Proven Need

Reading Time Time to read: 3 minutes

Beginner’s guide to solving problems with Applied AI

Akriti Dokania

Akriti Donkania

There’s often a misconception held by entrepreneurs that you need PhD-level technical knowledge to work with Artificial Intelligence (AI). However, there’s an important distinction between Core AI and Applied AI which we will discuss in this article.

Key Takeaways:

  • It’s straightforward to spot practical applications of AI use cases

  • There’s three steps: Identify the behaviour, identify the input, and apply the AI technique

  • Most common techniques include: classification, regression / prediction, optimisation, computer vision, pattern recognition and NLP

“Core AI” startups are mostly focused on building the technology infrastructure for the AI industry as a whole. They are somewhat similar to cloud services, you solve real world problems by building on top of this infrastructure.

An example of a Core AI startup would be DeepMind, a Google company that has built learning technology on top of multilevel neural network algorithms. The technology can enable machines to play video games like humans, make food production more efficient (sorting the good cucumbers from the bad), as well as automate the analysis of human eye samples in hospitals.

On the other hand, Applied AI startups enable companies to perform certain tasks using AI, or apply AI to specific industry verticals.

Here’s some examples of Applied AI startups in our portfolio:

  • An algorithm that learns your style preferences and recommends clothing / outfits (Thread)

  • An intelligent matching platform, capturing legal counsel requirements from SMEs and putting them in touch with relevant legal experts (Lexoo).

Finding a scalable market opportunity using Applied AI can be broken down into three steps:


1) Identify the behavior

Look for activities in your own life that are repeated, and could benefit from automation. In particular, those routines that can be improved by learning from the past.

For example, think of the routine questions that customer service agents are asked repeatedly, the team behind Gluru noticed this recurring behaviour and used AI to automate customer service support.


2) Identify the input

Input.png


The second step in this process is figuring out the input. These data inputs can be broken down into three types: text, voice, image (or a combination of two or all three). You need to break down the task you have identified and see what data you can extract from it.  
 

3) Apply one of these techniques

At Forward Partners, our investment team invited AI experts to help us build an exhaustive list of Applied AI use cases and match them to common machine learning techniques:

Technique

Definition

 Use Case Example

Classification  

Clumps data into buckets. Categorization and organization of data.

Health Automation - organizes all the health data to motivate users to challenge themselves Eg: TicTrac

Recommendation engine - bucketization of user preferences to provide ecommerce recommendations. eg: Thread

Regression/ Prediction

Finds a relationship among sets of data

Pricing prediction - Hotel room price prediction. eg Pace

Optimization

Reduces large sets of data to give the most efficient output

Motion planning - to provide the simplest cost effective urban mobility through public transport eg five

Pattern Recognition

Watches for trends in data sets and magnifies these trends

Voice recognition - Transcription as a service eg Trint

Analytics - Financial consumer spending application eg. Cleo

Computer Vision

This involves analyzing, processing and acquiring of digital images or videos

Image recognition - Understand emotions based on people's faces eg Realeyesit

Natural Language Processing

Analysing and interpreting language and speech and using the same to imitate human behavior

Content generation - News content creation in local languages eg Signalmedia


In conclusion

You don’t need to have a PhD in AI to build an Applied AI startup. However it’s important to get familiar with the different AI techniques available, then you can start applying them to real-world problems.

Akriti Dokania

Akriti Donkania

While pursuing her MBA at London Business School, Akriti joins us to be an Investment Associate at Forward Partners. Prior to this, she has been a product manager for the user experience team at Microsoft and Alexa at Amazon. Right before business school, she was an entrepreneur, running a fintech company. She helps with deal sourcing, screening and researching new areas of investment. Her other interests lie in working on new product designs and communicating via gifs.

Apply for Office Hours

We’re looking for great entrepreneurs with great ideas.

Apply here

0 Comments

Post your comment

Similar Guides