Business analysis · AI · Financial services
Bridging business and AI by governing what it knows
I have spent more than ten years in IT for capital markets, asset management and banking, much of it as a business analyst on complex systems. This is my view of what AI needs from us: not more context, but the right context, chosen and approved by people.
My view
AI is only as good as what it knows
Models keep getting better. What they produce still depends on the context we give them: the documents, the rules and the decisions behind a project. Give them too much, and what matters gets lost. Give them something wrong or out of date, and they are steered off course.
Financial institutions learned long ago to govern their golden data: trusted sources, business rules, clear owners. I think AI needs the same discipline for its context, with one difference. A golden context has no deterministic rules, so a person has to choose the inputs, review them and approve the result.
That person is what the Context Guardians idea calls a Context Guardian: someone who curates, validates and maintains the context AI works from. AI checks, a human approves, and the human stays accountable for what is governed.
The idea is set out in the Context Guardians manifesto. What you read here is my view of it, from the projects I have worked on.
Where the view comes from
The experience behind it
IT in financial services
More than ten years in capital markets, asset management and banking: trading platforms, a large banking-system migration, model risk governance.
Business analysis on complex systems
Turning what the business needs into what systems do, on projects where a misunderstanding is expensive.
CFA charterholder
The finance behind the systems: the instruments, the risks and the reporting they support.
Before any solution
The questions that decide a project
No methodology fits every project. The ones I have seen succeed had a clear goal, a high-level plan, a working structure, and someone asking the right questions before diving into a new solution:
- What specific business outcome are we targeting, and how will we measure success?
- What impacts will these changes have on existing business processes and on the ecosystem as a whole?
- Does this initiative clearly align with our strategic priorities, and why is it prioritized above other options?
- Could an existing process or tool solve 80% of the problem and lead to a less complex, more manageable application portfolio?
- Which regulatory, operational or reputational risks does this introduce, and how will we mitigate them?
With AI in the project, the same questions apply, and one more: what does the AI know, and who approved it?
Follow the conversation
I write about business analysis, AI and the context it works from. New posts appear on LinkedIn first.