The Internal Tools Revolution: Why Every Company Needs a Design Thinker Who Can Vibe Code

Provocative Essay | ~7 min read

Here's a dirty secret of the software industry: the tools companies build for their customers are beautiful. The tools they build for their own teams are garbage.

I've seen this pattern for 15 years across every industry I've consulted in. The customer-facing product gets user research, design systems, performance optimization, and A/B testing. The internal CRM gets a spreadsheet duct-taped to a form. The reporting dashboard was built by an intern in 2021 and nobody's touched it since. The onboarding system is a 47-page Google Doc that no new hire has ever read completely.

It's not that companies don't care about their internal tools. It's that building good ones has always required the same expensive resources as building the product. Engineers, designers, product managers, sprints, roadmaps, prioritization.

And internal tools always lose that prioritization battle.

Until now.

The Vibe Coding Shift

In February 2025, AI researcher Andrej Karpathy coined a term that caught fire: vibe coding. The idea was simple. Describing what you need in conversational language and collaborating with AI to bring it to life. Not traditional programming. Not no-code drag-and-drop either. Something new. A conversation between human intent and machine capability.

The adoption has been staggering. Searches for "vibe coding" jumped 6,700% in 2025 alone (Exploding Topics). According to a Sleek design report, 67% of design teams at mid-to-large companies have integrated AI generation tools into their workflow as of early 2026. Muzli called it "the most significant shift in design practice since the move from print to digital."

But here's what most of the vibe coding conversation misses: everyone's talking about building consumer apps, SaaS products, side projects. Nobody's talking about the most obvious, most impactful application:

Internal tools.

When building a custom dashboard takes 15 minutes instead of 3 sprints, you build the dashboard. When creating an interactive workshop tool takes an hour instead of hiring a vendor, you build the tool. When generating a client documentation portal takes 15 minutes instead of 15 days, you build the portal.

That changes the math on organizational effectiveness completely.

Why Design Thinkers, Not Just Developers

Here's where I'll make a claim that might sound counterintuitive:

If AI does the coding, the person directing the AI shouldn't be a developer. It should be a design thinker.

I can already hear the objections. Developers understand code, they can review output, they can debug issues. All true. But the more important skills are different.

The hard part was never the code

Building an internal tool was always 20% coding and 80% everything else: understanding the problem, talking to stakeholders, figuring out what data matters, designing an interface people will actually use, and iterating based on real feedback.

AI just automated the 20%. The 80% still requires a human. Specifically, a human with facilitation skills, empathy, and systems thinking.

Design thinkers ask better questions

A developer handed a CSV thinks about data types, schemas, and efficient rendering. A design thinker handed the same CSV thinks about: Who looks at this data? What decisions does it inform? What's the story it needs to tell? What action should someone take after seeing it?

The quality of the output depends on the quality of the input. The quality of the input comes from asking the right questions.

Internal tools are service design problems

Every internal tool is a touchpoint in an internal service. The onboarding dashboard is part of the employee experience. The client reporting tool is part of the consulting delivery service. The knowledge base is part of the organizational learning system.

Design thinkers see these connections. A dashboard isn't just a screen. It's a moment in a workflow that either enables or frustrates the person using it.

A confession

I didn't always think this way. For years I was the designer who handed off specs and moved on. I knew what the tool should look like but couldn't build it. Then I started vibe coding. First tentatively, then obsessively. Something shifted.

The first internal tool I built with AI was a workshop polling system. I needed something like Mentimeter but customized for my facilitation method. I described what I needed. An hour later, it existed.

I stared at it. Functional, interactive, deployed. In one hour.

I'd been waiting for "a developer to pick it up" for months.

The bottleneck in my practice was never engineering capacity. It was the artificial separation between the person who understands the problem and the person who builds the solution. AI erased that boundary. It changed everything about how I work.

What "Vibe Coding" Actually Looks Like

Let me demystify this. A real scenario:

The problem: Your consulting team tracks projects in a spreadsheet. Every Monday, someone spends 30 minutes manually creating a status report for leadership. It's always slightly out of date by the time it's presented.

The vibe coding approach:

You open an AI coding tool. You upload the spreadsheet. You say:

"I need a project dashboard for our consulting team. It should show all active projects with their status (on track, at risk, blocked), timeline progress, assigned team members, and next milestones. Leadership needs to see a summary view with KPIs at the top. Total active projects, percentage on track, revenue at risk. Team leads need to filter by their projects. Make it look professional and clean."

Within 10-15 minutes, you have a working dashboard. You look at it, notice the status colors aren't intuitive, ask for a change. You realize you also want a Gantt-style timeline view. You ask for it. You notice the mobile layout is off. You mention it.

Three rounds of conversation. Maybe 30 minutes total. You have a tool that replaces a manual weekly process, gives leadership real-time visibility, and looks good enough to present to clients.

What skills did this require?

• Understanding the workflow problem (people spending time on manual reports)
• Knowing who the users are (leadership wants summary, team leads want detail)
• Having opinions about information hierarchy (KPIs first, then detail)
• Being comfortable with iteration (first output isn't final)
• Communication clarity (describing what you need in plain language)

Notice what's not on the list: JavaScript, React, CSS, database queries, API design.

The technical skills have been absorbed by the AI. The human skills (empathy, communication, systems thinking, facilitation) are what drive the quality of the outcome.

The Organizational Opportunity

Scale this up.

Imagine a mid-size company (50-200 people) where 5-10 people across departments are comfortable with vibe coding. Not full-time. Just people who know how to describe what they need and refine the output.

Suddenly:

• Sales ops builds a pipeline visualizer that updates in real time instead of waiting for the BI team's quarterly dashboard refresh
• HR creates an interactive onboarding experience instead of the 47-page Google Doc
• Product builds a customer feedback analyzer that categorizes and prioritizes input instead of manually tagging tickets
• Finance generates investor-ready reports from raw accounting data instead of spending two days in Excel before every board meeting
• Knowledge management goes from "everything's in Notion somewhere" to a connected, searchable, AI-powered knowledge base

Each of these is a real problem I've encountered in client organizations. Each went unsolved for months or years because they never won the priority battle against customer-facing features.

With vibe coding, they don't need to win that battle. They can be solved locally, by the people closest to the problem, in hours instead of sprints.

Why This Is Different from No-Code

I can hear the objection: "Isn't this just no-code tools rebranded?"

No. The difference matters.

No-code tools give you a canvas with predefined components. You can build within their constraints. When you need something outside those constraints (a custom visualization, a specific data transformation, an unusual interaction pattern) you hit a wall.

AI-powered creation has no canvas. No predefined components. You describe what you need, and the AI generates custom code. The ceiling isn't the tool's feature set. It's your ability to describe what you want.

This means:

• No vendor lock-in. The output is standard code (React, HTML, Python).
• No feature limitations. If you can describe it, it can be built.
• No template constraints. Every output is custom to your specific need.
• Full ownership. You own the code. Modify it, host it, extend it.

The trade-off: you need to be more articulate about what you want. No-code gives you guardrails. AI-powered creation gives you a blank canvas and a very capable collaborator. Design thinkers, trained to navigate ambiguity and articulate vision, are positioned well for this.

The Skills Stack for This New World

If you're a designer, product person, or business operator who wants to be relevant in this reality, here's the skills stack that matters:

Foundation: Design Thinking

• Problem framing (what are we actually solving?)
• User empathy (who uses this and what do they need?)
• Systems thinking (how does this connect to everything else?)
• Facilitation (how do we align stakeholders on what to build?)

New Literacy: AI Collaboration

• Prompt craft (clearly articulating what you need with the right context)
• Iterative refinement (steering toward better through conversation)
• Output evaluation (recognizing when generated solutions are good, mediocre, or subtly wrong)
• Context management (giving the AI the right background to produce relevant results)

Amplifier: Visual Communication

• Information hierarchy (what should people see first?)
• Data visualization principles (which chart tells this story?)
• Interaction design patterns (how should this flow?)
• Accessibility awareness (does this work for everyone?)

Connector: Domain Knowledge

• Understanding the business context and workflows
• Speaking the language of the stakeholders
• Recognizing which tools will actually get adopted vs. ignored
• Knowing when "good enough now" beats "perfect in three months"

This isn't a traditional design education. It isn't a traditional engineering education either. It's a builder's skill set for the AI age.

Getting Started: A Practical Path

If you're curious about vibe coding, here's a path that doesn't require bootcamps, certifications, or career changes:

Week 1: Try one thing. Pick a spreadsheet your team uses. Open Claude, v0, or Bolt. Describe a better version of what that spreadsheet is trying to do. See what happens.

Week 2: Solve a real problem. Identify a manual process on your team. Something that takes someone an hour a week and is boring. Build a tool to replace it.

Week 3: Share and iterate. Show your team what you built. Get feedback. Refine. Notice how the conversation changes when people can see a working prototype instead of discussing an abstract idea.

Week 4: Think bigger. What if your team had a custom dashboard for every important metric? What if every manual process had a tool? What if your internal knowledge was actually searchable and connected? Start mapping the opportunities.

One spreadsheet. One afternoon. One open mind.

The Revolution Is Quiet

Every company I work with has the same latent demand: better internal tools, faster reporting, connected knowledge, custom workflows. The demand has existed for years. What's changed is the supply side.

Building custom internal tools has gone from a major engineering investment to a conversation. The people best positioned to direct that conversation aren't the ones who know how to write code. They're the ones who know how to understand problems, design solutions, and communicate clearly.

If you're a design thinker, a facilitator, a systems thinker, a product person: the tools have caught up to your skills.

The internal tools revolution won't be announced. It'll just happen, one painful spreadsheet at a time, built by the people who were tired of waiting.

What's Next

This is the second in a series exploring AI-powered creation for organizations:

• Previous: From Spreadsheet to Dashboard in 10 Minutes (coming soon) — The data visualization opportunity
• Next: The Knowledge Engine (coming soon) — How AI transforms scattered files into collaborative intelligence

And if you want to develop these skills in community: Design Answers is where we're building this practice together. Real files, real tools, real problems. A mentorship cohort for people who want to go deep.

Details coming soon.

Vitali Gusatinsky is a design consultant and builder with 15+ years of experience specializing in AI-powered internal tools, dashboards, and knowledge systems for B2B organizations. Creator of VitaliOS and founder of Design Answers.