The Institutional AI Framework
Eight operating decisions for moving from AI experimentation to a capability an institution can own, govern and fund.
Read the framework
AI is becomingInstitutionalInfrastructure.
How institutions adopt AI, govern its authority and build the capabilities to operate it.
Established in Singapore · 2024Illustrative image
The institutionis the unitof change.
WCAI helps institutions navigate the transition from experimenting with AI to operating with AI.
A model can make a task possible. An institution must decide who owns the outcome, which actions it will authorise, and whether the improvement survives the complete workload and cost.
WCAI connects those decisions through research, practical resources and implementation.
Four connected fields
Explore our focus areasInstitutional AI
From promising experiments to operating capabilities an institution can own and fund.
ExploreAI Governance
Accountability, meaningful oversight and evidence for the authority an AI system receives.
ExploreSovereign AI & Strategic Infrastructure
Control of data, models, compute and the dependencies that shape institutional choices.
ExploreAI & Financial Systems
Payments, digital assets and the operating responsibilities of delegated financial action.
Explore
The Institutional AI Framework
Eight operating decisions for a capability the institution can own, govern and fund.
A proposed decision method, not a validated maturity score.
Read the frameworkStrategy & leadership
Which outcome warrants investment, and who can change or stop it?
Evidence to inspectBaseline, sponsor, alternatives and staged funding decision
Operating model
Who performs, accepts and supports the work?
Evidence to inspectProcess map, responsibilities, exception queue and operating agreement
Governance & assurance
What use and residual risk can be authorised?
Evidence to inspectTested scope, failure evidence, approval rules and incident route
Data & knowledge
Which records may the system use, and which are authoritative?
Evidence to inspectData flows, access tests, provenance and disputed-record process
Infrastructure & architecture
Can the service meet its obligations and withstand dependency changes?
Evidence to inspectDependency map, service limits, recovery test and exit evidence
Workforce & change
Can the funded team operate the changed process?
Evidence to inspectRetained tasks, training, reviewer capacity and fallback rehearsal
Agent readiness
What actions may the system initiate or complete?
Evidence to inspectEnforced authority, action-bound approvals and duplicate/retry tests
Value realisation
What benefit survives the complete cost and workload?
Evidence to inspectCapacity, cash, quality, backlog and downside comparisons
- Strategy & leadership
- Operating model
- Governance · Data · Architecture · Workforce · Agent readiness (examined together)
- Authorise a bounded operation, restrict it, or hold
- Value realisation
- Revisit scope, funding and controls when evidence changes
A proposed decision sequence, not a validated causal model.
Selected from WCAI
When AI agents control money
A practical approach to delegated spending: identify the principal, bound the mandate, enforce limits and preserve evidence when an agent initiates a payment.
Read the insightSovereign AI does not necessarily mean sovereign models
A practical way to decide which AI capabilities a country or institution must control, and where partnerships can meet the requirement.
Read the insightWorking resources
Content and commerce experiment brief
Design a comparison of AI-assisted content production that accounts for audience assignment, contribution, costs and uncertainty.
OpenSustainability evidence register
Trace a sustainability reporting line from source to review using a CSV register and fictional examples of corrections and missing evidence.
OpenReconciliation evaluation pack
Plan a fair reconciliation evaluation with a study record, fictional CSV fixtures and logs for proposals, allocations and final decisions.
Open
Advisory for consequential AI decisions.
Senior decision support, combined with hands-on implementation alongside the teams responsible for the outcome.
Built by WCAI
Connect an AI decision to a bounded workflow, tested permissions and a service people can support.
Explore how we approach implementation and the evidence an operating team needs.
Founded in Singapore.
Informed by worldwide experience.
Founded by Randeep Melhi, WCAI brings an institutional finance, consulting and operating perspective to the decisions surrounding emerging technology.
About WCAIThe WCAI Briefing
A monthly perspective on the AI developments that change institutional decisions.