AI assistants
In the AI assistant and chatbot niche, trust has become scarcer than quality. The real ceiling of the market is 's 72 points: none of the 57 apps cleared the 80 mark, and 47 of the 57 turned out inflated or questionable.
The giants , Claude, and Gemini hold their share through scale and real usefulness on specific tasks: replacing search, working with documents, research. But even they let users down regularly: Gemini quietly switches the model from Pro to Flash mid-session, degrades with every update, Claude loses history across devices. Below the top three the picture is stark: a hundred wrapper clones competing for position through review inflation, not product.
The money sits where the user trusts the brand enough to pay for several subscriptions in parallel. That is exactly what analysts do, paying for and Claude at the same time.
A builder who solves the trust problem through verifiable accuracy, a stable personality, and predictable model behavior gets a market the giants have outgrown upward and the clones never reached.
Market overview
The niche is split into two completely different markets with different trust logic: the top five (, Claude, Gemini, , Copilot) gather a professional audience with real payment behavior, while the other 52 apps are shells over the same APIs with inflated ratings, a dying audience, and aggressive monetization that destroys trust faster than it can be built.
- Size
- 17,314,626ratings across 57 apps · 22,227 reviews read
- Concentration
- 66%of all ratings held by the top three
- Downloads
- 3.4 B+installs across the top 12 on Google Play, led by Google Gemini
- What people pay
- $20/mo$20$40$50prices cited in real reviews
- Leaders
- Revenue estimate
- What the niche's top apps make a year. The number opens together with the ideas. Unlock
- Trust
- 47 of 100apps have an inflated or doubtful star, only 0 are genuinely good
- Discoverability
- 54 of 100a new app's chance to break in: the top three hold 66% of ratings, 51% of the shelf is gamed, only 1 apps are genuinely strongComputed from leader concentration, gamed share, count of strong apps and demand size. Rough, order of magnitude.
- Money
- Who pays in this niche, for what, and why most players lose money. It opens together with the ideas. Unlock
and Claude earn loyalty on real scenarios, but neither solves the basic problem: the user does not know which model they are talking to at any given moment or why the quality changed. Gemini turns this into a systemic bug: switching from Pro to Flash wrecks the output of an hours-long session, and the user only finds out after the fact. The companion platforms , CHAI, Linky, and BALA lost their audiences the same way: every update broke the memory of a character the user had been building for months. Aggregators like Poe and Genie sell access to other people's models without adding value of their own, and are the first to lose users whenever a provider changes its pricing. The wrappers at the very bottom of the rating, AI Chat Assistant ChatNow (22 points against a 4.8 storefront) and AI Chatbot Pixi (14 points), build traffic through false ads for features that don't exist and last exactly until the user compares.
Audience
"AI assistants" is not one customer. Inside are different people with different jobs, and they pay very differently. First you choose who you build for.
Where the money is
Where users come from
Channels visible right in the reviews: people say themselves how they found the app and why they installed it. This is the niche's distribution.
Moving from Character.AI and ChatGPT6
The most powerful stream in the niche is refugees: Character.AI tightened limits and added a paywall, so people search for a c.ai alternative on purpose, while others got tired of ChatGPT or came from a dead service. A new app just needs to position itself as the alternative outright and not repeat the same limits.
I started using this app after C.AI got locked up behind a paywall, and i absolutely love it!
Persona Ai lead me hear after it shut down I like it and I’ve been on the platform for 2-3 years
TikTok and social media ads3
Both flavors work: an organic TikTok recommendation brings roleplayers tired of limits, while paid ads promise avatars and features the app doesn't have and harvest angry reviews. One developer even asks users to make TikToks for rewards.
Ive been using character AI for the past few months but i couldn’t take the ads and all the limits but on TikTok i saw someone recommend swerve and its been way better
When watching the ad for ChatNow on social media it shows a visual avatar responding to the user. When purchasing and downloading it there is no visual avatar.
Word of mouth: friends3
Roleplay apps spread through friend circles: when the usual service goes bad, friends drag each other over to the replacement. A friend's recommendation instantly overrides distrust of an unknown brand.
I was suggested this app by a friend ever since C.ai went off the rails.
Really great app my friend got me into it.
App store search for the job3
Some people aren't looking for AI in general but for a specific tool: a work assistant, an affordable image generator. They compare prices and capabilities and leave fast if the listing's promise doesn't match the product.
I was looking for a work assistant, but I found a machine that does it all.
I was looking for a reasonably priced image generator. This one was $6.99 a wek and with a free 3 day trial-unlimited.
Honest rating
The same hundred apps in two scoring systems. Switch and watch the storefront star diverge from what people actually write in reviews.
ChatGPT4.8 in store · genuine · 8,263,023 ratings72our scoreA general-purpose AI assistant that has replaced Google for millions of people, but gets noticeably worse with each model update.
Users describe specific tasks it covers every day: household organization, emotional support, document work. For a significant chunk of users it has become the only tool for information search.
Several reviews note a decline in writing quality in newer versions compared to older ones. Custom assistants lose their memory and drift off-topic over time.
A broad audience for daily tasks, especially people who have switched from Google to conversational search.
Claude by Anthropic4.7 in store · genuine · 186,833 ratings68our scoreThe best model for deep document and text work, but the mobile app falls noticeably short of the web version.
Strong structured reasoning and precision on research tasks. People pay for it alongside ChatGPT and stick with it specifically for document work.
The mobile app loses sync across devices and chat history disappears. Memory does not clear correctly and drags context from old sessions into new ones.
Analysts, editors, and researchers anyone who works with long texts and documents.
Google Gemini4.7 in store · genuine · 1,920,130 ratings62our scoreA capable Google assistant that lets you down in a non-obvious way: it silently switches between models mid-session.
Some users appreciate the professional tone without filler, depth on niche topics (history, science), and the feeling that it answers factual questions faster than Google.
Switches from Pro to Flash within a single session without any notice, which can ruin hours of work. Loses the thread and starts hallucinating after a few thousand words.
Users who want a strict information assistant without conversational fluff, as long as sessions are short.
Sakura - Chat with AI Bots4.8 in store · genuine · 20,508 ratings62our scoreA roleplay chat with customizable characters that is more open and responsive than most alternatives.
Characters maintain conversational context and do not ask repetitive questions. Developers are actively present in the community and respond to feedback.
Aggressive filters blur images without clear violations. Free-tier limits have gotten stricter recently.
Roleplay chat fans who want a less censored experience with better memory.
Microsoft Copilot4.8 in store · genuine · 399,994 ratings60our scoreThe standalone free Copilot works noticeably better than the enterprise version, though its inability to remember anything between sessions remains a real weakness.
Users describe it as a tool for specific tasks: work reports, planning, learning. Professional tone and accuracy in explanations get consistent praise.
The assistant remembers nothing between sessions, forcing users to re-explain context from scratch. Filters trigger on neutral requests and block them without explanation.
Working professionals for regular tasks: documents, planning, learning no Microsoft ecosystem required.
Perplexity - AI Search & Chat4.8 in store · doubtful · 481,040 ratings58our scoreAn AI search engine with citations that works significantly better on desktop than in the mobile app.
Users value that answers come with sources for verification. Response style adapts to the nature of the question, which stands out compared to citation-free answers.
The mobile app is consistently worse than the web version: content fails to display, streams cut off. 'Not available in your region' errors appear without explanation, even for paying users.
Researchers and professionals who need cited sources, preferably from a desktop browser.
Solvely - AI Study Tools4.8 in store · doubtful · 38,350 ratings58our scoreAn AI study assistant for students: step-by-step photo-based problem solving, explanations, and notes.
Reads problems from a photo and explains the solution step by step. Users returning to school especially value the ability to ask follow-up questions.
Loses context between parts of multi-step problems. Accuracy on harder questions is inconsistent users report error rates up to 95% on specific tests.
Students and high schoolers who need step-by-step help with math and hard sciences.
Grok - AI Chat & Video4.9 in store · doubtful · 1,268,188 ratings55our scoreA chat with a distinctive personality and image generation, but the product experience is broken by hard free-tier limits.
Users note the model's unconventional approach to tasks and fast image generation speed. A few writers use it for grammar and text structure work.
The model repeats already generated content instead of producing new variations. Long-conversation context retention is weaker than competitors, by users' own accounts.
Users looking for an AI with an unconventional edge and fast image generation who are willing to work within tight quota limits.
Next: the review findings
You already see the market, the audience and the honest rating. A free sign-in opens the first finding: what people value, what drives them away and why, with quotes.
Free, by email or Telegram
Nearby niches
Breakdowns next door: same audience, adjacent jobs.
AI image generation
AI photo
Photo editor
Wallpapers & widgets