Frequently Asked Questions About Aliena Cai AI-Augmented Design Method
22 answers covering everything from basics to advanced usage.
// Basics
What is the Double Diamond design process?
The Double Diamond is a two-phase design framework. The First Diamond is the problem space — diverging through research, then converging to a well-defined problem. The Second Diamond is the solution space — diverging across solutions and prototypes, then converging through testing to a validated, engineering-ready design. Aliena's method maps where AI helps in each diamond and where human judgment must be protected.
What does 'Severe Shrinkage of the First Diamond' mean?
It's the common failure pattern where business stakeholders pre-define both the problem and the solution before designers are involved, leaving designers with only a visualization role. When you're handed a finished PRD and told to deliver screens, the problem space has been skipped. Aliena's Three Stakeholder Emails are the countermeasure to reopen it.
What is the 'fake project' problem in design portfolios?
A fake project is a portfolio piece — typically a redesign of a famous app — that ignores business reality, competitive context, and revenue viability. Hiring managers read it as evidence that a designer lacks business sense. The fix is running AI-assisted market analysis and revenue modelling upfront and producing a PRD, so the work demonstrates commercial acumen.
What is primary research and why can't AI do it?
Primary research is first-hand user learning — direct interviews, observations, and contact with real users. AI can't do it because online searching and chatbot answers are not first-hand contact; they miss the nuance, contradictions, and unspoken context that only direct interaction surfaces. AI can draft recruiting emails and interview questions, but the synthesis and judgment stay with you.
// How To
How do I run AI-assisted market validation on a design idea?
Prompt ChatGPT or an equivalent with your idea and request at minimum a market analysis, a revenue model design, and a revenue prediction. Repeat this for three to five different ideas before committing to one. You're aiming for business sense, not just user sense — the goal is to select the idea with the strongest commercial case before touching Figma.
How do I write the Three Stakeholder Emails with AI?
Ask AI to draft each email from the business stakeholder's perspective to make the argument more persuasive. Email 1 goes to the project owner requesting discovery-meeting access before the PRD is finalized. Email 2 is a kick-off alignment invite to all stakeholders before the problem is locked. Email 3 proposes a short primary user interview plan to the research team and PM.
How do I use Figma's Replace Content feature effectively?
Select the text elements you want to update — you can select multiple simultaneously — then describe what each element represents so Figma AI generates contextually accurate copy. Batch-replacing this way is faster than copy-pasting from an external AI tool. Do this as soon as you begin laying out screens to eliminate Lorem Ipsum early.
How do I test layouts for international markets before handoff?
Use Figma's Translate To feature to translate all text within a selected frame into your target language in a single action. This stress-tests your layouts against longer or shorter strings in other languages before engineering handoff, catching truncation and overflow issues while they're still cheap to fix.
How do I document iterations without repetitive strain?
Keep the latest version at the top of your Figma file and allocate one page per major iteration. On each page annotate — in large, clear text — what changed, why, and who requested it. Use ChatGPT to convert rough spoken or written notes into clean point-form annotations, and paste AI meeting summaries from tools like Granola as a baseline.
// Troubleshooting
My AI-generated prototype looks great — should I just ship it?
No. Getting emotionally attached to the first AI-generated prototype is a core pitfall. Always run multiple parallel Figma Make variants with the same prompt, varying the AI model per run, then evaluate all versions and extract what works in each before converging. Diverge before you converge — this applies in the solution space just as much as the problem space.
An engineer said my interaction is 'too hard' to build — what now?
Don't present static annotations for complex interactions — they invite feasibility vetoes. Instead, select the frame, send it to Figma Make, describe the interaction in plain language, and generate a working prototype in minutes. Demoing a functional interaction shifts the conversation from 'is it possible?' to 'does this feel right?' and gives you far more design authority.
I was told to deliver designs immediately — how do I fit in research?
Recognize that if the problem and solution were pre-defined, the First Diamond has been shrunk. Flag this explicitly, then send the Three Stakeholder Emails to reopen the problem space — request discovery access, propose a kick-off alignment meeting, and pitch a short primary interview plan. Even a small research window can meaningfully change direction and expand your influence.
I keep over-polishing before testing — how do I stop?
Adopt the 7–8 out of 10 quality threshold from the Flight of the Bumblebee workflow. The goal at the prototype stage is to test fast, not to perfect. Iterate in fixed order — structure and happy path first, then content, then visuals — and get in front of real users before adding more polish. Over-polishing before validation wastes time.
// Comparisons
How does this method compare to a traditional non-AI design process?
A traditional process is slower in the problem space and rarely stress-tests business viability early. This method uses AI to aggressively widen the First Diamond with market and revenue analysis, accelerate PRD creation, and generate multiple prototype variants — while keeping primary research, synthesis, and human empathy human-led. It's faster and more business-credible without sacrificing rigor.
How does this compare to fully AI-driven 'vibe designing'?
Vibe designing hands the entire process to AI and ships the first plausible output. This method rejects that — it mandates primary research, parallel variants before converging, documented iterations, and returning final work to a proper Figma file with design system adherence. AI is a strategic accelerator here, not a replacement for judgment, which is what keeps you irreplaceable.
Is a PRD really better than a design brief?
Yes, for anything beyond a pure visual exercise. A design brief carries only the user perspective and omits business goals and technical constraints, which makes personal projects feel fake. A PRD defines business objectives, success metrics, scope, constraints, dependencies, and risks — anchoring all solution-space work and demonstrating systems thinking to stakeholders and hiring managers.
What's the difference between Figma Make and Figma's Replace Content?
Replace Content swaps placeholder text for contextually accurate copy within existing designs. Figma Make generates functional, interactive prototypes from text prompts or existing frames. You use Replace Content to fix content in static layouts, and Figma Make to prototype complex interactions or spin up full parallel prototype variants in the Flight of the Bumblebee workflow.
// Advanced
How do I maximize divergence when generating parallel prototypes?
Paste your PRD into Figma Make and generate a prototype, then immediately start two to four additional Figma Make projects with the exact same prompt. Because each run produces different output, and using different AI models per run pushes them further apart, you get genuinely varied directions. Evaluate all of them and extract what works in each before converging.
How does user testing double as demand validation in startups?
In startup contexts, testing with real users isn't just usability validation — it doubles as early waitlist building and demand signal. When you demo a 7–8 out of 10 prototype to real users and they want access, you've validated both the experience and the market appetite before investing more time. Internal role-play testing is acceptable but always secondary to real users.
What is the Happy Path and why does it come first?
The Happy Path is the core experience roughly 80% of users go through. It comes first in the Flight of the Bumblebee iteration order because if the primary flow isn't super intuitive, no amount of content polish or visual refinement matters. Nail structure and the happy path before touching content, then visuals — in that fixed sequence.
How do I propose a feature on a page owned by another team?
Take a screenshot of the existing page, start a new Figma Make project, paste the screenshot, and describe your proposed interaction in plain language to generate an interactive prototype in minutes. Demo it live in the stakeholder meeting. If it moves forward, bring the design back into a Figma file, annotate it, and clarify the demo is a reference — not a spec — and the final follows the design system.
Why is being human framed as a competitive advantage against AI?
Because AI amplifies what you bring but can't substitute it. Empathy with difficult clients, levity in tense meetings, and leaving people feeling at ease or inspired are the qualities people remember — and the ones that surface your name when opportunities arise. As AI commoditizes execution, these durable human qualities become the differentiator that protects your career.