Why your AI pilot dies at the handover
Why AI pilots stall before production and use stays shallow: depth of use, stopping criteria, workflow redesign and baseline.
November 4, 2026 · 3 minRead article →
AI readiness · Meridian
AI readiness isn’t having AI.
It’s knowing whether your data, processes, people, governance and infrastructure can carry what comes next. We measure that in two weeks — and the deliverables are yours.
2 weeksto the score
5 axesmeasured
12 monthsof plan
0–100
Readiness score
01 Data02 Process03 People04 Governance05 Infrastructure
What people think
What people think AI readiness is:
- Buying an AI tool
- Building a chatbot
- Training the team on prompts
- Hiring AI talent
- Running pilots
- Automating processes
- Deploying agents
- Choosing a vendor
- Investing in infrastructure
- Following trends
AI readiness
What it actually is · five axes
What AI readiness really is:
- Data quality and governanceData
- Data the process can reachData
- Value stream mappingProcess
- Documented processesProcess
- Use case prioritisationProcess
- Leadership sponsorshipPeople
- People engagementPeople
- Change managementPeople
- AI usage policyGovernance
- Risk managementGovernance
- Architecture ready to scaleInfrastructure
- Security and accessInfrastructure
Research · five axes
Feeling ready isn’t being ready.
On each axis, what leaders say and what the same survey finds right after.
What they sayWhat the same survey shows
Data
88%say their data is ready for AI
43%name data as their biggest obstacle
Process
52%say they’re prepared on strategy
21%on business processes, the last of seven areas
People
86%of CEOs say their people have the skills
25%of the workforce uses AI regularly, by their own count
Governance
71%expect to use AI agents within 12 months
21%can govern them today
Infrastructure
87%say they have the infrastructure they need
42%name it as their biggest obstacle
All self-reported, and each pair is two different questions: what matters is the contradiction, not the subtraction. Readiness goes along with results, but we found no study proving it causes them.
In Brazil
The most prepared group fell from 29% to 18% across three editions of Cisco’s index. Worldwide, from 14% to 13%.
Cisco AI Readiness Index, 2023 to 2025. Companies with 500 people or more, n per country not published. Cisco sells infrastructure.In law
Brazil has no general AI law. Bill PL 2338/2023 awaits a report in the Chamber, and the LGPD already applies. As of 2026-11-04.
Chamber of Deputies, bill record.Five axes · 0–100
Five axes. One score.
Each axis gets a score from 0 to 100. The lowest decides where the plan starts.
01 · Data
Data
Does the right data reach those who decide?
Quality and governanceAccess and integrationData classes
Score 0–100
02 · Process
Process
Where does AI change the work — and in what order?
Value stream mappedDocumented processesPrioritised use cases
Score 0–100
03 · People
People
Who sponsors, who uses, who resists?
Leadership sponsorshipTeam capacityChange management
Score 0–100
04 · Governance
Governance
Does legal know how to say yes?
Usage policyRisk matrixVendor clauses
Score 0–100
05 · Infrastructure
Infrastructure
Can the architecture scale?
Architecture and integrationSecurity and accessOperating cost
Score 0–100
Meridian · deliverables
Two weeks. Three deliverables.
Yours, whether or not we continue together.
Illustrative examples · not client data
01
Score per axis
A score from 0 to 100 for each axis, with the evidence behind it.
02
Gap register
Gaps ranked by what it costs to wait — not by what’s easiest.
03
12-month plan
The sequence for the next 12 months, quarter by quarter.