Empowering experimenters with AI ‘vibe coding’

The challenge

Development resource is rightly focussed on big challenges making ‘hygiene’ content features difficult to prioritise. Getting enough resource to even test them to find out if they’re worth building can be challenge enough to not bother.

Long term damage to reputation as you fall behind on ‘the basics’ is difficult to quantify or provide evidence of. Users don’t tend to ask for “more video”, “audio summaries” or “personalisation”…they just go elsewhere.

Internally, it’s not good either. A negative cycle between stakeholders and the engineering team emerges where each loses confidence in the other to bring about necessary change. Stakeholders begin to believe the engineering team can’t deliver; the engineering team begins to believe stakeholders are focussing on the wrong things.

The goal

Rebuild stakeholder confidence that we’re able to move at pace and respond to change while keeping the engineering team focussed on the biggest problems.

Create an environment where there is a ‘no engineers necessary’ route to trying out an idea (and an ability for non-technical team members to maintain those tests as content changes with daily and weekly news cycles).

Overall, rebuild a positive cycle of testing, learning and collaborative prioritisation.

The strategy

• Use generative AI to remove the need for upfront development resource to test out ideas
• Use generative AI tooling to provide authoring capabilities for non-technical team members to be able to maintain experiences through testing periods as content changes
• Empower team members (that have ideas and interest) with generative AI and AB testing skills to scale discovery from the ‘grass roots’
• Share successes far and wide to anyone that will listen (Yammering, internal newsletters, email, All Hands meetings, Forums, etc)