The same applies to financial advice – notwithstanding the fantastical allure of a machine or robot designed to replace a human being.
There is a lot of hype around the promise of such technology, however we are still some years from the promise becoming reality.
In any event, I believe that while the human race can produce a driverless car that is safe, by replicating human motor and sensory skills, a fully automated financial advice system that can predict markets with human behaviour is really quite the impossible.
And just as Elon Musk’s renowned Tesla vehicles now offer autopilot functionality, the truth is that most conventional vehicle manufacturers will not bring any new technology – especially driverless software based systems – to market without effective and thorough product testing.
As the regulators in Australia start to get their head around the pace and depth of the automated digital advice world, this blog is to offer a quick snapshot of the various development and automation levels of so-called ‘robo’.
I have broken this into four levels, and taken the fourth as my starting point, and we are projecting forward in time when looking at the fully autonomous notion of digital automated advice.
Level 4: autonomous
Just like the driverless car, level four is the advice industry equivalent to ‘automated’ technology.
All functions of advice and compliance are performed automatically, with the exception of components that require algorithm applications to predict and mimic human behaviour across an infinite spectrum, like predicting markets to human behaviour.
The theory of replicating the brain’s thinking or feeling a ‘certain way’ at a ‘particular time’ or ‘unforeseen events’, is a very complex problem to solve. It is not a like the replication of a motor skill or sensory function.
Autonomous takes the client automatically through the fact finding process, setting out the client’s entire actual position (data aggregation of cash flow and net worth).
The system can determine the most optimal strategy and actions to meet all his or her goals and objectives, all within compliant preliminary parameters.
The output from client engagement tools map directly to corresponding strategies, they are optimised (via engine algorithms) to achieve the goals and objectives of the client.
Learning and intelligence via data aggregation and algorithms enables the system to learn aspects of human behaviour. The system is constantly learning a client’s cash flow habits, learning how to pre-empt issues and ways to reward and motivate customers based on them.
There’s no limiting the client to a risk profile, instead the system learns how people feel and respond to risk, teaching them how they can meet their goals.
The system populates applications for ‘straight through processing’ (STP) to investment and fund platforms and auto acceptance of loadings for automated insurance underwriting services.
The technology is robust, accurate with performance scalability of the calculation and analytics engines. It is designed to handle complex data flows and has rigour around the testing of its algorithms and compliance obligations.
It has a ‘human-centric’ user experience. The system is convergent across insurance, superannuation, investment and taxation.
Level 3: semi-autonomous
The semi-autonomous definition consists of the parameters of which advisers worked within a manual way, and built these into automated parameters.
This level has managed to enable at least four functions of the digital advice process to become automated, similar to cruise control on a car.
It means that the adviser is disengaged from physically operating a particular part of the advice process.
It has an enabled fact find to database (CRM) merge, client engagement tools, needs analysis to product/asset allocation, digital document editor with no post merge editing in word.
The system can produce 10,000 times more permutations than a human paraplanner, in seconds (with pre-defined parameters for running scenarios to get the best decision output for the client’s situation).
The parts of the advice process automated in this scenario include:
- Digital fact find (client’s actual position & net worth),
- Needs analysis and;
- A ‘guided decision model’ so that the target product (investment, insurance, super) or asset allocation is suitable.
It is driven around the client’s needs and decisions. It is fully automated and client driven from responsive technology, on any device.
Client engagement tools are semi-automated, navigated by the adviser as a conversation lead experience.
The advice engine is semi-automated and goals-based, with the ability to model any number of client goals, asset classes, portfolios, investment strategies with any mix of economic, regulatory and behavioural rules automated in milliseconds.
Client net worth, actual position and cash flow is automatically determined in real time with slight human intervention via data aggregation. Data feeds are automated, expanding across all advice areas.
Processing applications with STP has automation with select providers. Advice and services are tracked automatically. Reporting and analytics are fully automated.
Compliance legislative changes are maintained – real time – with advice suitability parameters automatically monitoring advice and content to prevent non-compliance.
Level 2: partially automated
This is the industry’s current position with incumbent technology platforms.
And by partially-automated, I mean that most functions are controlled by an adviser (human), in a manual setting.
Fact-finding has become digitally automated, modelling can be carried out in a partially automated way and requires an adviser to drive the inputs manually, it’s still limited with linear algorithmic calculations as the software is driven by Excel spreadsheets, built workflows to bring together.
Product comparators are automated. There is no automation for tracking client goals against advice strategies and it does not incorporate all facets of holistic advice.
It has limited digital client engagement tools to assist the adviser conversations with the client, for educating advice concepts. Data feeds are automated, but not across all advice areas. Processing applications is mainly manual and forms (paper) based.
Level 1: minimally automated
This describes the financial advice sector's past. Level one was mostly manual and highly administration intensive.
The adviser (human) controlled the end-to-end advice delivery process using paper-based forms and duplicate data entry points for fact finding a client’s actual position. Typically, this level used a whiteboard to draw scenarios and explain advice strategy concepts.
Excel spreadsheets for modelling or linear calculator tools that worked in isolation to automate some silo aspects of the advice process.
It was difficult to track client goals against advice deliverables with transparency, and was traditionally product focused. Compliance? Labour intensive, and reporting of practice and client information was virtually impossible due to multiple data sources.
Why no level five? Well, we may one day add a level five, ‘fully’ process – once we can predict human behaviour and financial markets. Mars can be our limit – in line with Elon Musk’s Mars adventures.
Jacqui Henderson is the founder and chief executive of Adviser Intelligence.
