Startup Daisee applies AI to business problems
- 23 February, 2018 07:25
Daisee, an Australian startup focussed on applying artificial intelligence to commercial problems has been formally launched after six months in stealth mode, revealing how it has been able to optimise supply chain issues for one of its first customers.
Daisee is cofounded by CEO Richard Kimber, Google’s first regional managing director for Australia and New Zealand, and COO Frederick Charette, who has held multiple senior executive roles with some of the world’s largest financial services and utilities companies.
Kimber said Charette brought consulting and client engagement skills to the company. “A lot of AI companies come at it from a very technical viewpoint. Selecting which commercial opportunity to work on is as much of a challenge as deciding what technology to use.”
Daisee offers two AI applications: Daisee Harmonee for application to structured data and Daisee Ekko for unstructured data.
Both are based on the Maestro AI platform developed by Deakin University, which is an investor in the company along with Alium Capital, Thorney Investments and a number of private investors.
Kimber told Computerworld: “We looked at all the different university applications of AI and came across the AI platform at Deakin University, which had been developed over several years and has been used predominantly in academic research.”
He said the university had already made some moves to commercialise Maestro. “They have been quite reactive, responding to inbound enquiries rather than going out and seeking applications.”
Daisee will apply a range of enhancement to the platform to enable its use in enterprises. “We are taking a proven industrialised platform that has worked on vast data sets across multiple application areas and applying it to business problems,” Kimber said.
“It is an exciting opportunity because it gives us access to a proven methodology, toolset and code base.”
Kimber described Maestro as an orchestration engine. “It is about bringing all the components together, on the training side of the models and the run time elements. That’s the tricky part of AI: there are many components to need to bring together.”
Daisee is also working with UNSW, but that university does not have equity in Daisee. Kimber said: “UNSW has contributed IP around AI and we have a link into their academic research.
“They have a different focus, more on robotics. Deakin is much more focussed on time series data. Our intent is to blend the best of the research to provide a ‘best of’ commercialistion vehicle.”
Kimber said one of the strengths of Maestro was how it enabled data scientists to optimise their input in the development of applications.
“We believe one of our key selling points will be our ability to leverage our data scientists. The second part of our strategy is to leverage the data science teams in the universities. That is part of our agreements with Deakin and with UNSW.”
Flower wholesaler taps AI to anticipate demand
The first named customer for Daisee Harmonee is Brisbane based flower wholesaler Fresh Flowers Group. It engaged Daisee to help it better manage supply and demand.
Kimber said this was particularly challenging for the industry: its products are extremely perishable and demand fluctuates wildly, driven to a large extent by events such as Valentine’s Day.
“Daisee developed an AI system to provide some rules and guidelines to enable staff to know how much to buy. We took a whole bunch of data about significant calendar events and fed in data from their sales system to look at their past history, and created rules based on that data to make predictions about likely purchases and tested that against three months of data.
He said the project had taken three weeks and had projected cost savings of 30 per cent. “We compared our solution to running a traditional regression that you would run in an Excel spreadsheet and it is dramatically better.”
Fresh Flowers CEO Peter Lynch said: “There were peaks and troughs in our business that I did not realise were there. We had totally missed a couple of peaks periods.”
Kimber said the application could be adapted to anticipate demand and optimise supply in other industries.
“Supply and demand problems are in every single industry. This application has huge impacts for the bottom line of many industries.
“Our approach to going to market develop a generalised AI application then apply it to a particular company so we will have repeatability and reuse of the application.”
Daisee Ekko analyses every call
For the other application, Daisee Ekko, Kimber said: “We are in trials with a number of customers but we are not allowed to share their names. They are primarily in the insurance sector.
He said these early applications applied AI to the analysis of conversations between contact centre agents and customers.
“Daisee Ekko can listen to every conversation, convert speech to text and run AI algorithms to emulate what the human quality teams do to score that call. It will replicate the best human scorer.
“It will also help you determine sentiment: it can detect emotion and determine when a client is upset or angry.”
He added: “It is extremely easy to deploy because it is cloud based and call centres are moving to being completely cloud based.
He suggested the application could find a receptive market with Australian financial services organisations now the subject of a royal commission.
“Australia’s financial services companies need to take more steps to find out what is happening in their call centre, that is where a lot of mis-selling occurs.”
Australian businesses ready for AI
At its launch Daisee released a report on Australian business sentiment towards AI saying it revealed “high levels of awareness and a willingness to invest in the technology’s potential to reveal critical business insights and new market opportunities.”
Overall, Daisee said: “Australian companies forecast spending significantly more on AI between now and 2022. In particular, the report noted a doubling in the number of Australian companies expecting to invest over $1 million in AI between now and 2022, rising to 13 percent from six percent today.”