Having worked at startups that wanted to take Alexa into the enterprise space, I'll say that Amazon has been absolutely opposed to the idea, which feels like a missed opportunity.
Up to a point, I can admire the product discipline that Amazon and Apple (Alexa and Siri) have shown in keeping their voice products focused on consumers. But after so many years, when do we finally get voice for the enterprise? At one time, I was very hopeful about this space, see: "I believe in Enterprise software for the Amazon Echo"
In late 2015 and early 2016 this space, of voice tools, was moving quickly and I thought that surely, within a year, we would see better support for enterprise applications. I was mistaken.
Here is an example of the excitement of those times: "Invoxia will allow Alexa to figure out who is speaking"
At that time, it felt like we were on the cusp of using voice tools in business. There was certainly a market. For instance, https://www.jwt.com/history/ has about 9,000 employees in at least a dozen countries. They reached out to me to talk about their "Pangea" project, which was a huge internal project to better map what teams and skills and resources they had internally. I had a meeting with them in the summer of 2016 and they asked "We are thinking we could put an Amazon Alexa in every meeting room, and then whenever we need someone with a specific skill, we could ask, hey, Alexa, do we have someone at this company who has this skill?"
At that time, I told them of some of the difficulties, but I was also hopeful that things would change a lot of over the next year. I was wrong.
Among the many, many problems we faced was simply getting Alexa to recognize the names of companies, when the sound of the name did not match a standard English spelling:
TATCHA
L’Oréal
L’OCCITANE
What was needed was a way train specific conversions of phonemes to text, for both Alexa and Siri. There are a small number of tricks that are available, such as programming a glottal stop:
However, in general, neither Amazon nor Apple wants to give us the tools we need, and there has been little progress in recent years.
I know that many people on Hacker News have read my book "How To Destroy A Tech Startup In Three Easy Steps" in which we were using NLP technologies to try to allow salespeople to send text messages to Salesforce:
As you can imagine, we hoped to move on to voice-to-text so that salespeople could simply talk to Salesforce.
It seems to me there is a wave of new startups that are waiting to move forward in this area, but no small startup has the resources to build a voice-to-text system from scratch. We need the larger firms to help set up the ecosystem, so that we can move forward in this area.
Where is the company that is going to give us the tools we need to start building great voice driven enterprise solutions?
Generally in BigCo land you need a value prop that will allow your buyer to do 1 or 2 things (or a combo) - take out massive cost, or drive large incremental revenue.
Most of the voice solutions I've seen kind of nibble around the edges of this, don't get traction, and slowly die a slow death somewhere deep in the middle of the RFP or evaluation process.
But this is something that enterprise customers keep asking for. I think the "pain point" is well known: companies can easily spend $10 million or $20 million customizing their Salesforce implementation, but then the salespeople hate it and refuse to use it. And then, if you are the VP Of Sales, a lot of your job becomes an endless cycle of harassing your salespeople to record their interactions with customers. This is how I explain it in my book:
-------------------
I asked him to run me through the pitch, and John gave a practiced recital:
Most salespeople are human-centered and enjoy talking with other individuals — but they hate dealing with computers. If a salesperson is selling shampoo to Sheraton Hotels, the best part of their day will be talking to the customer; the worst part will be when they have to go back to the office and deal with their company's reporting software. More likely than not, this will be Salesforce, the most widely used software for tracking sales.
Salesforce is ugly. Their interface is clunky. The poor salesperson has to sit down, bring up the website, click on a bunch of buttons, and navigate through a bunch of forms. The worst day of high school math was probably more fun for them.
Celelot aimed to change that. Instead of dealing with Salesforce, the salesperson would simply pull out their phone and send a text message to the Celelot system. For example, "Spoke to Carol. I just sold 1 million bottles of shampoo to Sheraton Hotels, rev 500,000. Contract August 1. Delivery September 1." We would use a set of computer techniques known as Natural Language Processing, or NLP, to take a message like that and pull out all the fields that were significant to Salesforce:
Contact: Carol Harrington
Customer: Sheraton Hotels
Product: Shampoo
Quantity: 1,000,000
Revenue: $500,000
Close Date: August 1
Delivery Date: September 1
Celelot would automatically identify who sent the message, connect it with their Salesforce account, and log the information in the system. Salespeople would never have to interact with Salesforce directly.
Apart from streamlining the reporting process for Salesforce specifically, Celelot could potentially become the default interface for all sales-reporting software (a category officially known as Customer Relationship Managers, or CRMs, of which Salesforce and Pipedrive are two well-known examples). That would be game changing.
Yeah, this is a classic 'nice to have' solution - it's a pain point in the process and so you hear feedback and requests for solutions.
But the real question to determine if it's a viable product/business is does it drive enough value that people will actually buy it, use it, and remain paying customers?
I'd say in your example above, it likely won't massively decrease costs (the salespeople are still having to interact with a CRM) or drive increased incremental revenue in any real way.
That's the same as saying the CRM doesn't drive value, in which case we might ask why companies spend tens of millions of dollars customizing their CRMs? If you're the VP Of Sales, presumably you need some way to manage the sales pipeline, which is why you got Salesforce in the first place. But if your salespeople won't use it, then that's the same as not having it.
The real problem with sales management systems are crossed incentives: The value of a sales person is their exclusive knowledge and contacts so nice voice interfaces to the knowledge vacuum hose are not going to make a change.
Up to a point, I can admire the product discipline that Amazon and Apple (Alexa and Siri) have shown in keeping their voice products focused on consumers. But after so many years, when do we finally get voice for the enterprise? At one time, I was very hopeful about this space, see: "I believe in Enterprise software for the Amazon Echo"
http://www.smashcompany.com/business/i-believe-in-enterprise...
In late 2015 and early 2016 this space, of voice tools, was moving quickly and I thought that surely, within a year, we would see better support for enterprise applications. I was mistaken.
Here is an example of the excitement of those times: "Invoxia will allow Alexa to figure out who is speaking"
http://www.smashcompany.com/technology/invoxia-will-allow-al...
At that time, it felt like we were on the cusp of using voice tools in business. There was certainly a market. For instance, https://www.jwt.com/history/ has about 9,000 employees in at least a dozen countries. They reached out to me to talk about their "Pangea" project, which was a huge internal project to better map what teams and skills and resources they had internally. I had a meeting with them in the summer of 2016 and they asked "We are thinking we could put an Amazon Alexa in every meeting room, and then whenever we need someone with a specific skill, we could ask, hey, Alexa, do we have someone at this company who has this skill?"
At that time, I told them of some of the difficulties, but I was also hopeful that things would change a lot of over the next year. I was wrong.
Among the many, many problems we faced was simply getting Alexa to recognize the names of companies, when the sound of the name did not match a standard English spelling:
TATCHA
L’Oréal
L’OCCITANE
What was needed was a way train specific conversions of phonemes to text, for both Alexa and Siri. There are a small number of tricks that are available, such as programming a glottal stop:
http://www.smashcompany.com/technology/using-a-glottal-stop-...
However, in general, neither Amazon nor Apple wants to give us the tools we need, and there has been little progress in recent years.
I know that many people on Hacker News have read my book "How To Destroy A Tech Startup In Three Easy Steps" in which we were using NLP technologies to try to allow salespeople to send text messages to Salesforce:
https://www.amazon.com/dp/0998997617?psc=1&ref=ppx_yo2ov_dt_...
As you can imagine, we hoped to move on to voice-to-text so that salespeople could simply talk to Salesforce.
It seems to me there is a wave of new startups that are waiting to move forward in this area, but no small startup has the resources to build a voice-to-text system from scratch. We need the larger firms to help set up the ecosystem, so that we can move forward in this area.
Where is the company that is going to give us the tools we need to start building great voice driven enterprise solutions?