Enterprise hits and misses - is it a bad time to buy enterprise software? And is generative AI "instant mediocrity" for the enterprise?

Lead story -is it a bad time to buy enterprise software?

Brian thinks so - and he's issued a cautionary scorcher to enterprise buyers everywhere: It's a bad time to buy software - here's why! The crux of Brian's argument: AI/ML represents a sea change. Most software vendors aren't ready:

Right now, this isn’t a software selection, it’s gambling. Software buyers might be taking big technical and economic risks when they buy new application software today. The question is how to avoid obsolescence risk and a bunch of duplicate costs?

Brian sees big changes afoot:

Software firms are trying to figure out where it makes sense to use large language models (LLMs), Generative AI and other advanced technologies. Some of these capabilities can replace entire applications or material portions of an existing application.
Until they figure the previous point out, vendors have no idea how they’ll price these new capabilities.

Moreover, whatever you license or subscribe to now may not count towards the use of an enhanced or replacement solution. Do you want to pay twice for the same functionality?

Is Brian literally saying, "Don't buy any enterprise software"? No. But he's wary of big purchases - and he urges customers to ask the tough questions:

Put the existing vendor on notice – Let them know that: You are concerned with all the ambiguity re: advanced technologies, privacy, roadmaps, pricing and security.

I (mostly) agree with Brian, though I would extend his argument to implementation partners as well. If now is the time to re-evaluate software vendors, the same certainly applies to systems integrators.

I'll add these:

If you are one of those somewhat-rare firms with a deep internal data science and AI development team, your situation may be different. You may be more focused on accessing your software data without hassle than relying on vendors for AI innovation.

When it comes to AI, I believe you can more safely purchase software from firms that demonstrate the ability to play nice with others - including external AI models - and provide well-supported APIs and pre-built integrations.

These questions are also critical:

What is your AI pricing model? Do I have access to it in the core release, or do I need to upgrade (e.g. to your SaaS/cloud release).

Some vendors hope to "delight"/surprise customers by including all AI innovations in the existing software. Others will make it available via additional licenses, or some combination thereof. Doing a major upgrade is more viable IF your vendor of choice plans to embed AI in its core release, without additional licensing.

In the long run, vendors won't able to get away with charging for AI-enhanced software or additional AI releases (specialized AI apps are another matter). But in the 1-3 year time frame, the licensing issues Brian flags are a big cost factor. Vendors probably don't have this nailed down yet, so pricing assurances must be secured.

We can debate how much of a revolution generative AI will prove to be, but there is no debate that this type of AI is currently expensive to run. Operational costs may go down with further innovation, but for now, computing cost is one of the under-discussed factors in adding generative AI across workflows. Generative AI software licenses are likely to reflect this.   

Diginomica picks - my top stories on diginomica this week

• Why data management matters in meeting ESG targets - some organizational tips - Cath with timely advice on making ESG count: "It certainly seems there is a long way to go at every level in getting ESG data management right – despite the pivotal role data plays in enabling organizations to hit their ESG goals."
• Is Rust the strong foundational code CIOs need? - Not the question I was expecting, but Mark Chillingworth has an interesting answer: "Future technology developments will need to help the organization be more environmentally sustainable. Harris at Red Badger says Rust is one of the first languages he's worked with that offers both - at the same price."

Vendor analysis, diginomica style. Here's mya three top choices from our vendor coverage:

• Adobe Summit EMEA - with Firefly for enterprise and Express upgrade, Gen AI makes everyone a content creator - Phil reports back on one of the most well-thought enterprise vendors when it comes to generative AI announcements: "This is a bold move by Adobe, which might have been expected to aim to protect its existing Creative Cloud products by resisting the urge to make content creation more universally accessible."
• Celonis aims to make process mining more intuitive, open and intelligent - Derek on Celonis: "It’s worth highlighting that these strategic levers are being fueled by Celonis’ recent investments in what it is calling an ‘expanded data foundation’ - where it is shifting focus to object centric process mining."
• Bringing enterprise clout to generative AI - C3.ai CEO Tom Siebel on what makes a difference - Stuart on a firm with a familiar enterprise name at the top: "As the consumer-buzz over ChatGPT shifts over to the enterprise market, the tick list of competitive differentiators that he’s claiming for his firm’s generative AI offering provides a compelling menu for adoption, if they can be delivered ‘on the ground."

Salesforce Connections 2023 coverage - Stuart provided the virtual coverage; I was on the ground in Chicago. The show's generative AI themes obviously took center stage - with more coverage to follow, via Salesforce's New York City "AI day" today. AI may be the messaging priority - understandably so - but we focused on the customer views/reactions.

A diginomica use case selection:

Jon's grab bag - Cath revives the dashboard in Dashboards aren’t dead! How charities and health organizations use data visualization to make key decisions. (Though in my view, dashboards were never dead - we just need to understand their limitations. Dashboards don't make better decisions on their own accord; they work best in conjunction with an alerts-based infrastructure). Finally, Chris grapples with a fresh crypto report like only he can in UK crypto report - we need regulation, but don’t ask us for details.

Best of the enterprise web

My top six

• The impact of generative AI on software team productivity is... complicated - Joe McKendrick's latest has me asking one question: c'mon Joe, why do you have to be such a buzzkill sometimes? Can't we just enjoy the prospect of auto-magical productivity gains for a singly Scooby-Doo moment? As McKendrick writes: "Despite the potential for vast productivity gains from generative AI tools such as ChatGPT or GitHub Copilot, will technology professionals' jobs actually grow more complicated?"
• Instant Mediocrity: a Business Guide to ChatGPT in the Enterprise - An essential enterprise AI piece comes by way of Amalgam Insight's Hyoun Park. His twist on "instant mediocrity" is not as negative as some might think. As Park writes: "The truth is that instant mediocrity is often a useful level of skill. If one is trying to answer a question that has one of three or four answers, a technology that is mediocre at that skill will probably give you the right answer... If you want to remember all of the standard marketing tools used in a business, a mediocre answer is just fine. As long as you don’t need inspired answers, mediocrity can provide a lot of value." I objected to Park's generative-AI-as-intern analogy. I've had interns that could run circles around ChatGPT; a good intern would be able to prompt engineer ChatGPT in no time, and be far better than both alone.
• Hybrid Workforce Management: Navigating the Complexities of a Diverse Workforce in the Modern Era - Hyoun Park with another meaty post for us to ponder: "When 40% of labor consists of either part-time, contractors, or on-demand workers, a workforce management solution that only looks at full-time payroll, onboarding, time, attendance, and benefits is no longer sufficient."
• Driving An Octopus - Lora Cecere on her valiant quest to redefine supply chain planning: "My inbox is full of articles on probabilistic planning to improve safety stock. My response is a big YAWN."
• Two models of AI oversight - and how things could go deeply wrong - Gary Marcus on the AI hearing he testified in at the US Senate - and the divergent AI scenarios we are facing (one good, one not-so-lovely).

• Apple Vision - Stratechery's Ben Thompson is too smart and thoughtful for me to tag him as an Apple fanboy, but I did struggle to separate a whiff of fanboyism from his Apple Vision review. Thompson does make an important point about Apple Vision being much more Augmented Reality than Meta's full VR immersion headsets, but as I said on Twitter:

Virtual reality may struggle beyond sizable niche audiences, but mainstream augmented reality glasses feel inevitable to me. But not quite yet:

Whiffs

So Microsoft's 'Clippy' Designer Says He Was 'so Embarrassed' by Creation. Why be embarrassed? A super-annoying, not-smart-enough virtual assistant was clearly ahead of its time. At least Clippy tried to be cute while (uselessly) disrupting your work. Oh, and as for this:

Which is why pieces like this totally miss the point: Why greedy, lazy transactional lawyers should be very scared of GPT-4. Getting to 90 or 95 percent accuracy is plenty good enough for many AI use cases, but it's not good enough for going to trial, unless you want to join the shame-and-spank tunnel in front of disapproving judges. The cool part - and I think there is a cool part - is extending reasonably decent legal know-how to those who have no attorney except the Internet.

Oh, and this:

Maybe someone can FedEx a copy of that playbook along, before the next event. Plane is boarding, let's stop here.

If you find an #ensw piece that qualifies for hits and misses - in a good or bad way - let me know in the comments as Clive (almost) always does. Most Enterprise hits and misses articles are selected from my curated @jonerpnewsfeed.