Organisations do not usually lack ideas for AI. They lack a way to kill the ones that should not be built. Strategy, discovery and architecture are that filter: where AI belongs, what data and ownership would make it possible, and whether the answer is a Linear Horizon product, a bespoke system, or nothing.
The people doing that work should still be involved if you later build. Otherwise the presentation and the system diverge.
Linear Horizon advises from the same practice that trains and ships. You can enter here and stop. You can continue into Build or a product deployment. The case is to know what you are buying before the demo becomes a commitment.