Skip to content
ADVERTISEMENT
Technology

AI Matters: Two ways to buy AI, and how to tell which one you need

Matthew Elliott of Nivo discusses how brokers can use AI to reduce administration without replacing human judgement.

Opinion
5 min read
Share
AI Matters: Two ways to buy AI, and how to tell which one you need
Matthew Elliot
ADVERTISEMENT

How many documents and how much administration does it take to secure a loan? Well, take one commercial case as an example. The business banks with three banks, so that’s six months of statements from each. There are two directors, so two identity checks, two proofs of address, two personal forms. The accounts are a year old, so you need management figures. There is existing borrowing, so you need settlement figures.

Now count the emails. You ask. Nothing comes. You ask again on Friday. Half of it arrives. One statement is missing a page and you cannot tell which month until you check. You ask a third time.

15 rounds of email on a case that was never complicated. That is the job most brokerages are actually doing, and it’s the job nobody joined the industry to do.

I put it that way because the artificial intelligence (AI) conversation usually starts at the wrong end. Firms ask where they could use AI. The better question is where their people spend time they shouldn’t. Skip that question and you buy technology that makes a poor process run faster, which leaves you where you started, only now with a subscription.

ADVERTISEMENT

The prize is not speed. It is removing the work. Most brokerage admin exists because documents arrive incomplete, information gets checked twice, and someone spends three weeks chasing a single page. Cut that and you have created capacity, not efficiency. Capacity is what lets a firm grow without hiring.

ADVERTISEMENT

So, say you accept the problem. How do you buy the fix?

The fast start

You buy something already built. The gain is that you learn from real cases early. Within a fortnight you know whether the thing works on your business. That is a much cheaper way to be wrong than a nine-month build. The trade-off is fit. Off-the-shelf means someone else’s assumptions, and there will be cases where the tool does most of the job and hands you the rest.

ADVERTISEMENT

Here’s the part firms tend to get backwards. Eighty per cent of the task done right is not a shortfall. It is the difference between an adviser handling all of a case and an adviser handling the 20% that needs a human. Nobody looks at a new hire in week three and calls them a failure because they escalate the odd file. The tool that clears the routine work and flags the awkward case is doing exactly what a good junior does.

The firms that get least from this are the ones who hold out for a system that handles everything before they will start. They spend a year specifying the edge cases and gain nothing in the meantime, while the firm next door has been saving hours a case since March. Perfect is not the standard. Better than Tuesday is the standard.

ADVERTISEMENT

The bespoke build

ADVERTISEMENT

The other route is to design the process around your own business. Every rule, every edge case, every integration with the systems you already run.

Done well, this fits precisely, and it fits nobody else. That is worth real money if your process is genuinely to your advantage; if the way you package a case, or score a lead, or route work to the right adviser is something competitors cannot copy.

The cost is time, money and attention. Long projects need people inside your firm who can specify what they want, and most brokerages do not have those people spare. Scope creeps, because it always does. The requirements you wrote in January describe a business that has moved on by September. And in a field moving this quickly, a twelve-month build can go live already behind the market.

The failure I have seen most often is not a bad build. It is a good build of the wrong thing. A firm spends a year automating a process it should have deleted, because nobody stopped to ask whether the work needed doing at all.

ADVERTISEMENT

What we have learned running both

We work with clients on both routes. The firms that started narrow got to value faster and they ended up with better long-term systems. Not because narrow is better, but because running something real teaches you what you actually need. Requirements written in a workshop are guesses. Requirements written after three months of live cases are facts. Almost every client who began with a focused deployment came back with a sharper, shorter, more useful list of what to build next than the one they would have written at the start.

The bespoke projects that worked shared two things. Someone senior owned the outcome and stayed with it. And they went live on a narrow scope first anyway, then expanded from something that was already working.

The projects that struggled were the ones where ambition arrived before understanding.

So, the honest answer to “which route?” is that it depends less on your budget than on how well you understand your own problem. If you cannot yet name the task that eats the most hours, you are not ready to commission a build. Start small and let the cases tell you.

ADVERTISEMENT

What neither route should touch is judgement. The value an adviser brings was never collecting bank statements. It is reading a client’s circumstances, structuring the case and knowing which lender will take it. Automate the admin around that. Leave the decisions alone, and keep a person able to step into any case at any point.

One last piece of advice. Before you look at a single product, look at your own numbers. Pick the task your team repeats most. Measure what it costs you now: hours per case, days lost waiting, enquiries that went quiet and never came back. Then measure it again in three months

Matthew Elliott is co-founder and chief commercial officer at Nivo

Related

ADVERTISEMENT