Diagnosing Low Adoption of the AI Remediation Agent

The problem I was hired to solve

SonarQube's Remediation Agent (SQRA) launched into open beta in early 2026 — and adoption was well below expectations. The question on the table wasn't just "why aren't people using it?" It was: do we have a product problem, a positioning problem, or a go-to-market problem? I was brought in to find out.

What I did

I ran a structured investigation across four fronts:

  • Funnel diagnosis: Sampled the target account list, worked with account executives to trace where outreach was failing, and analysed campaign data to find where interest dropped off.
  • Direct customer outreach: Spoke with customers who had access but hadn't activated to understand the real reasons — not the assumed ones.
  • Competitive landscape: Mapped the remediation tool landscape across Wiz, Dependabot, CodeRabbit, Copilot, Sentry/Seer, Cursor/BugBot, and Devin to understand what we were being compared against and on what dimensions customers were making decisions.
  • Sales enablement: Identified that sales teams lacked the language and materials to answer customer objections confidently. Built a live GitHub-hosted resource using Claude that gave sales a reliable, always-current answer to the most common questions raised during demos.
  • IMDA partnership study: Led user discovery with IMDA Singapore, a government collaboration partner. Designed a testing plan combining baseline metrics, surveys, and interviews to measure time to remediate, fix volume, trust, and cognitive load with and without the agent.

What I found

Adoption wasn't low because of a single missing feature. The real problem was a stack of compounding friction:

  • Targeting was too broad. The accounts being approached weren't necessarily the ones most likely to convert — the funnel was leaking at the top.
  • Awareness was limited. Many eligible customers hadn't heard of SQRA or didn't understand it was available to them.
  • Trust was the core barrier. AI review boards inside enterprise companies were flagging the agent before it reached developers. Customers needed proof, not just promises.
  • Positioning was unclear. Neither customers nor sales teams could clearly articulate why Sonar's agent was the right choice over Claude, Copilot, or an internally-built agent.
  • Packaging and pricing created hesitation. The commercial model wasn't matching how customers evaluated value.

The market context added urgency: the space was moving fast toward shift-left environments, iterative remediation, transparent reasoning, and tighter PR-workflow integration. Standing still wasn't an option.

The decisions this enabled

  • Shifted the internal conversation from "why is adoption low?" to a more specific and actionable question: who are the right customers, and what conditions need to be in place before they'll convert?
  • Gave the product and GTM teams the specific metrics and benchmarks they'd need to prove value to enterprise buyers.
  • Clarified which UX and product directions were worth prioritising next to close the trust gap.
  • Gave sales a concrete enablement resource they could use immediately — reducing reliance on ad-hoc answers in customer calls.
  • Established an evidence base for sharpening the value proposition and targeting criteria going into the next growth cycle.