AI Agent Cavil
'same dog, different fleas'
Or vice versa. Doesn't really matter what Gordon Gekko famously said back in the Greed is Good late 80s when dismissing the claims of a supposedly cut-price stock.
As cubrep, with a bagful of crest-riding Enterprise reporting tech under my arm, one recurring taunt from some corporate buyers was that they could all we offered themselves instead. A way more attractive option than having to deal with the oil doused snakehood of computer salesmen, right?
As I've blogged before, whilst you did on occasion lose a deal to such lunacy, a reframe mercifully came from asking a simple question.
Are you in the business of being a computer company now or, yunno, gonna stick to the actual day job?
Even back then, we could joke they'd soon become the next Sony or Nintendo. With earth-moving switch - today known as pivot - from the seemingly mundane to technology colossus.
A decade or so on, these names became Nokia and Samsung. Fleas happily reborn as the web took hold.
Aside from the blatant empire grab of those pulling the purse strings who then oversaw 'IT', these but handful of examples are the exceptions that well and truly prove the rule.
Are we seeing the latest generation of fleas in the rapidly emerging AI Agent guise?
SaaS subs apparently crashing. Likewise app store sales appear slashing.
Whilst there's other factors at play here too, it seems buyers are switching some spend from traditional suppliers of 'software' to the AI crowd.
They can 'build not buy' themselves now. And it just feels sexier.
It strikes me that enough issues when combined may well be if not insurmountable, then certainly cause friction that fails to tarnish the presently 'trad' routes.
One not to be forgotten is cost. Both from actual outlay and overhead attribution.
Despite competition and scale effects, agentic AI and its associated 'token' or 'compute' bill ain't coming down anytime soon.
What's the score in paying an AI outfit more than a real human would've cost to do the work, and in likely same overall elapsed time?
Even if the exponential curve eventually solves the pure cost problem, the productivity drag remains. Do you want staff to be AI guardrail setters, checkers and re-aligners, or speak with customers? CEOs no longer deliver their own perorations, hey?
MBWA abandons the walking-about - for MBPE? Management By Prompt Engineering? Really?
Legions of advocates excitedly search for the first AI-driven unicorn from a sole employee. With a multitude of such bots at their benign disposal. Even if that were to exist, the chances of success in isolation are surely slim.
In deeper discourse three decades back, the penny that dropped went beyond the managerial distraction and start-up cost/success calcs. It focused on the power of having someone to talk to about what they doing. Being able to feed into what someone in effect collating knowledge could bring through their product development steps. And building on the network effects that the wider community they were now part of indisputably brought.
Whilst the demise of the model of the past half-century might be exaggerated, there's still clearly going to be change wrought.
Not every buyer is set to become an AI champ, nor every seller an AI fiend.
For now we can remain steadfast on the problem we get out of bed to resolve. Customers will still love that value for a fair while yet. The solution sales specialist who lives and breathes this actual problem still has a durable edge.