Skip to content
App Imp

AI assistants.

Limits & caps runs in 9 of the 9 apps here — nobody on this shelf has fixed it.

9 apps · 3,997 reviews held · 2,399 negative · 2026-08-24 snapshot

83/100 opportunity

6 of 6 factors · not advice
Nobody's fixed it9 of 9 apps100
How far above typical20.5% vs 2.8% typical100
How angrystar1.9 · 491 reviews70
People in the market2,496,408 ratings80
Room to enterleader holds 54% of ratings46
Willingness to pay8 of 9 charge · $5 entry89

Mostly solved: ux. 8 of 9 apps run below the store’s typical rate on it — while 9 of 9 still carry limits & caps complaints.

How this is scored

Weighted mean of the sub-scores, weights re-normalized over whatever is computable (nobody's fixed it 25 · how far above typical 20 · how angry 15 · people in the market 15 · room to enter 15 · willingness to pay 10). Fixed anchors map each count to 0–100: breadth and willingness-to-pay are straight shares; the gap’s heat tops out at 4times the store’s typical rate; anger scores star3 as 0 and star1.5 as 100; reach runs 104 to 107ratings on a log scale; room to enter is one minus the leader’s share of shelf ratings. A factor that cannot be counted is dropped and named — never imputed. The site’s one composite (#115, owner decision 2026-08-15); cited industry benchmarks are never blended in.

5 gaps.

01

Limits & caps — unfixed in 9 of 9 apps

extreme · 20.5% of this shelf’s 2,399 negative reviews, against 2.8% typical store-wide, at 1.95star average anger.

What they said · 491 reviews
“From Hope to Disappointment — What started as something genuinely new and exciting has turned into a total letdown. When Claude first appeared it felt different — a fresh way of interacting with AI, full of promise and potential. For a short time it actually surprised us in a good way. That feeling didn’t last long. It has become the most moralistic, paternalistic, and restricted AI out there. Con”Claude by Anthropic · 1star · Aug 15, 2026
“What a fall from grace — Claude used to be amazing. Fable was unbelievable. And now we have limits that are so low you exhaust a 200$ sub in a few hours, Fable and Opus lobotomized, and Claude is changing words to create a watermark. Run away!!”Claude by Anthropic · 1star · Aug 15, 2026
“Paid Monthly Subscription, But Still Have to Buy More Credits Disappointing — I’m very disappointed with Perplexity AI’s subscription. I already pay for a monthly plan, but after using it for a while, it keeps asking me to buy more credits. What is the point of paying every month if I still have to keep purchasing additional credits? The subscription should include enough usage for paying custome”Perplexity - AI Search & Chat · 1star · Aug 15, 2026
How it compares
this shelf
20.5%
491 of 2,399
store-wide
2.8%
typical
breadth
100%
9 of 9 apps
02

Accuracy — unfixed in 9 of 9 apps

elevated · 16.0% of this shelf’s 2,399 negative reviews, against 11.5% typical store-wide, at 1.65star average anger.

What they said · 385 reviews
“The most horrible experience of my life — I wanted to write a general letter, I’m a researcher, science, I gave it a reason why I was contacting somebody for a research project. It shot holes all through the research project before I could even get it off the page. When it asked me questions about it, I gave it an old reference point something that happened in 1907, but then I also gave it all of”Claude by Anthropic · 1star · Aug 16, 2026
“Hallucinates — This app can’t answer questions correctly and then ARGUES when you present the correct answer. Literally, it spits out paragraphs about how it’s definitely right… When you get the answer and feed it into the algorithm. I asked about electric fire smells and it argued for multiple paragraphs. It’s dangerously wrong.”Claude by Anthropic · 1star · Aug 16, 2026
“Got my 1st question wrong 2 times — After giving Perplexity a SS of a GE appliance manufacturer label, I asked Perplexity, “how old is this freezer?” It said 1 yr and 1 month old. I asked it to research the question again. It said it was wrong, and said the freezer is one month old. I asked Perplexity to research the question again. It said the freezer is 12 years and one month old. Which is the c”Perplexity - AI Search & Chat · 1star · Aug 15, 2026
How it compares
this shelf
16.0%
385 of 2,399
store-wide
11.5%
typical
breadth
100%
9 of 9 apps
03

Billing — unfixed in 9 of 9 apps

elevated · 40.1% of this shelf’s 2,399 negative reviews, against 32.1% typical store-wide, at 1.71star average anger.

What they said · 961 reviews
  • Price judged too high for the value88 of 961
  • Trial converted without clear consent47 of 961
  • Cancellation blocked or broken13 of 961
  • Charge the user never authorised10 of 961
“Privacy nightmare — Cancelled my membership and you should too.”Claude by Anthropic · 1star · Aug 16, 2026
“Was the best agentic browser — The main reason to consider a new browser over Chrome is a standout feature. This was the ability to use Comet Browse to navigate and interact across pages. It was available on the pro tier ($20), but they removed it. Now, users might only get 4-5 agentic tasks per month, if they're lucky. They're promoting Perplexity Computer, which is good but costly and not worthw”Perplexity - AI Search & Chat · 2star · Aug 16, 2026
“Watermarking — I’ve been a loyal user and have spent $200 a month for than 2 years but finally ended my subscription after the idiot decision to watermark the text.”Claude by Anthropic · 1star · Aug 16, 2026
How it compares
this shelf
40.1%
961 of 2,399
store-wide
32.1%
typical
breadth
100%
9 of 9 apps
04

Performance — unfixed in 9 of 9 apps

12.7% of this shelf’s 2,399 negative reviews, against 16.9% typical store-wide, at 2.12star average anger.

What they said · 305 reviews
“SLOW — THIS APP IS SO DAMN SLOWWW, I HAVE VERY FAST INTERNET, I’M CLAUDE PRO AND I’M ON THE LATEST SOFTWARE YET IT TAKES FOREVER TO FOR THE MESSAGE TO ACTUALLY ARRIVE.”Claude by Anthropic · 1star · Aug 15, 2026
“Watermarking is a terrible idea — What a terrible idea. A good idea would be to provide a hashed disclaimer at the end of a project that explains the contribution. Example. Provided 20% of the code written under author’s guidance and author’s architecture. Debugged and tested code under author’s guidance. That is a very realistic explanation of the contribution. Not a watermark that can claim ow”Claude by Anthropic · 1star · Aug 15, 2026
“Got my 1st question wrong 2 times — After giving Perplexity a SS of a GE appliance manufacturer label, I asked Perplexity, “how old is this freezer?” It said 1 yr and 1 month old. I asked it to research the question again. It said it was wrong, and said the freezer is one month old. I asked Perplexity to research the question again. It said the freezer is 12 years and one month old. Which is the c”Perplexity - AI Search & Chat · 1star · Aug 15, 2026
How it compares
this shelf
12.7%
305 of 2,399
store-wide
16.9%
typical
breadth
100%
9 of 9 apps
05

Support — unfixed in 9 of 9 apps

6.3% of this shelf’s 2,399 negative reviews, against 9.2% typical store-wide, at 2.17star average anger.

What they said · 151 reviews
“The most horrible experience of my life — I wanted to write a general letter, I’m a researcher, science, I gave it a reason why I was contacting somebody for a research project. It shot holes all through the research project before I could even get it off the page. When it asked me questions about it, I gave it an old reference point something that happened in 1907, but then I also gave it all of”Claude by Anthropic · 1star · Aug 16, 2026
“No Help n Support — I have to no complaints about Claude tge AI. My complaint is that when you need help n support you are directed to a page unresponsive to touch to log in. Because it’s unresponsive to touch you cannot lodge in to log a complaint. I’ve been attempting to get help fof 8 days by hitting the thumbs done on the chat n repeatedly telling them the log in page is unresponsive to touch.”Claude by Anthropic · 1star · Aug 15, 2026
“Doesn't support alt keyboards, broke widgets in iOS 27 — I have swiftkey installed yet as of the newest version of Perplexity, swiftkey will not appear as an option. And on iOS 27 widgets are broken on lock screens and main screens”Perplexity - AI Search & Chat · 2star · Aug 10, 2026
How it compares
this shelf
6.3%
151 of 2,399
store-wide
9.2%
typical
breadth
100%
9 of 9 apps

Is there a market for a new ai assistants app?

9 apps compete on this shelf, holding 3,997 App Store reviews — and the widest unfixed complaint, limits & caps, runs in 9 of 9 of them at 20.5% of negative reviews against 2.8% typical. Those are counts, not a verdict — the page shows what is unfixed, not whether to build.

What do users complain about most in ai assistants?

1. Limits & caps — in 9 of 9 apps, 20.5% of negative reviews; 2. Accuracy — in 9 of 9 apps, 16% of negative reviews; 3. Billing — in 9 of 9 apps, 40.1% of negative reviews; 4. Performance — in 9 of 9 apps, 12.7% of negative reviews; 5. Support — in 9 of 9 apps, 6.3% of negative reviews.

How many ai assistants app reviews did App Imp count?

3,997 US App Store reviews across 9 apps, 2,399 of them negative — counted on the 2026-08-24 snapshot, never estimated.

Who’s in it

All 9 apps counted on this shelf · 2026-08-24.

App
Grok AI1,344,52942%15%53%8%
Perplexity - AI Search & Chat498,16714%20%46%15%
Claude by Anthropic245,73612%24%24%18%
Chatbot AI Assistant - Genie199,47016%10%63%9%
AI Chatbot - Nova124,77410%7%36%10%
AI ChatBot Assistant: AskEasy32,1645%12%59%9%
Lloyd - Your AI Superpower23,5802%6%6%22%
RoboAI - Chat & Ask AI Chatbot16,75213%13%42%10%
DeepSeek - AI Assistant11,23613%17%1%12%

Each cell: that app’s share of its own negative reviews. Bold from 1.15 times the store’s typical rate for that complaint, and only where an app holds 20+ negative reviews. Click a column to sort.

What they charge, and what it costs them

Prices from Apple’s public pages; the share counts payment-mechanics language only.

AppMoney complaints
Chatbot AI Assistant - Genie$4.99–$99.99 · 9 tiers15.4% of 221
AI ChatBot Assistant: AskEasy$4.99–$69.99 · 10 tiers13.2% of 76
AI Chatbot - Nova$4.99–$59.99 · 7 tiers8.6% of 81
Grok AI$5–$300 · 9 tiers7.2% of 707
Perplexity - AI Search & Chat$5–$200 · 8 tiers6.5% of 323
Claude by Anthropic$20–$279 · 7 tiers2.8% of 543
RoboAI - Chat & Ask AI Chatbot$5.99–$69.99 · 10 tiers1.6% of 182
Lloyd - Your AI Superpower$5.99–$69.99 · 3 tiers0% of 165

What each one owns, where each is exposed

Grok AI

1,312,973 ratings · 59% negative

Owns. Largest reach; real-time X data integration

Exposed. Weekly credit caps added post-sale; opaque moderation; silent limit changes on paid tiers

Perplexity - AI Search & Chat

493,862 ratings · 60% negative

Owns. Search-grounded answers; broad topic coverage

Exposed. $20/mo paid users hit credit walls with no reset transparency; escalation path is $200/mo only

Microsoft Copilot

406,151 ratings · 41% negative

Owns. Lowest anger rate in set; Microsoft brand trust

Exposed. Conversation-count caps appear without warning; interface changes undisclosed in release notes

Claude by Anthropic

224,520 ratings · 61% negative

Owns. Strong model quality; IAP up to $279

Exposed. Memory limits trigger after paid hallucinations; mobile-desktop sync broken for paid users

Chatbot AI Assistant - Genie

199,045 ratings · 73% negative

Owns. High IAP ceiling ($99.99); broad reach

Exposed. Highest anger outside bottom tier; billing confusion at scale

Meta AI

197,157 ratings · 68% negative

Owns. Free, no IAP — zero price barrier

Exposed. Interface instability; no monetization path sets price floor at zero for competitors

DeepSeek - AI Assistant

11,158 ratings · 38% negative

Owns. Lowest anger rate; free, no IAP

Exposed. Small reach; no paid model demonstrated — anchor for free-tier expectations

Browse the apps behind these counts

Every app we count, with the complaint it runs furthest above the store’s typical rate.

Open the app library
How this is counted

Every number is arithmetic over the reviews we hold, re-counted weekly — nothing modelled, nothing scored. Themes come from a frozen pattern classifier rather than a model reading reviews, so the same review always lands in the same bucket. A shelf rate is pooled across every app here; the store rate is the same count across 12,122 apps. A complaint is only called a gap when at least half the shelf carries it and at least 20 reviews are behind it — a problem concentrated in one app is that app’s bug.