ in parallelTime to cover 130+ developersWeeks, and stale by the time the last one is written upDays, and current the moment it finishesCost per developer assessedAbout an hour of senior time eachA fraction of that, roughly 80x lower on a conservative estimateDoing it again next quarterThe whole exercise, from scratchRe-run at near-zero marginal cost, and watch the curve move

The deeper point is not the cost, it is ownership. The scan is not a subscription that scales against Backstage, it is a re-runnable data asset they own.

The human stays deliberately in the loop. The agent-led interviews are not a simplification but a choice: they make scenario-based scoring reliable at scale, which a mass survey never could. And because the scan is repeatable, every new run becomes a data point, not a snapshot but a curve.

## Outcome

For the first time, Backstage IT has a way to see, rather than assume, where the organisation stands. Every one of its 130+ developers can be scored individually and rolled up by person, by department, and company-wide, putting the champions and the gaps in view instead of leaving them to guesswork.

Operationally, that changes how Backstage can run its AI rollout. Instead of training everyone at once on instinct, it can direct training precisely to the teams and the weakest ADKAR letter where the most ground is there to gain. The rollout becomes a plan instead of a guess.

Strategically, the shift is bigger. AI-readiness was a claim in a sales conversation and is now a repeatable AI-maturity measurement Backstage can put on the table, use to position its developers as demonstrably AI-strong engineers, and re-run to show the curve rising. It went from a talking point to a differentiator.

Discovery and build mattered equally. Without the anchored scoring model the scan would have been a fast survey, and without the build the model would have stayed a PDF. Together they leave Backstage with a measurement instrument that keeps working long after the first run is done.

## Results

- **130+** developers, each assessed by their own agent in days, not the weeks of one-by-one interviews
- **3 tiers** person, department, and company scores, every one traceable to a source
- **80x lower** cost per developer assessed vs. one-by-one interviews, on a conservative estimate

## Quote

> AI-readiness used to be a story we told. Now it's a number we can show, per developer and per team, and re-measure every quarter. That completely changes the conversation with a client.
> — Sander Geels, Co-founder & CEO, Backstage IT