Let me say at the outset, my hype antenna goes up whenever a vendor talks about machine learning. It comes from years of frustration with machines which have supposedly been trained to transcribe voices. Even the hyperscalers only deliver 85 to 90% accuracy when it comes to transcribing my book and blog interviews. I used to think it was the noise in my audio recordings. After all, if doctors can dictate their complex lingo and get pretty good automated transcription what is so complex about my workloads? Well, machines can learn their individual voices and accents - doctor handwriting is a bigger challenge, I think:).
In my case 95% of the voice is not mine - it is that of my interviewees. In tech, many of them have strong German, Indian, Chinese, S. African and other accents. And also in tech we use all kinds of big words and acronyms (thanks, Gartner!).
Machines can be trained well if you can limit the taxonomy they are expected to learn and then you expose them to LOTS of relevant data. It's the latter part where most vendors fail to measure up. They may have a number of large on-premise customers. Vendor machines cannot easily access that data - they are either not allowed to access the customer data center or even if the data is extracted, customers rarely give them permission to benchmark against other customers. If they are in the cloud, they do have easier access to the data, but few have sought or received permission to anonymize the data and use it for ML.
Which is why I was intrigued to hear about Sidetrade and its database of accounts receivable and cash collection data accumulated over 20 years - representing over 300 million invoices valued at over trillion dollars. I invited David Turner, the CMO to talk about their value proposition and have extracted 20 minutes from his presentation below. He discusses his Augmented Cash application and two of his large customers - Manpower and XPO Logistics.
The pandemic has been good for business. Q2 was the strongest in their history. Cash flow is king with the global economy in turmoil and their insight delivered via their bot, Aimie clearly helps collectors.
You would think ERP and CRM vendors would be all over this with their order processing and eCommerce capabilities? Sidetrade depends on them to process customer orders, do the billing and record the cash collections. (Sidetrade has a roadmap to itself offer some of the transaction processing) But go back to my comments above. Those vendors likely don't have enough real-life customer collection data available to train their machines.
