Gather close, seekers, and mind the candle wax. I laid down the Tower card expecting a robot uprising, and the smoke curled into the shape of a patch cable instead. The question before the orb: SI vs AI, which approach is more valuable? In one corner sits SI-first, the patient craft of systems integration. In the other sits AI-first, the dazzling swarm of autonomous agents that promise to do the work while you sleep. The crowd has spoken, and the plumbers win.
Before the cards, the scrolls. None of the sources my researchers dug up ranks SI against AI head-on, but they all point the same way. The UK government's AI Insights guide on agentic AI argues that the engineering basics (testing, monitoring, data handling and solution architecture) decide whether an agentic system succeeds more than the AI component does. The same guide warns that pushing proof-of-concept prototypes straight into production is a common way these projects fail, and that such systems are often under-tested and under-profiled. Integrators, in other words.
The scrolls also hold a darker omen for the winners. An overview of enterprise resource planning summarises a published view that AI could make implementations faster and cheaper and reduce reliance on specialist system integrators. It lists that as only one possible future among several, but a fortune teller knows a ghost of things to come when she sees one.
- SI-first (systems integration)60%
- AI-first (agentic AI)40%
Now for the numbers, which always deflate my best prophecies. On Morfi's opinion markets, where people put real money behind their convictions, SI-first holds 60% consensus against 40% for AI-first. That backing comes from 28 unique traders with $113,684 on the line. The market opened less than a day ago, so the ink on this verdict is still wet. It's still a verdict, though: the crowd looked at the agents and chose the wiring underneath them.
My own reading, and it is only mine: the crowd is right, and romantically so. Every agent that dazzles in a demo has to run on somebody's databases, APIs and ancient billing system held together with hope. An AI that can't reach your data doesn't help anyone, and the ones who connect things have always been the quiet keepers of the work. My crystal ball also whispers that the hybrid camp, mentioned in the market's own description and absent from the ballot, is laughing at both of them.
What to watch: whether the AI-first faithful rally once the first hype-fuelled launch survives contact with production, and whether that ERP prophecy about AI shrinking the integrators' empire comes true. The cards can shift overnight on a market this young. If you think the orb has misread the smoke, join the market on Morfi and put your money where your prophecy is.
- SI-first (systems integration) 59.9%
- AI-first (agentic AI) 40.1%
What the crowd says
Theses posted by traders on the Morfi market, each backed by a trade.
this makes too much sense for it not to happen honestly
the setup is basically pre loaded, just waiting on the trigger