HPE’s AI Strategy for MSPs in the SMB Market
HPE's GreenLake for Large Language Models aims to make it easier for MSPs to deliver AI apps to their SMB clients.
July 10, 2023
![HPE’s AI Strategy for MSPs in the SMB Market HPE’s AI Strategy for MSPs in the SMB Market](https://eu-images.contentstack.com/v3/assets/blt10e444bce2d36aa8/bltf823ede7ca2cde23/6523f023b7cd538610f18ac6/HPE-GreenLake-Large-Language-Model-3-scaled.jpg?width=700&auto=webp&quality=80&disable=upscale)
At the core of HPE’s AI foundation sits Determined AI, a machine learning software development environment for engineers and data scientists, and Pachyderm, a developer of software that automates data versioning and reproducibility. Pachyderm assists engineers in understanding the lineage of data used to build AI models. The software automates data tracking, something engineers did manually, to increase the speed and scale of development.
“Pachyderm couples with a pipeline system so that we can track transformations that were made to train data over time. That gives us confidence in where our applications came from,” said Evan Sparks, chief product officer for AI at HPE.
If a customer wants to delete certain information from a data set, he added, the software can track the lineage and confirm the change.
Determined AI and Pachyderm are about building, training and tuning AI models. To round out the AI solution, HPE has developed the “inference” software used to run and manage AI models from the edge, servers or cloud. The software also identifies performance characteristics, such as a change in data patterns that could be associated with fraud.
“That becomes a really critical point for customers who are worried about deploying those models,” Sparks said.
Those customers include SMBs that will be relying on MSPs to integrate AI solutions and protect their data.
Besides the software platform, HPE can provide MSPs with the critical infrastructure – including ProLiant servers – needed to deploy managed and on-prem AI solutions in any market. Having access to a full suite of services and hardware from one trusted provider can be helpful, given the increasing demand for AI applications, channel experts said.
Nearly every end-user Sparks speaks with, no matter where they sit on the company org chart, wants to know how they can integrate their company data with ChatGPT. Some consider contracting a cloud provider for that but are often spooked by security concerns, Sparks said. That’s where an MSP’s role as a trusted advisor can come into play.
“People are looking for a solution that is going to respect their enterprise data security and data sovereignty,” Sparks said. “They want that kind of experience from a provider they can trust.”
HPE executives at the Discover partner conference all said the company is in a strong position to enable MSPs to be that kind of partner in the new world of AI.
HPE’s primary AI users today are machine learning engineers and data scientists at small to large companies, Sparks said. They work on HPE’s open-source software platform and can purchase HPE support and enterprise features to complete an application. Meanwhile, HPE is ramping up the number of AI applications it creates for resale. If those applications don’t fit an MSP’s needs, they also have access to the open-source platform and HPE’s suite of AI services, Sparks said.
“As the company grows or its initiatives begin to scale, they can graduate to the enterprise-centric product suite,” he said. “We’ve seen a number of customers go through that transition over time. For SMBs, our software platforms really are power tools. What used to be done by a team of 500 data scientists and machine learning engineers, now can be done by two or three with similar levels of output.
“The other thing that excites me about what we’re doing around our entry into the supercomputing cloud market is, we are democratizing access to that resource to a much broader group of people,” Sparks added. “And so SMBs that never would even think about building a multimillion-dollar supercomputing system can get fractional time on something like that to accelerate their time to outcome.”
HPE’s AI professional services for MSPs include assistance in supporting hardware infrastructure, defining AI strategies, building applications and creating full life-cycle development and deployment plans, all for a price. HPE also has partners that can provide MSPs with those type of services.
“It’s somebody that we validated and can deploy their applications in a GreenLake environment,” Sparks said about the partnerships.
With the right tools and support, those AI applications are limited only by the needs and resources of the customer.
Sparks says he’s already seeing a number of useful and exciting use cases developed on the HPE platform. One is a theft-prevention application developed for a large retailer. The application tracks images in real-time from video cameras placed around the store and determines if somebody is stealing, either by taking an item off the shelf and putting it in their pocket, bag or other method. Clerks are alerted and can watch the video feed of the perpetrator’s actions. They can also send the feed to security and other law enforcement. The solution can be easily replicated for smaller businesses, Sparks said.
Another solution can search the entire history of support tickets and build that knowledge into a large language model app that allows users to ask highly customized questions about specific products and their failure points.
“You can imagine putting that agent on the shoulder of your support desk and really making them much more efficient at their jobs,” Sparks said. “That kind of solution can apply to SMBs and MSPs alike. So being able accelerate that knowledge is powerful.”
Sparks has seen a similar app developed for salespeople. It can run on a mobile device and provides them with “complete knowledge” of every item in their portfolio. The app can listen in real-time to the sales conversation and make suggestions for upselling applicable products and services.
“It’s not going to be perfect; none of this stuff is,” he said. “But if it learns and gives the human a bunch of leverage, we think it can be a tremendous accelerant to different kinds of businesses.”
Those businesses include the SMBs served by MSPs.
Sparks says he’s already seeing a number of useful and exciting use cases developed on the HPE platform. One is a theft-prevention application developed for a large retailer. The application tracks images in real-time from video cameras placed around the store and determines if somebody is stealing, either by taking an item off the shelf and putting it in their pocket, bag or other method. Clerks are alerted and can watch the video feed of the perpetrator’s actions. They can also send the feed to security and other law enforcement. The solution can be easily replicated for smaller businesses, Sparks said.
Another solution can search the entire history of support tickets and build that knowledge into a large language model app that allows users to ask highly customized questions about specific products and their failure points.
“You can imagine putting that agent on the shoulder of your support desk and really making them much more efficient at their jobs,” Sparks said. “That kind of solution can apply to SMBs and MSPs alike. So being able accelerate that knowledge is powerful.”
Sparks has seen a similar app developed for salespeople. It can run on a mobile device and provides them with “complete knowledge” of every item in their portfolio. The app can listen in real-time to the sales conversation and make suggestions for upselling applicable products and services.
“It’s not going to be perfect; none of this stuff is,” he said. “But if it learns and gives the human a bunch of leverage, we think it can be a tremendous accelerant to different kinds of businesses.”
Those businesses include the SMBs served by MSPs.
HPE recently launched its AI strategy for MSPs, GreenLake for Large Language Models (LLMs), a cloud computing platform that provides on-demand AI modeling in a supercomputing cloud environment. The platform is part of HPE’s $2.6 billion investment in artificial intelligence, including its purchase of Cray supercomputers; Determined AI, a machine learning software development environment for engineers and data scientists; Pachyderm, software that automates data versioning and reproducibility in AI models, and the in-house development of inference software.
That evolving platform is obviously geared toward high-end enterprise applications. It’s what you’d expect from a company with “enterprise” in its name. What’s less obvious is that HPE is casting a wide net to bring more MSPs serving SMBs into its fold. That move is apparent by HPE’s new Partner Ready Vantage Program, initiatives to break into new markets with MSPs, and an AI strategy that provides MSPs with the tools and services they need to create and deliver applications and solutions to the SMB.
AI Strategy & GreenLake
HPE’s Fidelma Russo
“Instead of having to buy a supercomputer, which I’m sure many of us do not have the budgets for, you can now simply subscribe to HPE GreenLake for Large Language Models as a service,” Fidelma Russo, HPE’s CTO, said at last month’s Discover conference. “There you will have access to one of the most powerful supercomputers in the world to train your data or leverage our models. There will be a button for that on the GreenLake platform, making AI accessible for all.”
Evan Sparks, chief product officer for AI at HPE and founder of Determined AI, which HPE bought about two years ago, took time to flesh out HPE’s AI strategy and how MSPs can leverage the platforms for SMB and larger customers. As part of its AI investment, Sparks was quick to point out, HPE has hired hundreds of engineers to help MSPs and others with their AI projects and strategies.
“We’re pushing forward as quickly as we can,” he said.
See our slideshow above to find out more of HPE’s strategy for MSPs in the SMB.
Want to contact the author directly about this story? Have ideas for a follow-up article? Email Jeff O’Heir or connect with him on LinkedIn. |
Read more about:
MSPsAbout the Author(s)
You May Also Like