Micro-friction teardown
Why most auto-tagging projects fail in Zendesk
Seven ways auto-tagging programmes actually fail in Zendesk, in the order they usually bite.
Visit the essay →
22 April 2026
Labs.
Practice-led teardowns and the patterns we're testing.
Where we document operational problems, test the patterns we're considering for products or services, and publish what works and what doesn't. Biweekly essays. Occasional tools and reports. Collaborations with partners.
Latest
Recent essays.
Coming biweekly
- Coming soonThe AI Triage: when NOT to use an LLMAI where it doesn't belong
- Coming soonThe declarative-outer, generative-inner patternHybrid AI architectures
- Coming soonYour QA sample is lying to youOperational teardown
Tools & templates
Free artefacts, no gate.
The Kanso Filter
Six questions before a change ships. A decision process for service ops change.
Maturity Mapping assessment
Seven-dimension AI-readiness review. We fund a limited number. Apply with the metric you want to move.
AI readiness self-score (short form)
Coming soon
A faster version of the full assessment, for teams early in their thinking.
Coming soonQA sampling template
Coming soon
A smarter way to pick calls for review, from a forthcoming Labs essay.
Coming soon
Incubation
In the workshop.
Some of what we test here gets built into software. We prove a tool against real service team problems first, then promote the ones that hold up into full products. Right now we're exploring hybrid AI patterns, lighter QA sampling, and a few small tools that aren't ready to show.
Ask for early access →Flagship reports
Longer-form reports and references. Published quarterly-ish.
Coming soon
Why 74% of AI CX programmes fail
Research-backed, practitioner-interpreted. Published failure modes and what to do about them.
Coming soon
The EU AI Act for service teams
Canonical reference for what the Act means, where the risk tiers bite, what compliance looks like in a Zendesk-shaped world.
Coming soon
The AI Readiness Gap Benchmark
Annual benchmark drawn from anonymised Maturity Mapping self-assessment data.
Set the agenda
Want us to look at something?
If there's a service team problem you'd like the Labs team to take apart, tell us. Give us the details: what's happening, where it hurts, and what you've tried. We read every one and pick the ones we can learn the most from.
Submit a problem to Labs →Monthly digest.
One email a month covering what we published and what we're testing next. No sponsorship, and no thin content calendars.