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 essayFewer steps between your team and a resolved request.
Your service tools were meant to help. Now they're the work. Kanso removes the steps that get in the way, so the metrics you're judged on, from handle time to customer effort and CSAT, move in your favour.
For the people who own customer service, employee service, ITSM and the contact centre, including teams already running Zendesk, or similar, who want more value from what they own.
簡素 (kanso): the elimination of the unnecessary.
We work on one or more metrics you own in two ways: the engagements we run with you, and the software we make. We sell our own software and refer the rest, so the recommendation isn't bent by what we happen to sell.
Services
Where your AI and service spend actually returns value.
Six service lines, from AI advisory and readiness through support-platform delivery, contact centre and voice, automation, reporting, and multilingual support. Practice-led, scoped engagements. We've run the service teams we advise. Each one is tied to one or more metrics you own (handle time, customer effort, CSAT, or another business measure you and your sponsor agree up front).
See the engagement catalogueSoftware
Our own software, built where a product earns its place.
When the right fix is software we can stand behind, we build and sell it ourselves; when an existing platform does the job better, we point you to it. Each app is focused on one job: take real friction out of your team's day and stand against one or more metrics you own.
We build software for the platforms our clients already run, Zendesk among them.
The Kanso Filter
One filter governs everything we build and everything we recommend.
The Kanso Filter: six questions before a change ships. Asked in order, before we ship or recommend it.
Does it remove a step?
Does it move a metric you own?
Does it improve accuracy or quality?
Does it minimise maintenance burden?
Does native beat custom?
Can you name the metric it should move, and how you'll measure it?
Four or more yes: ship, with notes on the gaps. Fewer than four: don't ship. The scope isn't right, or the change isn't worth it.
Use the Kanso FilterWhat we do
Six service lines. One that starts with the numbers.
We work across six service lines, scoped to your situation. AI Advisory & Readiness is the line that sets the direction: Maturity Mapping across the Seven Pillars is its cornerstone, showing where AI spend actually returns value before any budget is committed.
- AI Advisory & Readiness
- Support Platform Delivery
- Contact Centre & Voice
- AI Automation & Assistants
- Reporting & Analytics
- Multilingual & Real-Time Translation
AI Advisory & Readiness
The other five lines are scoped to your situation. Book a call to talk them through.
Diagnose.
From £3,000
Short, focused engagements that examine the operational pain: queues, handle time, customer effort, admin load. 100% credited toward Strategise within 60 days.
Recommended starting point
Strategise.
From £8,000
Fixed scope, board-ready artefacts, a business case tied to the metrics you own. 50% credited toward Transform within 90 days.
Transform.
From £25,000
For teams ready to act on the strategy. Milestone-based payments, quality guaranteed to an agreed standard.
Prices shown for your region, exclusive of tax. VAT/GST may apply. See terms.
We fund two Maturity Mapping assessments per month. A senior practitioner spends real time on your problem, so we cap it at what we can do properly. Tell us where you're stuck in a few lines. We'll take the two we can help most this month, and point the rest to a better fit, or a faster route that isn't us.
Partners & platforms
- Foundational AI modelsAnthropic, OpenAI, Mistral, Z.ai, Microsoft
- Cloud & infrastructureAWS, Microsoft, Google
- CX & support-platform ecosystemZendesk, Stylo, agnoStack, CloudSet
- Voice & contact centreTwilio, Amazon Connect, ElevenLabs
- Data & analyticsMicrosoft Fabric, Power BI, Amazon Redshift, Amazon QuickSight, Amazon SageMaker AI, Amazon Textract, Geckoboard
- Language & translationDeepL
See all partners & platforms. Where we have a commercial interest, we say so. Read the Principles.
Labs
Practice-led teardowns of real problems.
Where we document operational problems, test the patterns we're considering for products or services, and publish what works.
AI where it doesn't belong
The AI Triage: when NOT to use an LLM
Our named method for sorting a job into declarative, classical ML, generative, or hybrid, and the cases where an LLM makes it worse.
Coming soon
Hybrid AI
The declarative-outer, generative-inner pattern
Combining declarative safeguards with generative models inside service workflows.
Coming soon
Principles
Five principles about what we leave out.
We don't resell technology.
We sell our own. We refer the rest.
We don't recommend AI where it doesn't add value.
Including when a declarative process beats an LLM.
We don't build custom where native solves it.
The core vendor's roadmap first.
We don't take on work without a measurable outcome attached.
A number you choose, measured in your systems.
When we have a stake, we say so.
We partner with other technology companies, and we build our own.
Decades of running service teams behind every decision.
Kanso is new; the practice behind it isn't. We've run the teams we advise: a global QSR brand's AI stack (CSAT 35 to 75 percent), a national property company's real-time AI (£250K of capacity recovered), and Zendesk rollouts measured on customer effort, resolution times, and CSAT.
Read more about KansoStart where it's easiest.
We fund two Maturity Mapping assessments per month. A senior practitioner spends real time on your problem, so we cap it at what we can do properly. Tell us where you're stuck in a few lines. We'll take the two we can help most this month, and point the rest to a better fit, or a faster route that isn't us.