Loading…
Free
A free, ten-minute questionnaire scoring five dimensions of AI readiness. A PDF report with dimension scores, three prioritised next steps and a comparison to the median UK SME — in your inbox within one working day.
What you get
No account, no credit card, no follow-up sequence. The report is the output — a standalone artefact your team can act on.
A score per dimension, one finding each, an overall readiness rating. Built for an exec audience — no eval jargon.
Concrete, sized and sequenced — not “invest in data”, but the specific first move drawn from your lowest-scoring dimension.
Where you sit against the median across Canarlo's assessment dataset — useful for prioritising spend and calibrating expectations.
Row-level security, never used to train anything, UK GDPR throughout. Anonymised into the benchmark after 90 days.
How it runs
The same four stages run every engagement. The service decides the scale, not the method.
01
Planner
Twenty-odd questions grouped by dimension. Multiple-choice and short-scale, not essay. No account required.
02
Hub
Five dimensions scored by a person, not an algorithm alone. A disagreement across functions becomes a finding.
Pricing
Ten-minute questionnaire, PDF report scoring five dimensions, three prioritised next steps, comparison to the median UK SME.
FreeThirty minutes with the engineer who wrote your report. No obligation, no sales sequence.
FreeAlready past readiness and running AI in production? The AI System Audit is a fixed-fee diagnostic.
From £8,000Proof in production
Categorisation, summarisation, and triage automated across an operations team's daily workload. Graceful degradation when the…
View projectA documentation platform that reads code and emits prose a non-engineer can act on. Used internally on every Canarlo build.
An honest AI readiness checklist for UK businesses: 10 yes/no signals, a simple score, and what to…
Read guideFive practical AI quick wins you can set up this week — the tools, setup time, monthly cost, hours…
Read guideQuestions
About ten minutes. Twenty-odd questions grouped by dimension — multiple-choice and short-scale, not essay. No account, no credit card. The PDF report lands within one working day.
Responses land in our database. An engineer — not an account manager — scores the five dimensions and writes the PDF. A booking link inside the email opens a thirty-minute review call if you want one.
Yes. Responses sit behind row-level security — only Canarlo engineers can read them, only to write your report. Not shared with third parties, not used to train anything. UK GDPR throughout.
What we score
01
Data Readiness
Whether the data an AI would touch is available, labelled, and reachable. The typical failure: ten years of knowledge locked in PDFs and a shared drive nobody has indexed.
02
Use-Case Clarity
Whether you have a named task, a user, and a success metric — or a board slide that says “we want to use AI”. One is buildable. The other stalls at week three.
03
Technical Capability
Engineering bandwidth, a person who has shipped an AI feature before, and infrastructure that won't fight you. The common gap: nobody has ever evaluated an LLM in production.
04
Risk Tolerance
Your appetite for hallucination, latency drift, and UK GDPR exposure. Low tolerance is not disqualifying — it changes the architecture, not the ambition.
05
Team Buy-in
Leadership alignment plus end-user willingness to trust the output. The classic gap: a leadership team ready to go and a support team waiting to be replaced.
Talk to us
Scoped and priced before anything is built.
03
Forge
Findings ranked, three next steps sized and sequenced. Delivered as a PDF within one working day.
04
CanOS
Thirty minutes with the engineer who wrote your report, if you want it. No obligation either way.
How to automate your business without overbuilding: a build-vs-buy framework for your first AI…
No. The PDF is the output — a standalone artefact your team can act on. No upsell, no follow-up sequence.
Yes — recommended. One submission per team, filled in together. If functions disagree on a dimension, that's a finding worth surfacing — submit twice, one per function.