Article9 min readUpdated
Do I Need a UI UX Designer in the AI Era? A Plain Answer
Do I need a UI UX designer when AI can draw screens? A practical look at what a designer does that a tool does not, with Indian examples.
Imagine a founder in Siliguri who runs a small business selling packed spices. She types a prompt into an AI tool: "Design an ordering app for spice buyers." In two minutes she has six clean screens. They look professional. She sends them to a developer and asks for a quote.
Three weeks later the problems start. The developer asks what happens when a UPI payment fails halfway. Nobody knows. A shopkeeper in a small town opens the app on a mid-range phone with weak network, and the home screen shows a blank white space. The founder realises the six screens were only the start of the work, not the end.
So, do you need a UI UX designer when AI can draw screens? For a throwaway experiment, no. For a product that real people will pay for and use daily, yes. A designer does a set of jobs that an AI tool does not do, and those jobs decide whether the product works.
What AI is genuinely good at in design work
I want to be fair first. AI tools are useful, and a good designer uses them as assistants.
They are good at:
- Producing many rough layout ideas quickly, so you can react to something instead of a blank page.
- Writing first drafts of placeholder text, button labels and sample data.
- Suggesting colour combinations and type pairings to start a conversation.
- Summarising long notes from a meeting into a short list.
- Making small variations of an existing screen.
If your question is "can I get a picture of an app quickly", the answer is yes. If your question is "can I get an app that works for my customers", that needs more than a picture. The rest of this article explains the gap, in the order I meet it during a project.
A designer talks to your real users
An AI tool has never met your customer. It predicts what a typical screen looks like, based on patterns from many other products. It cannot know that your buyers are wholesale traders who place orders by sending a WhatsApp voice note today.
When I start a project, the first stage is understanding. I ask the founder who the users are. Then I ask to speak with a few of them, even for 15 minutes on a phone call. I listen for the words they use, the tools they use now, and the moment they get stuck.
Here is a small example. Suppose a coaching centre with 300 students wants a fee payment app. The founder believes parents want a dashboard with charts. In three phone calls, parents say they want one thing: a reminder with a payment link on WhatsApp, and a receipt they can show to a relative. The dashboard is not needed. That finding changes the whole product, and no prompt would have produced it.
A designer decides what not to build
An AI tool is eager. Ask for an app and it gives you every screen it has seen in similar apps: profile, settings, notifications, wishlist, referral, dark mode. Each screen looks reasonable. Together they become a product that takes eight months to build and confuses new users.
A designer's most valuable skill is often subtraction. I ask: what is the one thing a user must be able to do in the first week? Everything else waits for version two.
For a food ordering app, the first version needs: browse a menu, add to cart, pay, track the order. A loyalty programme can wait. A founder who skips this thinking pays a developer for features nobody uses.
This is a business decision as much as a design decision. It depends on your budget, your timeline, and your team size. An AI tool does not know any of these three things unless you tell it, and even then it does not weigh them against each other. A designer who has worked with you does.
A designer handles the screens nobody prompts for
Most generated designs show the happy path. The user logs in, finds the product, pays, and everything works. Real life has many other paths. These are called edge cases: the situations that happen less often but hurt a lot when they are handled badly.
Here are the screens I design that a prompt usually misses:
- Failed payment. The UPI app opens, the user approves, and the money leaves the bank, but the order shows as pending. The screen must say clearly what happened, whether money will be refunded, and how long it takes. See UPI payment UX and checkout design for how I think about this.
- No network. A user in a small town loses signal in a lift. The app should keep what they typed, show a plain message, and retry when the signal returns.
- Empty states. The first time a seller opens the dashboard, there are no orders. A blank screen feels broken. A good empty state explains what will appear here and offers one clear next step.
- Wrong input. A user types a 9-digit phone number. The error message should say "Enter a 10-digit mobile number", not "Invalid input".
- Session expiry, slow loading, permission denied, long names, and very long addresses.
Each of these needs a decision, and each decision affects what the developer builds. When these screens are missing, the developer invents the answer on the spot. The result is inconsistent and often confusing for the user.
A designer keeps fifty screens consistent
An AI tool generates each screen as a separate event. Ask for ten screens one after another and you often get ten slightly different buttons, spacing, and icon styles. The first three look fine. By screen twenty, the product feels like it was made by several strangers.
Consistency is not decoration. It is how users learn your product. If the "Save" button is green on one screen and blue on another, users hesitate. If the back arrow moves, they feel lost.
Designers solve this with a design system: a shared set of colours, type sizes, spacing rules and reusable components. In Figma, I build these as components once and reuse them everywhere. When the founder says "make the buttons less rounded", I change one component and fifty screens update together.
For a startup, a small system is enough. I explain how to start in design systems for startups.
A designer knows your local context
Patterns from global products do not always fit India. A tool trained on many screens will lean toward what it has seen most. A designer who knows your market will lean toward what your users need.
Some examples from Indian products:
- UPI. Users expect to pay with the app they already trust. The payment screen should show the UPI option first, handle the app-switch, and recover when the user returns without paying.
- Regional languages. A screen that fits in English can break in Bengali or Hindi, because the words are longer and the letters are taller. Buttons must grow. Fonts must be chosen with care. I wrote about this in regional language app design in India. My first language is Bengali, so I check these things closely.
- Cheap phones. Many users have mid-range Android phones with limited storage and slow networks. Heavy images and animations make the app feel slow. See designing for low-end phones and slow networks.
- Trust signals. For loans, tax and health apps, users are afraid of making a mistake with money. Small things such as clear wording and visible support options matter a great deal.
An AI tool can produce Bengali text. It does not know that your user in Burdwan will not trust a loan screen that asks for a PAN number before explaining why.
A designer is accountable and hands over something buildable
When a design fails, someone must answer. A user cannot complete a payment, an elderly customer cannot read the text, a developer cannot build a screen. With an AI tool, there is nobody to ask. The tool does not follow up, does not own the result, and does not remember your project next week.
A freelance designer is a person who agrees to deliver something, who explains each decision, and who fixes problems that appear. That is why I include revision rounds and why I stay available after hand-off. Accountability is not a feature of the file. It is a relationship.
This matters for accessibility too. Contrast, text size, and tap targets affect real people. A designer checks these on purpose.
Buildable hand-off
A screen image is not a specification. A developer needs to know: what happens when the user taps this? What is the maximum length of this field? Which states does this button have: normal, pressed, disabled, loading? What happens on a small screen?
At the last stage of my process, prototype and hand off, I prepare the Figma file with named layers, spacing values, component states and notes on each flow. I also stay available to answer the developer's questions. A generated image, by contrast, is flat. The developer must guess every detail, and every guess costs time.
My checklist for this stage is in Figma developer handoff checklist.
So who should do what
The honest answer is a split. Here is how I would divide the work for a typical small project.
| Task | AI tool | Designer |
|---|---|---|
| First layout ideas | Good for speed | Chooses and refines |
| Placeholder text and sample data | Good | Rewrites for your users |
| Talking to real users | Cannot do | Does this |
| Deciding the first version scope | Cannot weigh your budget | Does this |
| Error, empty and offline screens | Often missing | Designs and documents |
| Consistency across screens | Drifts over time | Builds a design system |
| Local context: UPI, language, devices | Generic patterns | Applies local knowledge |
| Developer hand-off | Flat images | Specified Figma file |
| Responsibility for the result | None | Yes |
If your budget is very small, you can use AI to explore ideas and then pay a designer for the parts in the right-hand column. That costs less than a full project and avoids the expensive mistakes.
When you may not need a designer
I do not want to oversell. Some situations do not need a professional designer yet.
- You are testing an idea with five friends and will throw the result away.
- You are building an internal tool for three people in your own office.
- You need a rough picture to explain your idea to a developer or investor.
In these cases, an AI tool or a simple template is a sensible choice. The moment real customers, real money, or a real team depend on the product, the situation changes.
If you are still deciding between options, how to hire a freelance UI/UX designer in India helps you compare. If you want to talk through your own product, you can get in touch and describe what you are building.
Frequently asked questions
Can AI replace a UI UX designer for my app?
+
AI can produce first drafts of screens quickly. It does not talk to your users, design error and empty states, keep many screens consistent, or take responsibility for the result. For a real product, you still need a person for those parts.
Is UI UX design still relevant now that AI tools exist?
+
Yes. The tools make drawing screens faster, but the harder work is understanding users, choosing what to build and handling edge cases. That work is still done by people.
Should I hire a designer or use AI for my startup app?
+
Use AI for early ideas and rough notes. Hire a designer for the flow, edge cases, consistency and hand-off to developers. Many founders do both to keep costs sensible.
What is an edge case in app design?
+
An edge case is a situation that happens less often, such as a failed UPI payment or lost network. It needs its own screen or message, or users get confused.
When can I skip a designer?
+
If you are testing an idea with a few friends, building a small internal tool, or making a rough picture for a developer, a tool or template is often enough.
Working on something like this?
Tell Tithi what you are planning. You get a clear scope, a fixed price and a timeline before any work begins. I reply to every enquiry within 24 hours.