Opportunities we're most excited by
The future of language generation
The state-of-the-art in Natural Language Processing domain has been advancing at a blistering pace, to the point where current models have an incredible capability to perform various tasks with very few training examples. GPT-3 was recently released by Open AI, and early results demonstrate its capability to reason over a very wide variety of text and tasks. We are excited in particular by the applications of Natural Language Understanding (e.g. Q&A systems, intent recognition) and Natural Language Generation (e.g. Chatbots, writing auto-suggestions). We are looking for new ventures that would deploy this powerful technology alongside strong commercial use cases.
Example startups: OpenAI, Replika, Casetext
Interactive storytelling and synthetic media
What does the future of media and gaming look like as AI technologies become more commoditised? With advancements in fields such as text generation and image generation, we see a potential for all types of media to be personalised to the audience and create the most engagement. We are looking for new ventures that leverage the newest developments in AI to create interactive experiences, allowing users to experience a unique story based on their choices and preferences. We are also interested in businesses pushing the envelope in synthetic copywriting, marketing and ad creation.
Example startups: Phrasee, emotif, Narrative Science
The no-code revolution
The no-code, low-code revolution continues, helping individuals of all stripes to become creators and entrepreneurs. Yet most low or no-code tools quickly reach the limits of their utility, with most companies requiring a significant engineering and development team to build tech products. Can AI bring about the next wave of no-code tools that democratise tech for everyone as well as help companies launch and test new products fast, in market, in a resource efficient way?
Our areas of interest include, but are not limited to, coding and testing in multiple languages, product design and prototyping, end-to-end ML, hardware designing / CAD sketching.
Example startups: debuild, Uizard
Back office hyperautomation
Many labour intensive, low value add back-office processes are still performed by humans; automating these processes would enable companies to re-direct their workforce towards higher value add activities, leading to cost savings for organisations and better quality of work for individuals. We believe there's potential to use a combination of AI data extraction, classifier methodologies, RPA and generative AI transformer models to revolutionise back-office operations. We are interested in new businesses either solving one complex vertical, or building a horizontal platform that enable a company to quickly automate any back office function.
Example startups: Blue Prism, UiPath
Machine learning operations
Similarly to how dev-ops automated the process of building, testing and deploying code to production and created new frameworks such as continuous delivery and integration, ML ops aims to formalise processes to deploy machine learning models into production. Going from first model experiments to deployable code in production is a big challenge in ML with a lot of effort spent in ensuring replicability, and consistency, as well as monitoring the performance of ML systems over time as incoming data may change. ML Ops is still in its infancy, with no standardisation yet in the market, so tech companies have typically built their own ML infrastructure and deployment automations.
We think there is an opportunity to create an end to end ML ops infrastructure product as well as to build products that automate a slice of the deployment process and can be plugged into popular architectures or products.
Example startups: MLflow, Seldon, Faculty
Augmented Intelligence marries the computational power of machine learning with the creativity and intuition of human intelligence to enable better decision making across the entire organisation. Combining human and machine intelligence becomes particularly important in systems where the cost or risk of failure is too high or the AI is not evolved enough to take humans out of the equation such as identifying fraud, cybercrime or when making high stake business decisions.
We think that a smart frontend and UX that helps humans better understand and interpret machine learning predictions will help them make better decisions based on those model outputs. Intuitive leaps and the interpretation and decision making outcomes made by humans can then be used to reinforce machine learning models and further machine learning towards machine cognition.
Our world-leading accelerator program invests in founders who have a product in market and early traction, who want to leverage our team and partners to scale.
Our Venture Studio
Our venture studio invests in entrepreneurs who want to build a startup from scratch, with almost zero risk. Bring us an idea, or apply to launch one of our existing concepts.