Summarize with AI
Before demo day, fix the phone view of your homepage, not the desktop site. Investors in the room tend to open your link on a phone within minutes of your slot, often on crowded venue Wi-Fi, so the first mobile screen has to say what you do, for whom, and what to tap next.
That one screen is the whole website for the next ten minutes. You know the product well enough to pitch it in two minutes. The person in row twelve has heard thirty pitches today and gives your site about as long as it takes to load, so whatever shows up first is what they remember.
A black rectangle where the hero video should be
Picture an AI startup, used here as an example, that sells to independent pharmacies. When an insurer rejects a prescription claim, the product reads the rejection code, finds the fix, and resubmits the claim while the patient is still at the counter. It replaces a technician on hold with the insurer, or a script the pharmacy fills and never gets paid for.
The pitch goes well. The last slide shows the domain in large white type with a QR code under it. Three hundred phones come up in the room, and more on the livestream.
On a desktop the homepage works. It was redesigned six weeks ago. A full-width video shows a red rejection code turning into a green Paid tag. A thin bar above the nav reads Live in 40 independent pharmacies in Ohio. The subhead says Fix rejected claims before the patient leaves the counter.
Now open it on a phone, about 390 pixels wide, on the venue network. The video is a 6 MB file, and for the first several seconds it is a black rectangle with a grey play triangle. The thin top bar is hidden at the mobile breakpoint, so the pharmacy line is gone. The headline, Get paid for every script, is set so large it fills the screen. The subhead sits below the fold. The nav has folded into a three-line menu icon. A cookie banner covers the bottom third, and its Accept all button is bigger than the Book a demo button, which is also below the fold.
So the first screen says Get paid for every script over a black box. Nothing on it says pharmacy, claims or insurers. An investor who stops there could guess payments, film or telehealth, and leaves without knowing which.
What he reviewed, and on which screen
The founder did the work he was told to do. He rehearsed the pitch more than thirty times and cut it to the second. He signed off every section of the redesign on a 27-inch monitor, in design review, with the team around the same screen. The mobile frames were in the file too. He scrolled past them in the last review because they looked like smaller copies of the desktop.
After his slot, in the hallway, he opens the analytics on his phone. A spike of visitors, almost all on mobile, most gone in seconds. Then he opens his own homepage on that same phone, on the same Wi-Fi, for the first time since launch. He watches the black box sit there. That evening he writes follow-up emails with the deck attached, because the deck is the only thing he trusts to explain the company now.
This is the wound in its sharpest form. He knows the product so well that he reads Get paid for every script and hears the whole pitch behind it. The stranger in row twelve reads six words over a black box.
Ten minutes, three hundred phones, one round
Demo day compresses the first meetings of a round into a few days. Investors leave with a list of names and follow up on the ones they still remember. They are not going to email you to say the site would not load. They just do not ask for the meeting.
The cost is time, and time is runway. Every extra week the seed round stays open is a week the founder spends pitching instead of selling. Then the money has to stretch. Carta's Q1 2026 data puts the bar for an AI Series A near 3.5 million in ARR, where three years ago it was roughly one million. Over all sectors, about 70 percent of seed companies never get to a Series A, and the ones that do wait a median of around 616 days. A round that closes late or small begins that climb with less fuel.
The site will not win the round. But it can quietly lose a meeting, and on demo day the meetings are the round.
Why the desktop redesign is the wrong project
Most advice says to redesign your website before demo day, because investors will look. They will look. The mistake is the screen they look on. A full redesign spends weeks on the version that gets opened later, at a desk, by the few investors who already asked for a meeting. The phone view, the one everyone sees first, gets whatever the breakpoint produces.
So here is the rule. If demo day is less than a month away, do not start a redesign. Spend the time on one phone screen, the page weight and the path after the tap. A tired site that loads fast and says the thing plainly beats a new one that loads a black box. If you want the slower desktop checks investors do later, the team page and the proof, they are covered in what investors look at on your website.
Skip the special demo day landing page too. It splits your attention in the busiest month and goes stale in a week. Investors who come back later land on the real homepage anyway. Make the real homepage work on a phone and point the QR code at it.
One screen that repeats the pitch
The first phone screen should say what your last slide said, in the same words. If the room heard Fix rejected pharmacy claims before the patient leaves the counter, the headline on the phone should say that, not Get paid for every script. Matching words tell the investor they reached the right company. A clever new line makes them work out whether it is the same thing.
Under the headline, one line of proof that a stranger can check: who uses it and how many. Move it out of any bar the mobile breakpoint hides and into the hero itself. Then one button, visible without scrolling, with a label that says what happens. A still image of the product, one real rejected claim with the fix beside it, belongs under that, not a video.
Picking the words is a choice, not a template. The Mi-VAD website was written for two readers at once: investors judging whether the company is credible and clinicians judging whether the device is safe. Its homepage makes the case for small before it says a word about specifications. That is the job of your first screen too. Say the one thing the room needs to believe, before any detail.
Weigh the page on bad Wi-Fi
Venue Wi-Fi with three hundred people on it behaves like a slow phone network. The 2025 Web Almanac from HTTP Archive puts the median mobile home page at 2.6 MB in total. The hero video in our example is more than twice that on its own, before the fonts, scripts and images load.
You do not need a performance engineer to fix this. Serve a still poster image on phones and load the video only on larger screens or when someone taps play. Compress the hero image. Remove the chat widget and the cookie banner from the first screen on mobile, or make the banner a slim strip that does not cover the button. Then open the page again with your phone on cellular data, Wi-Fi off, and count how long it takes before you can read the headline.
The follow-up path after the tap
An investor who likes what they read on the phone wants one of two things: to save you for later or to ask for a meeting. Most homepages give them neither. The Book a demo button opens a calendar built for buyers, with questions about team size and current software.
Give the investor a path that fits a phone and a crowded hallway. A short line under the hero, such as Investors: email the founders, with a working email link, does more than a demo form. Make sure the team page loads on mobile and names who you are and what you did before. If you are taking meetings, say how. The pharmacy startup in our example might add a single sentence: Raising our seed now, meeting investors this week.
A seven-day phone check before demo day
You can do all of this without hiring anyone. Do it a week out, so nothing ships the night before.
- Day 1: open your homepage on your own phone with Wi-Fi off. Screenshot the first screen. Show it to someone outside the company for five seconds and ask what the company does.
- Day 2: rewrite the mobile headline so it matches the line on your last slide, word for word if you can.
- Day 3: move your one proof line into the hero so no breakpoint hides it. Delete any second button on the first screen.
- Day 4: swap the hero video for a still image on phones. Compress every image above the fold.
- Day 5: shrink or move the cookie banner and remove any pop-up or chat bubble on mobile.
- Day 6: add the investor line and test the email link. Scan your QR code from the back of a room and type the domain exactly as the slide shows it, with and without www.
- Day 7: freeze the site. No deploys until the day after demo day.
If you have more time after that, the next layer is how the company reads to a stranger at a desk, which is what makes an AI startup look like it belongs in the batch. The signals are in how to make your AI startup look like a YC company.
Studio Maydit is a web and product design studio for AI founders in the US, UK and Europe. We build in Framer, Webflow and custom code, and fixed-scope projects run three to four weeks, so start one only if demo day is further out than that. If you are choosing who should build the site, compare studios in the website design agencies ranking for YC-backed AI startups, and judge each one on its phone screens, not its desktop shots. If you want a second pair of eyes on your first mobile screen before your slot, book a 30 minute call.
