The AI Web Search Landscape

The present and the future

8 min readDec 31, 2024

The search market is massive! In 2024, Google continues to dominate the global search market, retaining a substantial 90.01% share and valuing its search business at approximately $1.7 trillion. Even a small portion of the global search market can be a fortune. AI search engines have been on the rise this year, challenging Google in the search market. In this overview, I’ll share the current AI search landscape with a focus on web search, a list of the AI search engines I’ve explored, and my thoughts on AI search engines’ features and future. Let me know if you have any insights or if I’ve missed anything!

Disclaimer: Opinions are my own and do not reflect the opinions of my employer.

Traditional search vs. AI search

The AI Search Engine Key Features

📝 Basic features: Given a user query, an AI search engine returns a concise summary alongside relevant sources from the internet.
✅ Quality control: While most AI search engines offer a similar user experience (quite boring actually 🥱), the quality of search results can vary widely. Ultimately, none of the additional features mentioned below would be meaningful if the search quality is poor.

Here are some advanced features I’ve come across while exploring various AI search tools. No single tool currently incorporates all these features, but they would make for a nice AI search engine:

🖼 ️Modality search capabilities: AI search engines should not be limited to text-based queries. With modality-based searches, users can search through images or videos, making the search experience richer and more accessible across different types of content.
✏️ Source editing and interaction: Allow users to interact with the sources returned. Users can directly add or edit sources or instruct the AI to update or refine its information, giving users more control over the results.
📂 Source filtering and categorization: An ideal AI search engine would categorize sources automatically, enabling users to filter results based on criteria like website, topic, sentiment, language, or relevance. This helps in cutting through noise to find the most useful information.
🌐 Additional sources: Incorporating diverse platforms such as Reddit, X could enhance results with real-time and community-driven insights. I’m surprisedly amazed that some AI search engines can actually search and provide results on very gated sources — it’s possible they have partnerships (I have my doubts 🤔).
📸 Contextual results: Building on the previous point, AI search results should go beyond traditional links by offering options to include related images, videos, tweets, and other multimedia content. For example, if I search for “Le Chat,” I’d like to see a summary from reliable websites alongside top tweets and Reddit posts about Le Chat. Having the ability to specify and customize additional sources would make the search experience more comprehensive and tailored.
🎨 Visual features: Some AI search engines offer visual tools like mind maps and tables to present information in an organized and intuitive way. More creative visuals could be used in presenting results.
🧩 Chain of Thought (CoT) Reasoning: By using CoT reasoning, AI can break down complex queries into step-by-step solutions or plans, offering more thought-through and well-reasoned outcomes.

The Future of AI Search Engines

💡 Personalized Search Experiences: AI search engines are expected to become more and more personalized, understanding individual preferences and context. For instance, if I search for “le chat,” the search engine should recognize that I’m talking about the AI tool instead of “cat” in French.
📚 Domain-Specific Search: Improved in-depth domain-specific search and tools are needed! While some specialized search tools exist, very few excel at truly effective domain-specific searches, and none of the general-purpose tools currently handle it well.
⚙️ Integration of Agents and Workflows: Combining AI search engines with agents and workflow tools could create an ecosystem where searches seamlessly integrate with other tasks.
🧐 Enhanced Evaluation and Thinking: By integrating deeper reasoning and evaluation capabilities, AI search engines could automatically analyze and assess sources, providing users with the most relevant, high-quality information.
🔗 Broader Source Exploration: Future AI systems could go beyond their current source pools, tapping into a wider range of platforms. This would enhance the comprehensiveness of results and provide more thorough answers.
🏢 Enterprise search: Obviously enterprise search is going to be a major topic 💰, but it’s beyond the scope of this article. Let’s talk about it another time!

App List

General-Purpose AI Assistants from Model Builders

Le Chat: Le Chat is a conversational entry point to interact with the various models from Mistral AI. Le Chat is my go to chat interface — it is free and provides high search quality. I highly recommend giving it a try. (Disclaimer: I work for Mistral AI. My preference might be biased.)

Other general-purpose AI Assistants with search capabilities that I’m not going to discuss in details here ; )

Google Gemini
ChatGPT
Claude
Meta AI
Deepseek

Browser-based

Browser-based AI search engines have lots of potential because browsers are an integral part of our daily lives. When implemented right, they can offer a seamless AI-driven experience, enhancing everyday browsing activities. Unfortunately, I have not seen a great browser-based AI experience yet.

Brave Leo: I love that Brave Leo AI is powered by a Mistral AI model.

Microsoft Copilot

Arc: Arc provides a few AI integrations in the browser including:

Ask on Page: Hold down Command + F on any page to ask a question and let Max answer it for you in seconds.
Second Previews: Hover and press shift over any link to generate a preview of the webpage, without a single click.
Ask ChatGPT: Start typing “ChatGPT” into the Command Bar, hit Tab, and get answers to your questions in fewer clicks.

Doubao: Doubao is an interesting and fast-growing Chinese browser . Doubao has the best and most comprehensive AI browser integrations. Too bad it’s only Chinese.

AI Search Engine Startups

Perplexity: Perplexity is an AI-powered answer engine that provides accurate, trusted, and real-time answers to any question. Launched in August 2022, it has quickly grown to serve 15 million active users.

you.com: You.com is an AI-powered productivity engine that began as a personalization-focused search engine. It has millions of active users and has served over 1 billion queries.

Poe: Poe is a generative AI chatbot owned by Quora. It was first announced in December 2022. As reported in May 2024, it has over 7 million monthly active users. Of course, Quora is not a startup at this point. But since Poe is a relatively new tool, let’s include it in the AI startups list for simplicity.

Phind: Phind is an intelligent answer engine for developers. It was founded in 2022 and it has over 1.3 million paid subscribers as of mid-2024.

KomoAI: Komo AI is another AI-powered search engine that was founded in 2024 by former Google Search PM.

Felo: Felo was founded in 2024 and headquartered in Tokyo. Felo AI does not publicly disclose its user count, but they mentioned that they have grant access to over 245 million academic publications, suggesting a significant user base. I love that Felo provides AI search for specific website like Reddit and X.

Genspark: Genspark was founded in 2023. It has interesting features like AI-powered data search and cross check.

Onion AI: Onion AI combines all the AI search engines. It’d be nice if it can compare results among different engines.

Flowith: Flowith was launched in 2024 with its user base growing to 100k during the beta phase. Unlike traditional chat-based AI tools, flowith’s interaction based on a two-dimensional canvas allows users to simultaneously interact with multiple AI models within a single workspace. It reminds me of more complex workflow tools like Gumloop and Memorality.

Andi: I like that Andi has a Chrome extension, can filter results, provides command shortcuts. However, the summarization results is not the best compared to alternatives.

ColomboAI: an AI search engine combined with social feed and shopping. I like that it can search images and videos. But these features seem limited.

Other

There are several open-source AI search engines like Khoj, Lepton, Turbo Seek, Perplexica, Sensei. However, their features seem limited and the user experience is often not yet ideal.

Additionally, some AI search engine companies optimize for better search results or visualization like exa.ai and Tako.

Of course there are also many domain-specific AI search engines. For example, some are specialized for research like consensus, reportify, limina, iris.ai, scholarai, SciSpace, Scite.ai, Research Rabbit. Many others specialize in the enterprise knowledge base, which would require a separate article to explore in detail — so we’ll leave that discussion for another time.

I actually quite like some of the domain-specific AI search tools. Unlike the general AI search engines that all provide very similar experience, some domain-specific tools are unique and different in their own ways.

There are also several other tools I haven’t had the chance to try yet. If you’re interested, there’s an awesome-ai-web-search list that features even more options to explore.

That’s all for now! I’m excited to see where the AI web search landscape evolves in the future — it has the potential for dramatic changes. I’d love to hear your thoughts or know if I missed anything. And feel free to connect with me on LinkedIn and Twitter ❤️