Flink Labs believes that machine intelligence works best when used to augment human intelligence, not to replace it.
We are not an AI company.
Artificial intelligence and Machine learning are tools we use to solve our client’s problems and deliver measurable value.
We have a pragmatic approach to Artificial intelligence, recognising that most organisations don’t have massive training datasets nor millions of dollars to spend on model development. Instead, we focus on practical applied Machine learning and Artificial intelligence methods that can be used out of the box, getting better over time, and providing measurable benefits right away.
We keep abreast of the current AI breakthroughs leveraging state of the art methods, such as transfer learning, convolutional neural networks (CNN) for Natural language processing (NLP) and pre-trained models such as GPT-3 and Word2vec to supercharge our AI projects.
Artificial intelligence and Machine learning play a role in almost all out projects. Clustering algorithms and visualisations are heavily used during exploratory data analysis phases. NLP based automated narrative generation is used to produce textual descriptions of data, and deep learning algorithms are widely used for classifying and analysing text across hundreds of thousands of user comments.
Automated narrative generation - We use NLP and statistical analysis to generate “robot text”; providing human readable insights and overviews in our data portals. The use of AI to generate insights and highlight outliers saves hundreds of hours for our clients, empowering their skilled analysts to focus on deeper discovery and analysis rather than writing “boilerplate” copy.
Image and text data mining - Working with IP Australia, we used advanced Natural language processing and image recognition to analyse patent applications, extracting metadata and structure for use in deep learning pipelines. We also used text classification and clustering algorithms to assist trademark examiners in assessing applications.
Customer movement and purchase behaviour - Combining supermarket trolley tracking data with point of sale purchase information, we used Markov chain analysis to construct patterns of shopper movement through a supermarket, identifying patterns in shopper behaviour that was used to help optimise store layout and product placement, resulting in an increase in average basket spend.
With backgrounds in intelligent agents and recommender systems, patents in predictive analytics and Artificial intelligence, and active AI projects across a range of clients we are uniquely positioned to help with your Machine intelligence projects.