Pretty much every vendor keynote finds a way to sneak in an Alexa demo - or a chatbot announcement with an adorable nickname.
It's easy to get cynical on this; I've made my share of wiseacre remarks. Still - customers want to know their vendors have a roadmap for incorporating that tech.
They don't want to be told next-gen tech will
change the game solve everything, or require them to move to an all-new platform. And they don't want to hear it will cost them an arm and a leg (e.g. new licenses to use it).
Last year's Acumatica Summit included one of the first Alexa demos I saw that pulled ERP data into a live Alexa query. But at this year's Acumatica Summit, Ajoy Krishnamoorthy, VP, Platform Strategy, presented a more comprehensive vision of how next-gen tech fits into Acumatica's roadmap.
Tackling IoT and AI - with a practical bent
The focus? Practical uses that tie back into ERP processes. I hope that doesn't sound too boring, because some of the walk throughs were pretty nifty. Such as: an appearance by Acumatica customer Cherrylake Tree Farms to demo their IoT agricultural soil sensors (pictured above - video here). I'll reflect on what I saw, and share a few offstage insights from my sit down with Krishnamoorthy.
After outlining development/platform priorities which included microservices, containers, and serverless concepts, Krishnamoorthy moved into intelligent apps and edge application building blocks. Queue the usual suspects: AI/ML, NLP, blockchain, and IoT/RFID.
He shared these scenarios:
- AI/ML - a real-time personalization demo showed Acumatica's work with customer Shoebacca, using Microsoft Azure and tech from Acumatica partner Kensium to implement product recommendations. Microsoft Azure's Recommendation Solution and fraud detection with Apruvd were part of the mix.
- IoT for real-time measurements of crucial soil data with Cherrylake Tree Farm.
- RFID for real-time tracking, utilizing tech from IMPINJ to trace items in real-time, from creation to sales. One standout: identifying errors in pack and ship. Showing his satirical/comedic side, Krishnamoorthy tossed aside a mug from Sage that did not pass RFID muster (check my Acumatica Summit review podcast with Brian Sommer, embedded below, for more color on that).
- Hyperledger Sawtooth supply chain blockchain project - Krishnamoorthy shared a blockchain project underway with Acumatica customer Bluefin. Using tech from Intel and AWS, the project tracks and authenticates supply chain activity.
- Natural User Interface with Google Dialogflow - CEO Jon Roskill joined Krishnamoorthy on stage to demo another "Natural User Interface" (NUI) ERP integration in Alexa, for scenarios like invoice approval. How was this an advance on last year's Alexa demo? Because it included the ability to authenticate and authorize the speaker's voice profile. Security is a major impediment to enterprise voice adoption. Krishnamoorthy had some fun with this one, showing how different voices couldn't get into the system - including Roskill's. The tech included Alexa for Business from Amazon Web Services. Alexa isn't the only voice game in town. Krishnamoorthy announced that Acumatica will also be working with Google Dialogflow to expand their use of NUI.
The strategy behind the demos - "no AI science projects"
Krishnamoorthy wanted me to know that the comic timing around Alexa giving people authorization grief was not pre-planned:
None of the jokes were rehearsed. People think, "Wow, you had the Alexa thing setup that you wanted to fail," and I'm like, "No, why would I want it to fail the first time?" They think all the jokes were preplanned.
Stagecraft exposed, Krishnamoorthy says Amazon has been eager to partner with them on ERP-meets-Alexa. He knows customers are wary of extravagant AI demos. Krishnamoorthy only wanted to show customer examples:
I'm a big believer in this whole pragmatic approach. To point, without a customer, I wouldn't do any of this. Otherwise, it becomes a science project.
Though the use cases weren't all generally available products, Krishnamoorthy wanted to show field examples:
It's validating technology for customer purposes, or for making an informed decision. That's why I chose Bluefin. That's why I showed Shoebacca.
Krishnamoorthy needed to see the tech for himself. He hopped on a plane to see Cherrylake Tree Farms in action, and talk IoT. It was eye-opening:
I flew into Orlando, walked on their farm... I got a feel to say, "Okay, this is real sh@t." When somebody explains what's happening in a citrus farm, that's as foreign as it can be for me. It's not an industry I grew up in. But when I sat down in a conference room, [showing how their soil sensors could tie into our reporting], the guys lit up seriously.
One of the most excited people? A Cherrylake farmer.
The guy who was the most excited in that conversation wasn't the tech guys... He is a producer. He actually is a farmer. His eyes were lit up.
So why was the farmer so excited?
Because he saw the opportunity. Because right now what happens is they have twenty people working there on the farm, measuring the humidity of the soil. Inserting into holes. Imagine putting a sensor and getting that info. Those twenty workers can go do some other crucial things. That's the type of stuff he was so excited about. He said, "Yeah, that's what I want to do."
Here's what the soil sensors pick up:
This IoT project is in the pilot phase, but I expect it to be live and integrated with Acumatica by next year's Acumatica Summit. Cherrylake already uses robots to rotate tree saplings for maximal sunlight. The soil sensor readings will provide vital info on soil humidity, and other factors that predict if a tree is going to survive or not.
The recommendation engine built for Shoebacca by Kensium is already running for several customers, and should be available out of the box in an upcoming Acumatica release. That fits Krishnamoorthy's goal of avoiding premium costs, and embedding "intelligent apps and services."
The other priority: don't build a service from scratch if a third party built one already. Take Apruvd for fraud detection:
I don't want to reinvent the wheel. I have no pride in saying, "I built this from the ground up." Apruvd has already done it. Signified has done it. A few others have done it. And Apruvd is already in discussion with our customer. I said, "Great. Customers are validating it."
Where possible, Krishnamoorthy wants Acumatica to share their own work. That includes Alexa:
Later this year, we'll put out what I demoed as a public skill. Anybody can download it, and start using it with Acumatica. That's the part I'm working with the Alexa engineering team.
A big validation for Krishnamoorthy came when Amazon's CTO demoed the Acumatica integration at an Amazon event, in front of 40,000 people. That brings us to Krishnamoorthy's third point: don't stick with one vendor's service when customers need choice. Thus their NUI work with Google's Dialogflow.
My take - an overall AI narrative is needed
Krishnamoorthy had more to say on blockchain use cases that I'll share later. The Acumatica approach of "no next-gen until we can show something with a customer" is a refreshing one. They'll need to show progress on general availability of these solutions for next year's event. But to be fair, Acumatica and their customers are on a similar course of readiness.
Given their midmarket focus, Acumatica's customers will be counting on them to deliver. Most customers don't have the budgets or in-house talent to build their own AI solutions from scratch. Whether they come from Acumatica-only, or from a series of well-integrated partnerships, doesn't really matter.
Acumatica is smart to avoid re-inventing the wheel on functions from fraud detection to recommendation engines. An accessible platform and a boatload of APIs is what counts - along with the ERP expertise to automate into transactions and reporting. Not to mention an intimate grasp of customer requirements in their chosen verticals. That's why it's good to see Acumatica joining the Open API Initiative. If blockchain gets traction in their customer base, I'd like to see them join Hyperledger as well.
Acumatica will face hard decisions on which AI capabilities to build into its own xRP platform. NLP capabilities are fine to source from a third party, but some rules engines may be better to build. Overall, Acumatica's partner focus should serve them well in their next-gen push.
One thing most people missed: Acumatica has done a stealthy good job of building a developer community in the last year. Thanks to new Platform Evangelist Mark Franks, they are building momentum, including some terrific Hackathon projects this year.
In our show review podcast, Brian Sommer went on his stump speech about how no ERP vendors are ready for the scale of relevant third party "big data" that will impact the business decisions of tomorrow (and by tomorrow, he means tomorrow - not next year). That's a big problem yet to be solved. Still, Acumatica has done a good job of last mile ERP automation from external sources.
I'd like to see Acumatica tie their practical AI approach into a bolder narrative on jobs and productivity. In the Cherrylake example above, you have twenty soil measurement workers no longer needed for that. It sounds like they will contribute in higher value ways. Still, customers are going to have hard questions about the reach of automation. They want to know how it might change their companies - and their own roles.
Most of my Acumatica customer use cases touch on how repetitive tasks have been automated via cloud ERP, and how workers have been freed up for more impactful roles. That's the journey from the "operations and compliance scramble" to a real customer-focused business - with time for strategic data crunching.
Most of Acumatica's customers are trying to make that journey, and using Acumatica to do so. The impact of AI and automation fits in well with that storyline, but it raises new questions as well. At next year's Summit, it would be great to hear Acumatica tell that story, without losing focus on the practical.
As for a cute chatbot, well, no sign of those. I would not expect to see one either - unless customers need it. If so, I've got an idea for a name...
Updated 2:30 pm UK time after I crossed the pond, with a number of small tweaks to improve readability. No opinions have been changed.