ShelVey AI, our first cohort: 17 people, one shared way of working with Claude Code.
ShelVey AI's developers were already experimenting with Claude Code, but each in their own way. Its 17-person technical team became Sthuthi's first organisational Claude Code cohort, at the Practitioner level, with the aim of a common, disciplined working method.
17
employees certified
1st
organisational Claude Code cohort we ran
2 weeks
5 – 16 January 2026
ShelVey AI is part of Sthuthi's own network. We include this first organisational cohort as an affiliated engagement, not as independent proof.
"Before the certification, our developers were already experimenting with Claude Code, but everyone was using it differently. What Sthuthi Tech gave us was a much more disciplined way of working with it. The team became better at giving Claude the right repository context, breaking larger changes into sensible steps, reviewing generated code properly and using it for debugging without handing over engineering judgement. For a small technical team, that consistency made a noticeable difference very quickly."
ShelVey Leadership · ShelVey, LLC
Who took part
- Software engineers
- Full-stack developers
- Automation engineers
- Technical product staff
The whole technical team selected for certification, at the Claude Code Certified Practitioner — Sthuthi Technologies' independent credential.
Focus
- Repository navigation and contextual prompting
- Implementation planning
- Code generation, debugging and refactoring
- Test creation
- Human review of AI-generated changes
When
- Cohort 1 · 17 people
- 5 – 16 January 2026
Each cohort followed the track's standard format: instructor-led sessions on real code, then a proctored, task-based assessment reviewed by a human assessor before a credential was issued.
What changed
For a compact team, the main improvement was standardisation. Instead of each developer treating Claude Code as a personal productivity tool, the cohort agreed a common way to provide context, request changes, check generated implementations and decide when an engineer should step in directly.
The team could then use Claude Code more consistently across feature work, troubleshooting and refactoring, without lowering its normal code-review standards.
Why one cohort
The 17 people covered all the technical employees selected for certification at the time, so the engagement ran as a single cohort.
Commercial terms, contract value and pricing are confidential and not shown.
Planning a cohort of your own?
Tell us how many engineers and which teams. We'll scope a pilot and quote it on one call.