Most AI platforms give you the power to automate almost anything. Then they hand you a blank configuration screen.
You know what you want: “Every Monday, pull last week’s client meetings, summarize the action items, and send each team member their list.” It’s simple to say. Turning that sentence into a working workflow usually means triggers, connectors, field mappings, conditional logic and a growing list of settings you don’t fully understand.
Somewhere between the good idea and the working workflow, most people give up. The idea wasn’t bad, and the technology could have done it. The path between the two was built for engineers, not for the people who actually know how the business runs.
The friction nobody budgets for
AI projects are failing at a remarkable rate. MIT’s 2025 State of AI in Business report found that 95% of organizations were getting zero measurable return from their generative AI investments (MIT NANDA report; Forbes coverage). The usual explanations are technical: the model wasn’t accurate enough, the integration broke, the data wasn’t ready.
Those things happen. But in our experience, one of the most common failure points is far more ordinary. It’s the gap between the person who understands the work and the tool that could automate it.
The managing partner, operations lead or practice manager knows exactly what needs to happen. The platform can technically do it. Connecting the two requires technical translation, which is slow, expensive and usually depends on someone who’s already busy. So a few workflows get built by whoever has the technical skills, everyone else goes back to the old way, and six months later leadership wonders why the AI investment never took off.
The problem wasn’t capability. It was friction.
Configuration screens are the wrong interface for most of your team
Configuration screens make sense for technical teams. They’re precise and powerful. But they ask accountants, attorneys, consultants, advisors and operations staff to think like software developers: break the work into triggers and actions, understand data structures, and debug when something doesn’t connect.
That’s not how people describe their work. They think in outcomes and steps: “When this happens, I want that to happen, and here’s what matters.” They explain it the way they’d explain it to a capable new hire.
So what if you could build a workflow the same way?
Building workflows through conversation
That’s the idea behind conversation-driven workflow creation. Instead of configuring a tool, you describe what you want in plain language. The AI asks clarifying questions, proposes how the workflow should run and builds it with you, step by step.
In practice, it looks less like filling out a settings page and more like delegating to a sharp assistant:
- You explain the goal in your own words.
- The AI asks what it needs to know: where the information lives, who should get the result, what to do if something is missing.
- It shows you its plan, so you can correct it before anything runs.
- It builds the workflow, runs it, and lets you refine it by saying what you’d like changed.
This is how we designed Sofia, the AI platform we build at FINdustries. It’s how we run our own firm, too: our daily task summaries and sales pipeline check-ins run on Sofia every working day.
Why this removes adoption friction
- The expert becomes the builder. The person who understands the work creates the workflow directly, without waiting on a technical translator.
- Ideas turn into results faster. When going from idea to working workflow is a conversation instead of a project, far more ideas actually get tried.
- Iteration is natural. Improving a workflow is as simple as saying what should change, so workflows get better over time instead of being “set and forget.”
- More people participate. AI stops being the domain of one technical champion and becomes something the whole team uses.
That last point matters most. AI adoption succeeds when it spreads across a team, not when it’s concentrated in one person.
Conversation doesn’t mean less control
A fair concern: if anyone can build a workflow by talking, is there less oversight? In a well-designed system, there isn’t.
At FINdustries, we believe control comes before delegation. A conversational system should show its plan before it acts, ask for confirmation on consequential steps, and keep a clear record of what each workflow does and why. The conversation isn’t a shortcut around governance. It’s a more human way to set it up.
In fact, conversation can make control easier. When the AI explains its plan in plain language, you actually understand what you’re approving, which is more than most people can say about a complex configuration screen.
Questions to ask when evaluating an AI platform
- Who on our team could build a new workflow on their own today?
- How long does it take to go from an idea to a working workflow?
- Can we describe what we want in plain language, or do we need technical configuration?
- Can we see and approve what the AI plans to do before it runs?
- How easy is it to change a workflow once it’s built?
The answers will tell you more about whether AI will actually be adopted than any feature comparison.
The bottom line
The biggest barrier to AI adoption often isn’t what the technology can do. It’s how hard it is for the people who know the business to put it to work. Blank configuration screens keep that barrier high. Conversation brings it down.
Want to see a workflow built in a conversation? Talk to FINdustries about Sofia.
Sources
- MIT NANDA, The GenAI Divide: State of AI in Business 2025
- Forbes, “MIT Finds 95% Of GenAI Pilots Fail Because Companies Avoid Friction” (Aug 26, 2025)
