Shadow AI Diagnostic

What it does

A founder invites their team. Everyone answers a short, anonymous survey. The founder gets one report showing:

• How ready the business is to use AI well
• How much "shadow AI" is happening (tools used without approval)
• Which teams are ahead and which are exposed
• Clear next steps for each gap

Problem

Most founders think they know how their team uses AI. Usually they don't.

People paste customer data into free tools. Nobody owns the policy. Then a board member asks about AI risk and the CEO has to guess.

A survey of their own team gives them real answers, fast.

My role

I owned it from idea to launch:

• Designed the scoring model and the anonymity rules
• Built the prototype and the live app
• Wrote the insight library, with every stat checked against its original source
• Planned and ran the go-to-market: landing page, CRM setup, tracking, email and paid social

Workflow

1. 01
   Set teams

   Founder signs up and sets up their teams.

2. 02
   Invite

   Each staff member receives a private, single-use link.

3. 03
   Protect privacy

   Only show team results once enough people respond.

4. 04
   Update live

   Refresh the report as responses come in.

5. 05
   Recommend

   Connect each insight to a practical next step.

Tools and integrations

• App: Next.js, Vercel, Supabase
• Build: Claude and Claude Code
• Go-to-market: HubSpot, LinkedIn Ads, Google Analytics, Webflow

Example output

The founder report includes:

• An AI readiness score
• A shadow AI risk rating (low, medium or high)
• A heatmap by team
• Top insights, each backed by research and linked to a next step

Results & status

• Live and taking sign-ups
• Improved once already, based on where people dropped off
• Early stage: not enough traffic yet to report conversion or pipeline numbers

Limitations

• Small teams get less detail, because anonymity comes first
• Long B2B sales cycles mean impact takes months to show
• Results depend on honest answers from staff

Related projects