Your Design Team Deserves Better Than Blind Prompting. Here’s the Dashboard That Fixes It.

Why prompt literacy — not prompt enthusiasm — is the skill gap quietly draining your AI ROI.

I’ve spent 20 years designing enterprise software at HP, Boeing, Software AG, and Dell. I’ve led design teams across time zones, shipped products to millions, and received 8 USPTO patents. But nothing in those two decades prepared me for the conversation I kept having in 2024 and 2025:

“Our designers are using ChatGPT every day. We’re innovating!”

No. You’re not. You’re spending.

Here’s what I actually saw leading the AI Experience Team at Dell Technologies: designers copy-pasting vague prompts, getting mediocre outputs, re-prompting four times, burning tokens, and calling it “AI-assisted design.” Leadership had no visibility into who was actually getting good at this, who was wasting budget, or whether any of it was making the work better.

The enthusiasm was there. The literacy wasn’t.

So I built the thing I wished existed.

The Problem Nobody Wants to Name

Let’s be direct. Most design teams today have an AI adoption strategy that boils down to: “Everyone has access. Go explore.”

That’s not a strategy. That’s a hope.

Here’s what happens without structure:

• Designers prompt without frameworks. They type what feels natural, which is usually vague, context-free, and produces outputs that need heavy editing. The AI becomes a slow, expensive brainstorming partner instead of an accelerator.
• Leaders have zero visibility. How many tokens did the team use this quarter? Which frameworks are being adopted? Is the junior designer who uses AI the most also the most effective — or just the most wasteful? Nobody knows.
• There’s no learning path. A designer discovers CO-STAR in a LinkedIn post, tries it once, doesn’t see magic, and goes back to freeform prompting. There’s no structured progression, no practice environment, no way to build skill over time.
• The ROI conversation is impossible. When the CFO asks “What are we getting for our AI spend?”, the design leader has feelings, not data.

This is the prompt literacy gap. And it’s expensive.

What I Built — and Why It Has Two Doors

I created the Design Intelligence Portal — not as a tutorial site, but as a working command center with two distinct views for two distinct jobs-to-be-done.

The Designer View

When a designer logs in, they see tools for getting better at prompting:

• Prompt Generator — A structured builder that walks you through frameworks like CO-STAR (Context, Objective, Style, Tone, Audience, Response), RTF (Role-Task-Format), and Chain-of-Thought. Not theory. A tool you use while working.
• Framework Library — Deep-dive reference on each framework: when to use it, when to avoid it, what makes the difference between a 60% prompt and a 90% prompt.
• Prompt Gallery — Real examples across UX domains: research synthesis, design briefs, stakeholder communications, portfolio writing. See what “good” looks like before you try it yourself.
• Vibe Search — Browse and borrow best practices, prompts, and frameworks from across the community. Think of it as a pattern library for AI craft.
• Learning Path — Personalized progression. Not “watch a 4-hour course.” Structured, modular, bite-sized upskilling tied to the frameworks you’re actually using.
• Articles — Deep dives tied to each module, from “Chain-of-Thought for UX Research Synthesis” to “CO-STAR Briefs That Survive Contact with Stakeholders.” Practical, not academic.

The Leader View

When a design leader switches to the Leader view (yes, it’s a toggle — same product, different lens), the entire interface transforms:

• Dashboard — A command center showing active budgets, over-limit alerts, org utilization, and month-over-month efficiency gains. The KPIs a leader actually needs.
• Token Efficiency — How much is the team spending? Where’s the waste? Which teams are improving their tokens-per-outcome ratio?
• Model & Energy — Which AI models are being used, at what cost, with what energy footprint? Sustainability isn’t a nice-to-have when you’re running 168 million tokens a month.
• Budgets & Limits — Set guardrails. Get alerts. Review overage requests. This is the governance layer that turns “everyone has access” into “everyone has access with accountability.”
• Analytics — Who’s learning? Which frameworks are being adopted? Where’s the proficiency trending up? This is the data that makes the ROI conversation possible.
• AI Design Jobs & News — Because leaders also need to hire for these skills and stay current on the market.

The dual-view architecture isn’t a UX trick. It’s a philosophical position: the same data, seen through different lenses, serves different decisions. A designer needs to get better. A leader needs to know the team is getting better. One product. Two jobs. Zero guesswork.

The Three Frameworks That Actually Matter

I’m not going to pretend there are 47 essential prompt frameworks. For design work, three do 80% of the heavy lifting:

1. CO-STAR (Context, Objective, Style, Tone, Audience, Response)

This is the workhorse. Most prompts fail because they skip Context and Audience — the designer knows who the output is for, but the AI doesn’t. CO-STAR forces you to fill those slots, and the difference in output quality is immediate. I’ve seen designers go from three rounds of re-prompting to one-shot usable output just by adding the Audience and Tone parameters they’d been leaving out.

2. RTF (Role-Task-Format)

This is the power move for design briefs and stakeholder-facing outputs. By explicitly defining the Role the AI should inhabit, the specific Task it needs to accomplish, and the Format you need the output in, you eliminate the “this is close but not what I meant” problem. Especially powerful for anything that needs to survive contact with a VP or a cross-functional review.

3. Chain-of-Thought (CoT)

This is the framework for complex reasoning tasks — research synthesis, competitive analysis, design rationale. Instead of asking the AI for a conclusion, you ask it to show its reasoning step by step. I’ve used this to turn 12 user interviews into validated themes in 90 minutes. Without CoT, the same task takes a full day and the output is less trustworthy because you can’t trace how the AI got there.

The Metrics That Make This Real

Here’s what the Leader dashboard tracks — and why each metric matters:

Active Budgets — How many teams have allocated AI spend. Not a vanity metric — it tells you organizational commitment vs. pockets of enthusiasm.

Org Utilization — What percentage of available AI capacity is actually being used. Low utilization means you’re paying for seats nobody’s sitting in. High utilization with low efficiency means you’re paying for noise.

Efficiency Gain (MoM) — Are teams getting more output per token over time? This is the single most important metric. If it’s flat after 3 months, your “AI adoption” is just AI access.

Token Health — Tracking 150M+ tokens a month across an org sounds impressive until you realize several budgets are over limit. The Token Health widget gives leaders the at-a-glance view: total spend, overage count, and a one-click path to the budget details. (These are illustrative figures showing what the dashboard surfaces — your org’s numbers will vary.)

Energy Footprint — Because every token has a compute cost, and sustainability reporting is coming for AI spend the way it came for cloud spend.

Why This Matters for Hiring (and Getting Hired)

If you’re a design leader reading this: the designers who will be most valuable in 2026 and beyond are not the ones who “use AI.” They’re the ones who use AI with frameworks, measurably, and improving over time. The Design Intelligence Portal is one possible answer to the question: “How do I know if my team is actually getting good at this?”

If you’re a designer reading this: learning to prompt well is not optional anymore. It’s a core craft skill, like wireframing or user research. The gap between “I use ChatGPT” and “I use CO-STAR to generate one-shot design briefs that survive stakeholder review” is the gap between a junior and a senior in this new landscape.

And if you’re looking for someone who thinks about these problems at the intersection of design craft, enterprise scale, and responsible AI — let’s talk.

Try It Yourself

The Design Intelligence Portal “concept” is live at puneetarora94.vercel.app. Toggle between Designer and Leader views. Explore the frameworks. Judge the thinking, not just the pixels.

This isn’t a demo. It’s a position paper in product form.

Puneet Arora is a Principal UX Designer and AI Experience Lead with 20 years of enterprise experience at HP, Boeing, Software AG/IBM, and Dell Technologies. He holds 8 USPTO patents and 8 defensive publications, and serves as guest faculty at design universities across India. He is currently exploring AI Experience Lead and Design Strategy roles.

Connect on LinkedIn

Tags: #UXDesign #AIDesign #PromptEngineering #DesignLeadership #AIStrategy #DesignOps #ProductDesign #EnterpriseUX #ResponsibleAI #DesignIntelligence #AITokens #TokenCost #AICost #TokenMaxxing #ArtificialIntelligence #AI #GenAI #PromptEngineering #Prompting #Anthropic #Claude