AI is changing implementation speed. It is not changing the first question a good technology conversation should ask: what is the business actually trying to be good at?
That sounds simple, but it is often skipped. A business sees a new AI tool, a vendor demo, or a workflow automation promise and jumps straight into the platform conversation. That is backwards. Before deciding what to automate, you need to understand how the organisation works, where the judgement lives, which workflows matter, and what creates advantage.
AI accelerates the work after the thinking is done
The useful promise of AI is not that it magically understands your business. It is that once the business has been understood, AI can help implement improvements faster. It can draft, compare, summarise, classify, generate code, assist with support, move information between systems, and help people make better decisions with less manual effort.
That is valuable. But it is only valuable when pointed at the right work. If the workflow is poorly understood, AI just accelerates confusion. If the data is messy, AI exposes the mess faster. If ownership is unclear, agentic systems can make accountability worse. Security-led AI adoption starts by understanding the workflow before automating it.
SaaS was great, but it came with a trade-off
SaaS platforms solved a lot of problems. They gave businesses mature software without needing to build everything from scratch. For non-core work, that was usually the right answer. If the process was commodity, conforming to the SaaS workflow was fine.
The trouble started when the workflow was closer to the heart of the business. You either conformed to the platform and lost some of what made the business different, or you customised and hacked at the platform until it sort of matched the workflow. That could work for a while, but it often created cost, fragility, and systems that were hard to secure or improve.
The real question is what should be standard
AI makes this conversation more important, not less. Some work should stay in SaaS. Some work should use standard AI tools. Finance, HR, ticketing, document management, and other commodity functions often benefit from boring, well-governed platforms.
But the workflows that define how the business competes deserve more attention. That might be how you price work, assess risk, design a service, handle customer context, manage operational knowledge, or make decisions that depend on experience. Those are the places where AI can be used to harness what the organisation already knows and accelerate it.
You still need someone to find the advantage
The hard part is not choosing a model. The hard part is identifying the core competency the organisation possesses and deciding how technology should support it. That requires business analysis, workflow design, security thinking, and enough commercial judgement to know where custom work is worth it.
Once that is clear, AI becomes practical. It can help turn a better workflow into a working process. It can help build the internal tools, knowledge flows, integrations, and agentic workloads that support the advantage. It can also help reduce the cost of implementation, because the distance between design and working system is shorter than it used to be.
Keep commodity work commodity
None of this means every business needs bespoke AI everywhere. That is the wrong lesson. Commodity work should remain simple. Use SaaS where the workflow does not differentiate you. Use standard AI tools where the risk is low and the task is common. Keep the operating model clean.
Then spend the design effort on the workflows that matter. Secure them. Measure them. Improve them. Use AI to help people execute them faster and with better context.
The practical path
Start by mapping the business, not the tools. Separate commodity workflows from advantage workflows. Understand the data, decisions, risks, and handoffs. Decide where SaaS is enough, where standard AI is enough, and where a more deliberate AI-enabled workflow will create value.
AI does not remove the need for that conversation. It rewards businesses that have it properly. The organisations that know what makes them special will be able to use AI to accelerate that advantage. The organisations that skip the thinking will mostly accelerate noise.