I Have No Life! Machine Learning, Time To Step Up!

You Can Follow Me on Snapchat Here to see all things behind the scenes at 20VC and SaaStr!

Every Saturday and Sunday, I spend 12 hours each day in the studio taking out ums, ahs and pauses, from the podcast episodes to be released for the following week. This is potentially one of the most monotonous and robotic tasks known to mankind. Step in machine learning, for those that are not aware; machine learning is a type of AI that provides computers with the ability to learn without being explicitly programmed? It has the ability to give me a weekend! ;) 

Question:

Why are there not elements of machine learning baked into Garageband? I have now done over 600 published episodes, an estimated 15,000 minutes, surely enough data has passed through Garageband to determine what is an um and what is a pause to what is not? They are very clear and obvious to determine and stand out when one becomes accustomed to how they appear in the audio file (see below). 

The Problem: 

Now this is a very nascent and real problem for me and many other podcasters. However, at present there are 180,000 active podcasts. Many of those have the same creator, I too also have SaaStr! So estimating that from 180,000, there are 90,000 podcast producers (optimistic suggestion), this is our TAM (Total Addressable Market) for our machine learning podcast editor. Going one layer deeper, I always believe that attaining 1-2% of your TAM is doing very well and although unlikely, a goal that the founder should be striving for. That gives us a realistic customer base of 900-1,800 customers. With a traditional pricing mechanism this would have a cost of circa $50. This gives our metaphorical startup an MRR of $45,000 and an ARR of $540,000. Although, decent figures, it is really not a venture-backable business unless we see some insane growth in the podcast production market.

What is The Future: 

So what happens to the mass of machine learning products like this that have very clear value propositions but do not have the ability to be financed and ultimately produced and sustained. Is this a case for angel investors who do not require the lofty $Bn exits that VCs require? Will this see the rise of a generation of lifestyle businesses, that are self-sustaining and make good money but can never be considered a home run? For machine learning in particular, will this see the creation of your IBM's or Oracle's producing 'out of the box', so to speak machine learning capabilities. 

I would love to hear your thoughts on this, please do comment and let me know what you think? 

 

Why Julia Roberts Gets VC Content Marketing?

Why Investor Branding Is Like Exercise?