Appian's Jake Rank on serious AI: put agents inside the process, measure the outcome, and use a business rule when it's cheaper

Résumé détaillé

Appian brought its APJ roadshow to Sydney's Darling Harbour under the tagline "Serious AI, built in process", with afternoon keynotes and a morning set aside for the technical...

Alex Zaharov-Reutt, Global AI and Technology Editor | Published 12 Nov 2026

Appian brought its APJ roadshow to Sydney's Darling Harbour under the tagline "Serious AI, built in process", with afternoon keynotes and a morning set aside for the technical sessions (a sensible running order I'd like more vendors to copy).

Between sessions I sat down with Appian vice president of product management, Jake Rank, for my Alex on Tech and iTWireTV video interview series. Rank leads product across AI, document processing, RPA and process mining, and demonstrated the lot on stage.

His message was consistent on camera and off: AI earns its keep inside a measured business process, and sometimes the smartest move is to skip AI altogether.

Our video interview is below, after which this article continues, please watch, and read on!

"Appian's helping our customers automate their most mission-critical work with AI and other technologies," Rank said. "Whether it's financial services or insurance or government, we're helping them take their mission-critical work and make it faster and more efficient."

He splits AI into two jobs. "One is to accelerate and improve business outcomes. That's things like automating the document processing or using agents as part of a flow to maybe automate some human tasks," Rank said. "But we also use AI to augment development, so you can use AI to deliver with Appian faster."

That framing came straight from the main stage. Appian founder and CEO, Matt Calkins, whom I've interviewed before, picked up the opening video's light bulb story: invented in 1879, it took 14 more years to light up the Chicago World's Fair.

"How do you use AI for critical work is the business question of 2026," Calkins said. His answer: "AI is probabilistic, and therefore it needs a deterministic layer of process in order to keep it aligned with the mission."

Début de l'article, extrait de iTWire. Pour lire la suite, ouvrez l'article chez le média.