I argued with a founder last week about app vs ai vs com for twenty minutes before either of us opened a registrar tab. He wanted .ai because investors “get it.” His cofounder wanted .com because their enterprise pilot still types extensions like it's 2014. I wanted them to stop treating the TLD like a personality quiz. Full disclosure: I've shipped on all three, and I've apologized for one hyphenated .com I approved in 2022.
If you're choosing a startup hostname in August 2026, the extension choice is a decision about trust, category, and how people type you — not a fashion contest.
I verified current retail patterns the boring way: registrar pricing screens, aftermarket asks, and how founders actually introduce themselves on calls. For registration policy context I still keep ICANN bookmarked. For sales comps when the clean string is taken, I open NameBio before I invent a mashup.
Quick answer: which should a startup pick — .app, .ai, or .com?
Pick .com when trust and direct navigation matter most; pick .ai when the product is clearly AI and the exact stem is speakable; pick .app when you're shipping a product brand that benefits from HTTPS-forced .app semantics — and treat .io as a strong niche alternative for developer audiences, not a universal default.
| Method / TLD | Typical cost signal | What it signals | Best for |
|---|---|---|---|
| .com handreg or aftermarket | Retail ~$10–$15/yr; premiums vary wildly | Default trust, email credibility | Broad B2B, consumer, fundraising optics |
| .ai handreg or aftermarket | Often ~$80–$100/yr renewals; premiums common | “We build AI” | AI products with clean stems |
| .app handreg | Mid retail; HTTPS required by design | Product / app identity | SaaS, mobile, tooling brands |
| .io handreg | Higher than .com retail historically | Dev / startup shorthand | Infrastructure, developer tools |
My decision sentence after the table: if the exact .com is available and speakable, take it unless the brand is intentionally AI-coded and .ai is cleaner. If .com is gone, don't hyphenate your way into regret — compare .app / .ai / .io on speakability next.
Method 1: Claim the .com when trust is the job
.com is still the extension people assume. Enterprise IT, older buyers, journalists, and random conference introductions default to it. I've watched demos die in the first minute because someone typed Company.com and landed on a parked page owned by a stranger.
How to claim a startup .com — step by step
- Write the spoken brand out loud. If two people spell it differently, fix the string before you shop extensions.
- Search exact match at a major registrar. Confirm availability without add-on spam blinding you.
- Run a quick trademark sniff in your primary markets. Exact legal clearance is your counsel's job — your job is not ignoring obvious conflicts.
- If available, register for at least two years so renewal panic doesn't hit during a launch week.
- Turn on registry lock / 2FA immediately. Losing a launch domain to account theft is a special kind of misery.
- If taken, pull comps on NameBio and decide aftermarket vs alternate TLD with a written max bid — not a vibe.
Method 2: Use .ai when the category cue is the product
.ai tells the room you're in the AI category before the homepage loads. That cue is valuable when your stem is clean — Think “School.ai” class names, not twelve-character compounds. Renewals often sit closer to the $80–$100/year neighborhood depending on registrar and promo, so model holding cost honestly.
How to register or buy a .ai — step by step
- Confirm the stem is speakable without “dot A I” confusion on a phone call.
- Check handreg availability at a registrar that shows real renewal price, not just year-one promo.
- If taken, search aftermarket venues like Sedo or curated marketplaces — BIN vs offer, written floor.
- Budget escrow and transfer time if you're buying premium. Launch calendars hate surprise auth-code delays.
- Point DNS, set MX, and put a real page up. Empty .ai names get treated like speculation even when you're building.
For deeper aftermarket mechanics on AI strings, our handreg AI domains ROI notes pair well with this decision guide — different angle, same August 2026 market texture.
Method 3: Choose .app when the brand is the product
.app is my favorite “this is software” signal that isn't trying to be a dictionary lecture. Google's .app namespace effectively expects HTTPS, which is a feature for product brands. Founders who live on Product Hunt and mobile install flows often feel at home here.
How to ship on .app — step by step
- Confirm the brand reads as an application name, not a holding company or personal consultancy.
- Register the .app at a registrar that supports the TLD cleanly — verify renewal, not only checkout bait.
- Provision TLS correctly on day one. .app is not the place for “we'll do HTTPS later.”
- Secure matching social handles where you can. Extension won't save a handle collision mess.
- If you also need email trust with older buyers, consider a .com redirect strategy only when you can afford both — don't half-do it.
When I want a concrete curated example of a product-native .app brandable, I point people at Aifolio.app — not as “you must buy this,” as a positioning reference for how .app reads on a pitch deck.
What's dead or not worth it
Long hyphenated .coms. Triple-word Frankensteins. “Get” prefixes that make you sound like a coupon site. I've approved one of those under time pressure. I still regret the sales calls where prospects typed the wrong segment.
Also dead as a strategy: registering five marginal TLDs instead of one speakable home. Defensive spray feels productive. It usually funds renewals.
When is .io enough — and when should I still fight for .com?
.io is enough when your buyers are developers or startup-native and already treat .io as normal; fight for .com when enterprise email trust or mass-market typing matters more than category fashion.
I've shipped infrastructure tools on .io without apology. I've also watched consumer founders lose press links because journalists guessed .com. Match the audience. Don't match Twitter aesthetics.
If you want a second opinion lane on investor framing around AI vs IO choices, skim AI vs IO domains 2026 after you finish this decision pass — then come back to your written ceiling.
What if the exact app vs ai vs com options are all taken?
Then change the stem before you collect junk extensions — a weaker word on a strong TLD usually beats a strong word trapped behind hyphens.
My process: list three alternate stems that still sound like a company. Test them on a voice call. Only then reopen TLD shopping. I've seen teams burn a week protecting a bad stem across six extensions. That's not brand strategy. That's attachment.
I still run a stupidly simple speak test in meetings: if two teammates write different spellings after hearing the name once, the extension debate is premature. Fix the stem. I've watched teams argue TLD fashion for an hour on a name that already failed the speak test. That's theater. Product teams don't have time for theater during naming week.
Another practical note on email: if your first ten design partners are enterprise-heavy, .com friction shows up fast. If your first ten are AI researchers and builders, .ai or .io friction is often near zero. Match the inbox culture you actually sell into this quarter — not the fantasy customer from a slide.
I've also started asking founders to budget renewal before they fall in love. A .ai that feels cheap at checkout and expensive at month twelve creates avoidable resentment. Write the three-year holding cost next to the brand on the whiteboard. If that number changes the room's face, good. Better now than at renewal.
On redirects: a .com that forwards to .app can work when you own both and the product brand lives on .app. A fake “we'll get the .com later” promise usually becomes a permanent apology in the footer. I've made that apology. Once was enough.
I still run a stupidly simple speak test in meetings: if two teammates write different spellings after hearing the name once, the extension debate is premature. Fix the stem. I've watched teams argue TLD fashion for an hour on a name that already failed the speak test. That's theater. Product teams don't have time for theater during naming week.
One more practical note on email: if your first ten design partners are enterprise-heavy, .com friction shows up fast. If your first ten are AI researchers and builders, .ai or .io friction is often near zero. Match the inbox culture you actually sell into this quarter — not the fantasy customer from a slide.
Browse AI domains when you need curated options instead of raw registrar roulette, and keep FAQ handy for escrow questions once an aftermarket ask gets real. Soft close from me: pick the job first — trust, category, or product cue. Register second.
