---
title: "The Search Box With No Ads: What Metaso's AI Answers Change for Ordinary Readers"
date: 2026-11-22
category: AI in Action
site: NeuroAI
canonical: https://neuroai.site/a/na-app-metaso-cited-search
language: en
---

# The Search Box With No Ads: What Metaso's AI Answers Change for Ordinary Readers

> A Chinese AI search product built on citations and no advertising offers a different bargain: answers you can check, in exchange for a narrower business model.

A student has fourteen browser tabs open. Three are paywalled, two are copies of the same article, and the top result is an advertisement that looks almost exactly like an answer. She is not short of information. She is short of a way to trust what she is reading.

This is the problem a group of Chinese AI search products has been built around, and Metaso (秘塔AI搜索) is one of them. Made by a Shanghai company, it takes a simple position: return a direct answer, show where it came from, and do not sell the placement. Whether that is a better bargain for readers is worth thinking through carefully.

## Key takeaways

- Metaso is an AI-native search product from Shanghai-based Mita Technology (秘塔科技), the company behind earlier writing and legal-translation tools.

- Its pitch is ad-free results with visible source citations, plus a free deep research (深度研究) mode that produces a structured report with traceable references.

- The product offers different search modes, including an academic mode (学术模式) aimed at literature and citations.

- Its source-first design traces to that lineage.

- The trade-off is familiar: a product that refuses ad revenue has to find another way to pay for inference.

## Search that answers instead of listing

Traditional search engines give you a ranked list of links and sell the space above them. An AI search engine (AI搜索) reads the sources and writes a summary. That shift sounds cosmetic until you have a specific question: what a drug was approved for, how a regulation changed, which model released a feature first. A summary with citations saves the reader from opening ten tabs. It also concentrates a lot of trust in the system that wrote the summary.

Metaso's answer to that trust problem is visibility. The product displays its references so a reader can click through to the original page. Its academic mode is designed around literature search and citation output, and the company's earlier products were writing assistants and legal-translation tools. That lineage matters: the design instinct is "show the evidence," not "sound confident."

The deep research (深度研究) mode takes this further. Instead of a single response, it runs several search-and-reasoning steps and assembles a report with a traceable chain of sources. For a university student or a journalist, the value is not that the report is brilliant. It is that the work of gathering and organizing sources has been reduced from an afternoon to a few minutes, leaving the reader to do the thinking.

## Why "no ads" is a design decision, not a slogan

An advertising-funded search engine has an incentive to keep you clicking. An answer engine that shows its sources has a different incentive: make the answer settle the question. Those are genuinely different products, and the difference shows up in edge cases. If you ask about a household product, an ad-funded results page has to balance relevance against revenue. A cited-answer product can simply be boring.

But there is a cost. Every AI answer requires model inference, and inference is not free. A product that charges nothing and runs large numbers of multi-step queries is relying on subsidies, a paid tier, API revenue or an eventual enterprise business. Readers should understand that "free and ad-free" is a funding question deferred, not eliminated. This is not a criticism of the company; it is the arithmetic that every AI search product faces.

There is also a subtler risk. A citation is only as good as the source behind it. If a model picks a credible-looking but inaccurate page, the citation makes the error look more authoritative, not less. The presence of footnotes is a starting point for verification, not proof of it.

## The China-specific part

Metaso is optimized primarily for Chinese-language queries, and that is where its usefulness is most visible. Search behavior in China differs from the English-language market: readers often want an answer assembled from Chinese sources, with academic and legal material that global engines index unevenly. A product built for that environment is not simply a local copy of a Western chatbot.

For global readers, the interesting lesson is architectural rather than linguistic. The competitive claim in AI search is shifting from "which model is biggest" to "which product lets a user check the answer." That is a healthy direction regardless of which company wins. A large language model (大模型) that confidently invents a citation is worse than a smaller one that admits uncertainty and links to a page.

This also explains why the product has attracted attention among professionals rather than only casual users. Lawyers, researchers and students have a habit of asking: says who? An interface built around that habit feels natural to them, even if it is less flashy than a conversational assistant.

## What a careful user does differently

The practical change is in how you phrase a query. With a link list, you search in keywords and then evaluate sources yourself. With a cited answer, you can ask the full question in ordinary language and spend your attention on the footnotes. That is a better division of labor as long as you actually open the footnotes.

A useful habit is to ask the same question in two forms: one broad, one with a constraint such as a date or a specific institution. If the cited sources change completely, you are looking at a system that is sensitive to framing, and you should verify the underlying facts independently. This is ordinary information literacy, but AI search makes it faster to practice and easier to skip.

## Honest limitations

This is a product-and-trend deep read, not a test report. The description of Metaso's modes and features is drawn from the product's own presentation and from public directory listings; I have not run a controlled comparison against other search engines, and no current user count, revenue figure or funding valuation is used because I could not confirm one against authoritative sources. Claims about search quality are therefore deliberately qualitative. The article does not assess any regulatory filing, data-handling practice or advertising policy beyond what the product states about itself.

## What readers can do now

- When using an AI search engine, open at least two of the sources it cites before relying on the answer.

- Ask the same question with and without a date constraint; compare which sources the engine chooses each time.

- If you work with academic or legal material, try a mode that exports citations and check that each reference resolves to a real document.

---

Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-app-metaso-cited-search
Free to quote with attribution and a link to the original.
