Advisory Finance
Sanlam
Research into how customers contact Sanlam came first, then a knowledge base that shows what they search for.
Find out where AI is worth using in your business.
We look at your processes, data and goals to find the uses of AI worth pursuing, test the strongest, and tell you plainly where AI isn't the answer.
Start with the decision, not the model.
Most AI plans begin with a tool and look for somewhere to use it. The useful ones begin with a process: a task people repeat, a decision that depends on information spread across systems, or a queue of questions that all need the same kind of answer.
From there, three questions decide whether AI is worth it. Is the task frequent enough to matter? Is the information it needs available and reliable? And what happens when the answer is wrong? A wrong product suggestion costs little. A wrong answer about a policy or a payment costs a lot more, and needs a person in the loop.
Use, configure or build
Most businesses need less custom AI than they are offered.
Use what you have AI features already in tools you pay for, switched on with sensible rules. Limited to what the vendor offers, and the data stays in their product.
Configure and connect An existing model connected to your own documents and systems. Needs clean source data and clear rules for who can see what.
Build A use that is central to how you compete and has no product equivalent. Takes longer, costs more to run and needs ongoing testing.
What we check before recommending anything
- How often the task happens, and what it costs today.
- Where the information comes from, and whether it can be trusted.
- What a wrong answer would cost, and who would catch it.
- Who owns the process once AI is part of it.
- How the result will be measured against the current way of working.
- The privacy and security rules the data falls under.
Our advisory work
Advisory Health and Wellness
Healthcare administration group
An independent digital assessment: 7 stakeholder interviews, 400+ pages reviewed and a prioritised transformation roadmap.
Advisory Other
Industrial solutions provider
An independent strategy engagement that sharpened direction and leadership alignment for a confidential industrial provider.
Advisory Other
Cape Town Child Welfare Society
Mapped 8 workflows, 5 user roles and 7 report types into a blueprint for a case management system.
AI strategy FAQs
Do we need a lot of data to use AI?
Not always. Many uses work with documents, emails and records you already have. We assess what's needed for each use.
Should we build our own AI?
Rarely. Most uses are better served by existing models connected to your systems. Building makes sense when the use is central to how you compete.
What does a proof of concept involve?
A small, time-boxed test of one use, with real data where possible, so you can see the result before committing to more.
Who builds it once we decide?
Realm's software team can build and integrate what the strategy recommends, or your own team can take it forward.
Latest insights
Tell us where you're considering AI.
Give us a sense of the processes involved and what's prompting the question.