AI tools can speed up individual tasks while leaving the surrounding business process untouched. In a SiliconANGLE opinion article, SnapLogic chief technology officer Jeremiah Stone argues that the gap between faster work and measurable business gains is where many companies are getting stuck.
Watch Desk analysis
What happened
Stone uses Amdahl’s Law to explain why a faster step does not necessarily make an entire process faster. An AI assistant may draft a proposal in seconds, he writes, but reviews, pricing and approvals can still take days. A coding tool can produce software quickly while testing, security and deployment remain the bottleneck.
The article says Forrester found that only 15% of AI decision-makers reported an increase in earnings before interest, taxes, depreciation and amortisation over the previous 12 months. Stone recommends tying AI projects to a specific business outcome, establishing a baseline and assigning a business leader responsibility for results. He suggests companies should expect firm return-on-investment data within about six months, while a process redesign may take 90 to 120 days to design and implement.
Why it matters
The practical advice is to examine the whole workflow before automating one piece of it. Stone says companies should look for bottlenecks, hand-offs, approvals, exceptions and rework, then judge projects by measures such as cycle time, quality, cost per outcome and financial impact, rather than prompts, seats or impressive demos.
That is a more demanding test than asking whether an AI tool can do a task faster. It also makes the awkward but useful possibility explicit: sometimes the right redesign is to remove a step, not give it a chatbot.
Our read
Stone’s argument is opinion from a technology vendor’s executive, not proof that every AI project needs the same timetable or will fail without a process overhaul. Still, the core test is sound: name the outcome, measure the starting point and check whether the whole operation improves. Before adding another pilot, teams should be able to say what success would look like and who owns it.
What to watch
- Whether organisations measure business outcomes rather than tool usage.
- Whether pilots include integration, security and change management from the start.
- Whether companies publish results against clear baselines, including costs per outcome.
Discussion spark: Should companies redesign whole workflows before rolling out AI, or is it better to start with small task-level pilots and prove value first?
Sources and evidence
- How to avoid AI's 'valley of despair' – SiliconANGLE (23 October 2026, 14:00 UTC)
- Source update (23 October 2026, 14:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.