One of the reasons “AI transformation” can feel overwhelming is that it’s often discussed at the wrong altitude — as a company-wide initiative, a strategic imperative, a competitive necessity. All of that may be true. But it doesn’t tell you anything about where to actually start.
At ArcTouch, we’ve been building apps and digital experiences since 2008 — more than 500 products across mobile, web, and emerging platforms. Over the past few years, we’ve woven AI into every phase of that process. What we’ve learned is that AI’s impact isn’t abstract. It shows up in specific places, in specific ways, and it’s most powerful when experienced specialists are the ones directing it.
Here are the four areas where we’ve seen AI make the most consistent, measurable difference — and what it actually changes about the work.
#1. Strategy and planning
Every digital product starts the same way: a lot of questions, not enough answers, and pressure to move faster than the research allows. Market sizing, competitive analysis, user interview synthesis, and requirements documentation; this work is essential, but a significant portion of it is information-processing, not strategic thinking. It’s time-consuming without being the part that requires your best people’s judgment.
AI compresses this phase significantly. In our Discovery Workshops, typically 3-5 intensive days where our team and client stakeholders align on the product opportunity, we used to wrap up with a set of notes and a plan. Now we have validated ideas and working artifacts.
In one recent workshop for a Fortune 100 technology company, we used AI to turn a photograph of a wall of handwritten sticky notes into a first draft of a Product Requirements Document before the session ended. What would have been two or three days of documentation work became a starting point by day three — freeing the team to pressure-test ideas instead of transcribing and synthesizing them. By the end of the workshop, we had a functional prototype that we tested with real users. The client approved the full build before anyone left the room.
#2. Design and user experience
Design work has always had two distinct components: the creative work of deciding what to build and how it should feel, and the production work of executing and documenting it. Historically, these have been interwoven with production tasks — meticulously preparing assets, annotating specs for engineers, adapting designs across screen sizes — consuming an oversized share of a designer’s day.
AI has created a meaningful shift in where designers now spend their time. AI-powered design tools now handle much of the tedious and time-consuming production work. Designers are liberated to go deeper into the work that requires human judgment: the interaction that feels natural under the user’s thumb, the visual decision that makes an interface feel trustworthy, the animation detail that turns a functional app into a lovable one. ArcTouch designers have always focused here; AI now gives them more room to do it.
We’ve also seen AI improve the handoff between design and engineering. By integrating Figma’s MCP Server with tools like Cursor IDE, our engineers can pull design layouts, color tokens, and type styles directly into their coding environment without switching tools. Fewer misunderstandings. Fewer revision cycles. Faster builds.
#3. Engineering and development
The conversation about AI in engineering usually starts with code generation, and that’s real, but it’s not the most important part of the story.
Yes, AI tools write code faster than a human. Tools like Cursor IDE and Claude Code can generate whole sections of code and identify errors before they compound. In a recent project, our engineers pulled design specifications directly from Figma via an MCP Server integration and implemented them in minutes rather than hours. That kind of acceleration is genuine.
But the more important shift is in what engineers do with the time they recover. They move from writing code to owning it. From generating syntax to thinking about architecture. From reactive debugging to proactive quality decisions. Our engineers were never just coders; they’ve always been technical problem-solvers. AI removes enough of the mechanical work that more of each day looks like the latter.
AI also changes the early stages of a project, where uncertainty is highest. We use short, focused investigations called tech spikes to test assumptions before committing to a full build. In a recent Discovery Workshop, we ran three tech spikes using AI in the time it would have taken to run one manually. The code from a tech spike doesn’t need to be production-ready; it just needs to answer a question or test a hypothesis. AI is well-suited to exactly that.
What doesn’t change: Our engineers own every line of code, AI-generated or not. Validating output, catching bad patterns, maintaining security and performance standards — those responsibilities stay with the human. AI-generated code that ships without expert review is one of the clearest ways AI introduces risk instead of reducing it. We’ve held this position from the start, and it’s reflected in how we train and work.
#4. Testing and quality assurance
QA is where AI’s impact is both significant and easy to underestimate.
Test scenario generation is the clearest example. For a recent client project, our QA team used AI to generate more than 50 test scenarios directly from the codebase in under five minutes. The same work done manually would have taken the better part of a day. More importantly, AI catches edge cases that a human working under time pressure might miss, not because the human lacks skill, but because there are too many combinations to track manually.
For more sophisticated testing, AI makes things feasible that weren’t practical before. For a communications firm’s website project, we used AI to generate a script that compared two versions of a 35-page site for visual differences. The full check took 20 minutes. Done manually, it would have been a full day. It immediately surfaced six errors that could be fixed before they reached users.
The QA professional’s role shifts in the same direction as every other role: from executing repetitive tasks to designing the strategy around them. Less time running tests, more time thinking about coverage, edge cases, and what the user will actually encounter.
#The common thread
Across all four areas, the pattern holds: AI takes on the high-volume, repeatable work, and returns that capacity to the people best positioned to use it on things that require judgment, taste, and experience.
Achieving these gains from AI isn’t automatic. It requires deliberate choices about tools, integration, and quality standards. It also requires specialists who know how to direct AI effectively. A junior team pointing AI tools at a complex problem gets a different result than an experienced team doing the same thing. The tools amplify what’s already there.
At ArcTouch, our experience of building digital products for nearly two decades gives us a baseline for how to use AI more effectively. It’s not a replacement for the expertise we’ve built. The result for our clients is concrete: faster timelines, better-tested applications, more creative design work, and earlier confidence in what they’re building.
#See what AI-powered development looks like in practice
ArcTouch offers AI Advisory Workshops for product, design, and engineering teams — a focused session to identify where AI can make the biggest impact to your specific workflow, and what a practical integration path looks like for your team.
Ready to start your next project and build responsibly with AI from day one? Let’s talk.