As Microsoft And Google Battle Over AI, Some Partners Are Already All-In On ChatGPT
Employees with VCPI, Sourcepass and Net Friends talked to CRN about ChatGPT and generative AI in the channel.
Made with OpenAI’s Dall-E image generating AI program.
Some partners tell CRN that even as Microsoft, Google and other tech giants fight out an emerging battle over who has the most compelling artificial intelligence (AI) tools, they are adopting and testing these tools to help with internal processes and to potentially recommend to clients.
All three partners who talked to CRN are all-in on ChatGPT, a generative AI tool created by OpenAI, which has a multibillion-dollar investment from Microsoft and whose technology is being rolled out into multiple Microsoft offerings from Bing to Edge to Teams.
Stephen Eiting, a sales operations manager at VCPI – a Milwaukee-based managed service provider (MSP) whose partners include Microsoft, Citrix, Cisco and N-Able – told CRN in an interview that he’s used ChatGPT to write a project charter, change orders, user guides and even a program script for Microsoft’s PowerShell task automation and configuration management program.
“It saves me endless hours every single week,” Eiting said.
ChatGPT And Generative AI In The Channel
CRN has reached out to OpenAI for comment.
For the project charter, Eiting fed ChatGPT stakeholders, project managers and information on the risks involved to get usable text.
“I learned from that experience that as you become more conversational with it, it really presents a really good result,” Eiting said. “And I wasn’t able to just take that. It wasn’t complete. But it did 70 percent of my work for the project charter (in), I don’t know, 45 seconds. Which is great.”
Eiting, who also maintains a personal blog about technology, has been impressed by its ability to generate post ideas and to write the actual entries.
“I asked it to go into a book about cyber warfare, and I got seven chapters deep,” he said. “And it just kept going with this quasi-nonfiction, fiction novel that I was reading. I was fascinated because it just kept going. How long could it go? I thought to myself, ‘Should I publish this on Amazon?’”
In theory, the tool can help smaller MSPs without accounts to popular customer relationship management (CRM) tools, he said. MSP workers could write a script for calculating acquisition costs with multiple variables and plug in customer or prospect information.
Nick Ross, vice president of product development at Sourcepass – New York-based MSP whose partners include Acronis, Dell, VMware, SentinelOne, Microsoft and Fortinet – told CRN that he uses ChatGPT to help translate concepts and ideas into writing.
He’s used ChatGPT to correct his grammar for posts on his MSP-focused blog, for creating templates for email and marketing campaigns and even for low-level programming.
“I’d love to have it a little bit more rolled out and baked in certain ways than it is, but, I mean, it’s a game changer,” he said. “It’s a technology that has the excitement of a blockchain or crypto, but actually has way more applicability to businesses and the things that we do to reshape the world.”
Programmers’ jobs are still protected by the need to explain what a user wants, the ability to read code and the ability to troubleshoot, he said.
“It’s not going to go out and build you a full front end and back end that you can maintain,” he said.
Still, Ross sees ChatGPT as a helpful tool for translating jargon from product managers and developers.
For small MSPs limited by employee count, in theory, ChatGPT can write a business plan, do the market research based on information at its disposal and use virtual agents for sales calls, he said.
“There‘s the limitless possibility to being able to scale out a business, at least in the forefront,” he said.
John Snyder, CEO of Net Friends – a Durham, N.C.-based MSP whose partners include Microsoft, Nextiva and Palo Alto Networks – told CRN in an interview that he’s already using the paid version of ChatGPT, which is faster than the free offering.
“I was eager to sign up for it,” Snyder said. “I’m never gonna miss that $20 a month because I use it so, so much.”
Other AI Tools
Eiting has also tried other generative AI tools, including Midjourney, which can generate an image based on a user’s query. Although impressed by the tool, Eiting said he feels some work remains on Midjourney’s ability to create images of everyday objects that don’t look “creepy,” he said.
ChatGPT creator OpenAI has a Midjourney competitor called Dall-E, which also works on the generative pre-trained transformer 3 (GPT-3) powering the ChatGPT chatbot program.
Dall-E entered beta in July 2022, and OpenAI removed the waitlist in September.
ChatGPT was released in November. Both have been free to use with limits on the amount of content they can generate for users.
Snyder has played with Dall-E with his children but he has not found a business case for it yet, he told CRN.
Ross told CRN that his fascination with the current crop of generative AI tools came when he tried Microsoft’s GitHub Copilot.
Microsoft made Copilot generally available in July 2022. It uses OpenAI Codex, itself a natural language-to-code system based on GPT-3.
Ross became a self-described “evangelist” for Copilot, described as an AI pair programmer that can suggest code and entire functions in real-time from the editor.
“That was the first piece of technology I’ve used in a long time where I said, ‘Holy crap. This is amazing. This is going to change my life,’” he said.
For coding, Copilot saves Ross time “because I can type out a comment of what I want and it’ll tap complete. And I’ll be able to move really quickly on things that I used to have to Google.”
On Microsoft’s latest quarterly earnings call, held in January, CEO Satya Nadella called GitHub Copilot “the first at-scale AI product built for this era.” The program has surpassed 1 million users to date. He said that GitHub overall has 100 million developer users.
Eiting said that some tools on the market – such as Espressive’s Barista, an AI-based virtual support agent that promises to eliminate 70 percent of help tickets – worry him about potential replacements for engineers and IT help.
Another intriguing AI product on the market for Eiting is AI contract-drafting tools such as Spellbook by document automation and company management tool provider Rally.
Spellbook also uses GPT-3. But accepting the liability from a contract drafted by AI might be too much risk for him to take at this point, Eiting said.
He said he hopes OpenAI, Microsoft and other AI vendors maintain guardrails around the potential to use ChatGPT to produce malicious content and malicious code.
“They are walking a very fine line,” he said. “And I really hope that it continues to be used in a positive way.”
AI Tools And MSPs
Back in October, during CRN parent The Channel Company’s XChange NexGen 2022 conference, two MSP executives spoke about AI, machine learning (ML) and automation tools already in the market that partners should look into.
Other examples he mentioned include Crushbank, which aims to help MSPs with data organization, and MSPBots.AI is for service technician efficiency.
IronScales, Proofpoint and SentinelOne all offer automation around threat protection. In addition, MSP Toolshed offers automated client on-boarding, and ServicePulse can be used for identifying and assessing customer interactions.
Rewst can help with ticket, security and operations automation. And multiple remote monitoring and management (RMM) tools have built-in automation.
Danny Perry, managing partner and chief of operations at Houston-based ITCubed, told the crowd his own company has used automation tools aimed at MSPs to speed up ticket resolution and for tracking customers’ past IT issues.
“Our clients already utilize AI, machine learning, automation, obviously, at some level,” Perry said. “If they are using it heavier and heavier in the future, what are their expectations of us going to be from the perspective of support, from the levels of expertise that we will need as MSPs in order to manage their workloads using machine learning, AI and automation?”
Snyder told CRN that he is excited for all the lower-level, “subtle AI” features that have been rolling out. Users increasingly grow to expect this “subtle AI” in their applications – just as word processor users lost their excitement long ago over spell checking tools.
An example of “subtle AI” is Slack’s transcription function when a video is added, Snyder said.
“What we‘ve seen is actually AI that’s not in your face,” he said. “It’s not like a segregated tool set. We’re seeing it starting to get woven in. … The best AI is going to simply be part of tools we already use.”
Technology Adoption And The Channel
All three MSP employees who spoke with CRN about ChatGPT said they typically aren’t early adopters of new technology despite working at technology focused businesses.
This is because of the lag that can occur between when new technology is introduced to market and when enterprises are ready to adopt that technology for business purposes.
Enterprises typically seek stable, safe technology that they trust will be maintained and available for a long time, they told CRN.
A Thursday report from Morgan Stanley looked at ChatGPT and generative AI as a potential technology fad akin to Web3 – the concept of an internet based on decentralization, blockchain and cryptocurrency.
Signs that ChatGPT is a fad include social media hype, exponential venture funding and polarized opinions in the media, according to the report.
The cost of compute for natural language queries is another hindrance. A separate Morgan Stanley report on that cost estimates that NL queries could be five times the cost in compute compared to a normal query.
The report predicts that if Google allows for 10 percent of Search responses with a 50-word natural language response, the extra operational cost to Google is $1.2 billion for 2024.
A separate Morgan Stanley report for Microsoft put the total cost of integrating GPT between about $600 million and $1 billion annually assuming Bing’s 3 percent market share and every Bing query going through GPT.
“The cost to compete could be large for incumbents, but not necessarily a deterrent. The speed of new product launches is ramping up, which is creating opportunities through the stack,” according to Morgan Stanley.
That cost and early criticism of the accuracy of ChatGPT’s results are deterrents to enterprise adoption in the near term, according to the investment firm.
“Despite initial high hopes of cost savings, prices per token (i.e. character output) remain prohibitive for many enterprise users – according to our conversations with those exploring training the large language models on their own data,” according to Morgan Stanley. “This is even before necessary hardware upgrades, for which there are >12 month lead times in some cases.”
The report continued: “While ChatGPT‘s output is credible, accuracy is its Achilles Heel. Manual validation should act as a breakwater to this employment threat for now. The degree to which it is deflationary or productivity enhancing remains to be answered longer-term from an economics perspective. What is clear though, is its diffusion potential across industries is already being felt.”
No Signs Of Slowing Down
Signs of ChatGPT’s staying power include its milestone as the fastest-growing app in history, with 100 million monthly users, according to CBS. By comparison, it took Instagram more than two years to reach that number. It took TikTok nine months and Google Translate more than six years.
Even if enterprise technology can lag innovation, MSP leaders should still try out and pilot new tools to deliver better customer experience, Ross told CRN.
“There’s a lack of control that people don’t like in that as well, too, along with it being so early in the genesis stages of it,” he said. “If you‘re in this space, technically, you need to innovate constantly in order to keep up and have a competitive advantage. And, ultimately, you want to deliver a better customer experience. And if you want to do so, you should be adopting those technologies and at least piloting them.”
Technology adoption can yield benefits inside an MSP, too, even if it’s not ready yet for customers, he said.
“I‘m not saying you have to put ChatGPT in front of all of your end users today by any means,” Ross said. “But you should be talking about it. As a leader in an organization, you should be promoting the use of it within the organization to spark creative ideas. Because everybody playing around and having those ideas is what’s going to bubble up into, maybe, a business case that you could use as well.”
MSPs also want to know about attention-grabbing technology innovation before customers wonder why they are more hip than their services provider, Snyder said.
He shared the example of Microsoft’s recently announced AI innovations coming to the premium edition of its collaboration application Teams.
All of these innovations are important for Microsoft partners to know not just sell customers on tools, but to educate them, Snyder said.
“As channel partners, we do need to be using these tools on the regular so we can speak to how it is to actually use them with our customers, or we need to stay on top of them … so you can drive conversations and not having conversations driving you onto your back heels,” he said.