Felo

What is Felo?

Felo is a free multilingual AI search and creation platform. You ask a question in your own language, it searches the live web, and it answers with cited sources — then, unlike a plain answer engine, it lets you turn what it found into a deck, an interactive page or a mind map without leaving the tool. Its own framing is compact: search, understand, and create with AI. The product is literally a search box first; the root domain issues a 308 redirect straight to /search, so there is no marketing homepage standing between you and the query field.

What separates Felo from the Perplexity-shaped field it competes in is where it comes from and who it is built for. It is made by Felo Inc., a company established in Tokyo in July 2024, and its stated mission is to use advanced natural language processing and large language models to build a seamless global information gateway that crosses language barriers. That is not a generic slogan here: independent coverage of the company describes its distinguishing technical strength as strengths in Japanese and other Asian language search. Where most AI search products are English-first and treat other languages as translation targets, Felo's premise is that the non-English web deserves first-class retrieval.

The company raised roughly 1.5 billion yen, about 10 million dollars, in a Series A round announced in July 2025, led by India's Peak XV Partners with participation from the venture arm of South Korea's Mirae Asset. Reporting at the time noted the company launched Felo Search in July 2024 and began offering Felo Enterprise in May 2025, with the new capital going toward research hiring, enterprise search, agent research and expansion into South Korea and Taiwan. Japanese outlets covering the round cited more than 1.2 million users and over 300,000 searches a day, concentrated in Japan, Korea and Taiwan — figures that originate with the company rather than an independent audit.

One correction worth making up front, because it shapes whether the product suits you: calling Felo simply a free AI search engine is now out of date. Search remains free and effectively unlimited at the basic tier, but the creation half of the platform — slides, interactive pages, mind maps, voice notes, image generation — runs on a credit meter, and those credits are consumed quickly.

Core Features

• Two-tier intelligent search: Everyday queries run through fast search, while heavier questions escalate. Pro Search costs 15 credits per run and the Research Agent costs 150, a tenfold difference that tells you exactly how the product weighs the two: one is a better answer, the other is an autonomous research pass that plans and executes an information-gathering task on your behalf.
• AI Slide for presentations: The most developed creation module. It generates decks page by page, supports template-based generation, allows adding or modifying individual pages, and exports to file. At the time of checking, AI Slide generation was priced at 30 credits per page against a list price of 100, with template generation at 100 against a list of 200.
• AI Page for interactive web pages: Turns a research result into a shareable interactive page rather than a static document. AI Page generation costs 375 credits per run, with regeneration priced identically — the most expensive single creation action on the platform, and a useful signal of how much compute sits behind it.
• LiveDoc and AI mind maps: LiveDoc is the document surface where material accumulates, described by the product as all docs, together, evolving. It includes a mind map tool; at the time of checking the mindmap tool was free for a limited period against a list price of 15 credits, with a separate standalone AI mind map generation also metered at 15.
• X Research for social retrieval: A dedicated module for searching X posts, which matters for questions where the useful signal is in real-time discussion rather than indexed pages. It was also running as a limited-time free feature against a list price of 5 credits per post.
• A broad utility layer: Batch search, knowledge base search and retrieval, code generation, large document summarization, business document generation, real-time visualization and voice notes are all separately metered. Notably the voice assistant is metered at 1,500 credits per hour, the single most expensive line item on the price list.
• Multi-model image generation: Rather than a single house model, Felo exposes several with explicit per-model pricing, including Nano Banana, Nano Banana 2 at four resolutions, Nano Banana Pro and GPT Image 2. Nano Banana Pro at 4K costs 300 credits, so image work is a deliberate spend rather than an incidental one.
• An LLM Playground and API: The navigation exposes Search, LiveDoc, Agent, LLM Playground and Slides as separate first-class modules, alongside an API and SEO automation — meaning the platform is positioned as something you can build against, not only a destination site.

Use Cases

1. Research across a language barrier: This is the product's sharpest use case. A Japanese, Korean or Taiwanese professional researching a topic whose best sources are in a mix of local and English material gets retrieval that treats both seriously, rather than an English-first engine that surfaces translated summaries of the local web. The reverse also holds: an English-speaking analyst covering Asian markets can reach primary-language material directly.
2. From question to deck in one session: A consultant needs a client-ready deck on a topic they do not yet know. The Research Agent gathers and structures the material, then AI Slide turns the result into pages. The value is not that either half is unique, but that the handoff between them does not exist — no copying findings into a separate presentation tool.
3. Competitive and market monitoring: X Research covers the part of the information landscape that indexed search misses, where product complaints, launch reactions and rumors circulate before they reach articles. Combined with batch search, it supports sweeping a set of related questions rather than asking one at a time.
4. Building a working knowledge base: Knowledge base search and retrieval let teams query their own accumulated material alongside the live web, with LiveDoc as the surface where documents and mind maps persist and evolve rather than being generated once and lost.
5. Enterprise reporting: Felo Enterprise, launched in May 2025, targets business users who need expert-level reports generated from search results. This is the commercial expression of the Research Agent — the same capability packaged for teams that need repeatable output rather than ad hoc answers.
6. Study and self-directed learning: For students, the free tier's unlimited basic search plus a small daily allowance of professional searches and file analyses covers a realistic study workload, with mind maps for structuring what was read. The constraint is the daily cap, not the price.

How to use Felo

1. Go to felo.ai. You land directly in the search interface — no signup wall stands between you and a first query. Create an account when you want history, credits and the creation modules.
2. Ask your question in whatever language you actually think in. This is the point of the product; translating your query into English first defeats the multilingual retrieval it is built around.
3. Choose the right search depth deliberately. Fast search for lookups, Pro Search when the answer needs synthesis across sources, and the Research Agent when you genuinely want a full research pass — and check the credit cost before triggering the last of these.
4. Read the citations, not just the answer. Every AI search product's failure mode is a confident synthesis of weak sources; the citations are what let you tell a well-supported answer from a plausible one.
5. Push the result into a creation module when you need an artifact. AI Slide for a deck, AI Page for something interactive and shareable, LiveDoc's mind map for structuring your own understanding.
6. Watch the credit meter as a first-class part of the workflow. Check the pricing page's credits table before committing to an expensive action, and note which prices are promotional — several were flagged as limited-time when checked, so the cost of a workflow you build today may change.

Tips & Best Practices

• Log in at least once a month if you rely on the free tier. The privacy policy states that if a free account goes 30 days without a login, the company may delete the associated data, including user content and files. This is the single most consequential piece of fine print for free users and it appears nowhere in the marketing.
• Match the search tier to the question. Running the Research Agent on a question that fast search would answer burns 150 credits for nothing. Conversely, using fast search for a question requiring synthesis produces a shallow answer you will not trust.
• Draft in the cheap surfaces, finish in the expensive ones. Iterate on structure in a mind map or a document before triggering AI Page generation at 375 credits, since regeneration costs the same as the original run rather than a discounted amount.
• Verify anything load-bearing against the cited source. This applies to every AI search engine, and it applies with extra force to claims that appear only in one language — cross-language retrieval widens your evidence base, but it also means you may be relying on a source you cannot easily assess.
• Treat promotional pricing as temporary when planning. Several of these prices are explicitly flagged as limited-time promotions; the list prices behind them are three to five times higher on some items, which changes the arithmetic of any workflow built on them.
• Use the language of the source material, not your own, when the sources matter more than the phrasing. Asking in the language the primary documents were written in generally surfaces better material than asking in your preferred language and hoping retrieval bridges the gap.
• Export what you need to keep. Given the free-tier deletion clause and the general volatility of generated content, anything that matters should end up in a file you control rather than living only inside the platform.

Who is Felo for?

• Japanese, Korean and Taiwanese knowledge workers: The clearest fit and the company's own primary market, with retrieval built around Asian-language strength rather than translated English results.
• Researchers working across languages: Anyone whose sources are split across languages and who currently pays a tax in manual translation and parallel searching.
• Consultants and analysts producing deliverables: The search-to-slides pipeline is most valuable to people whose work ends in an artifact rather than in an answer.
• Students and self-directed learners: The free tier is genuinely usable for study, with unlimited basic search and mind maps for structuring reading.
• Competitive and product intelligence roles: X Research plus batch search suits monitoring work where the signal appears in discussion before it appears in articles.
• Enterprise teams needing repeatable reports: Felo Enterprise exists specifically for organizations that want expert-level reports generated from search results as a routine output.
• Anyone dissatisfied with English-first AI search: If your experience of the mainstream tools is that they handle your language as an afterthought, this is the product built on the opposite assumption.
• Not a good fit for teams needing deep third-party validation: Buyers who require established independent review evidence before adoption will find that evidence thin, for reasons covered under limitations below.

Platforms

• Web application: The primary and most complete surface. The root domain issues a 308 redirect straight to /search, so the web app is the product rather than a companion to it.
• iOS app: An official app, Felo - Search the world, published by Felo Inc. It is free, carries a 4+ content rating, requires iOS 18.0 or later, and its then-current version was released on 18 July 2026.
• Language coverage on mobile: The iOS app lists eleven interface languages — English, French, German, Italian, Japanese, Korean, Portuguese, Russian, Simplified Chinese, Spanish and Traditional Chinese — which is consistent with the platform's multilingual positioning rather than the usual English-plus-a-few pattern.
• Multilingual web presence: The site itself is published in many languages, with a dedicated Japanese section at /ja; the company's about page alone is offered in sixteen or more languages.
• API and automation: The navigation exposes an API and SEO automation alongside the interactive modules, for teams wanting to build on the platform rather than only use it.
• Enterprise deployment: Felo Enterprise is offered as a team edition through a sales conversation rather than self-serve signup.

Pricing & Plans

The free tier, called Standard, is free forever and requires no credit card. It includes a daily allowance of 200 bonus credits plus a specific set of entitlements: unlimited high-speed search, five professional searches per day and three file analyses per day. Read that list carefully, because it is narrower than the product's free reputation implies — the thing that is genuinely unlimited is basic search, while the more capable search modes are capped daily and the creation modules run on credits.

Paid access is a single tier sold two ways: $14.99 per month, or $149.99 per year, which the site presents as $12.5 per month and labels a 16% saving and the most popular option. Both carry identical entitlements — 15,000 credits per month plus the same daily 200-credit bonus. The company translates that allowance into output terms, stating it covers approximately 300 professional slides, 40 interactive web pages, 150 professional documents and 140 images. Beyond subscriptions there is a direct Buy Credits path for topping up, and a Felo Enterprise team edition whose price is not published and which routes through sales.

The credit table is where the real cost model lives, and it rewards reading before you commit. Search actions are cheap relative to creation: Pro Search at 15 and the Research Agent at 150 sit far below AI Page generation at 375 or the voice assistant at 1,500 credits per hour. Note also that several of these prices are explicitly flagged as limited-time promotions — AI Slide generation at 30 against a list price of 100, template generation at 100 against 200, X post search and the mindmap tool both temporarily free — so today's arithmetic is not necessarily next quarter's. Prices and allowances change; the official pricing page is the authority.

Alternatives

• Perplexity: The category's best-known product and the obvious comparison. Far larger, more mature, English-first, and focused on search and answers rather than bundling a creation suite.
• ChatGPT with search: The general-purpose assistant with web access. Broader as an assistant, less specialized as a retrieval engine, and without the presentation and page-building output formats.
• Genspark and similar search-plus-agent products: A cohort positioned like Felo, pairing agentic search with generated artifacts, generally without the deliberate Asian-language retrieval focus.
• Gamma and dedicated AI presentation tools: If your actual need is decks rather than research, a purpose-built presentation tool will go deeper on design control than Felo's AI Slide module.
• Traditional search engines plus a separate assistant: Still the default workflow for many, cheaper and more transparent about sources, but requiring you to do the synthesis and the artifact-building yourself.

Limitations & Considerations

• User satisfaction evidence is thin and not especially strong. The one third-party rating that could be verified first-hand was the iOS app's: 3.38 out of 5 across just 21 ratings. That is a small sample and should be read as directional only, but it is notably below what a product's own marketing would suggest, and no stronger independent evidence was available to balance it.
• Independent review platforms could not be checked. G2 returned HTTP 403 across two independent fetch channels, so its rating and sample size could not be confirmed. Community searches surfaced mainly competitor-authored comparison pages and affiliate-style reviews, which were excluded. The practical consequence: this product has less verifiable independent validation than its funding profile might lead you to expect.
• The free tier deletes inactive users' data. The privacy policy states that if a free account goes 30 days without a login, associated data including user content, files and related records may be deleted to optimize system resources. Free users treating Felo as durable storage are relying on something the policy does not promise.
• The free allowance does not stretch to the creation features. The daily 200-credit allowance does not cover a single AI Page generation at 375 credits, and one Research Agent run at 150 consumes most of a day's budget. The free tier is a search product; the creation platform effectively requires a subscription.
• Promotional pricing obscures the real cost. Several of the most attractive prices are limited-time offers over list prices three to five times higher. Budgeting a workflow on today's promotional rates carries obvious risk.
• The training-data commitment is uneven. The privacy policy makes a strong, specific promise about Google user data — it will not be shared with any third party, naming OpenAI and ChatGPT explicitly, and will not be used to develop, train or improve any general or non-personalized AI or ML model. But that protection is scoped to Google user data. For ordinary user content, the policy says only that personal information may be used to improve our services and conduct research, with no equally explicit denial covering ordinary user content. If your queries are sensitive, that gap matters.
• Company details are not fully consistent. The About page lists two different company addresses in Tokyo, while media coverage gives a third ward. Minor in itself, but a reminder that self-published company information warrants the same scrutiny as any other claim.
• The product is young and moving fast. Founded in July 2024, with the enterprise product less than a year old at the time of the Series A and heavy promotional churn in the price list, this is an early-stage platform. That brings rapid capability growth and, equally, instability in what a given plan buys.

FAQ

Q1. Is Felo free to use?

Yes, with real limits. The Standard tier is free forever with no credit card, and gives unlimited high-speed search, five professional searches per day and three file analyses per day, plus a daily 200-credit bonus. The creation features consume credits, so free access is best understood as a search product rather than a full platform.

Q2. What does the paid plan cost?

There is one paid tier, sold monthly at $14.99 or annually at $149.99, which the site presents as $12.5 per month with a 16% saving. Both give 15,000 credits per month plus the daily 200-credit bonus. Credits can also be purchased directly, and a Felo Enterprise team edition is available through sales.

Q3. How do credits work?

Every action has a posted price. Pro Search is 15 credits per run and the Research Agent is 150; on the creation side, AI Page generation is 375 credits and the voice assistant is 1,500 credits per hour. The pricing page carries the full table, which is worth reading before committing to expensive actions.

Q4. Which languages does it handle?

Multilingual support is the product's core premise rather than an add-on. The iOS app lists eleven interface languages, and the company's own pages are published in sixteen or more. Independent coverage identifies its distinguishing strength as Japanese and other Asian language search.

Q5. Who makes Felo and where is it based?

Felo Inc., a company established in Tokyo in July 2024. It raised roughly 1.5 billion yen, about 10 million dollars, in a Series A led by Peak XV Partners with Mirae Asset's venture arm participating, announced in July 2025.

Q6. Is my data used to train AI models?

The policy is specific about Google user data — it will not be shared with third parties, naming OpenAI and ChatGPT explicitly, and will not be used to train general or non-personalized models. For other user content, the policy states only that information may be used to improve services and conduct research, without an equivalent explicit exclusion. Treat sensitive queries accordingly. Separately, the policy does set out a full set of user rights covering access, deletion, correction, portability, restriction of processing, withdrawal of consent and objection, exercisable through your account or by writing to the published support address.

Q7. What happens if I stop using my free account?

The privacy policy states that if a free account goes 30 days without a login, associated data including user content and files may be deleted. If you want to preserve work, export it or log in periodically.

Q8. Can it create presentations and documents, not just answer questions?

Yes — that is the second half of the product. AI Slide generates decks with template support and file export, AI Page builds interactive web pages, and LiveDoc provides documents and mind maps. All of these are credit-metered.

Q9. Is there a mobile app?

Yes. Felo - Search the world is the official iOS app from Felo Inc., free, rated 4+, requiring iOS 18.0 or later. Its user rating was 3.38 out of 5 from 21 ratings when checked — a small sample worth weighing accordingly.

Q10. How does it compare with Perplexity?

Perplexity is larger, more established and English-first, focused on search and answers. Felo is smaller and younger, differentiates on Asian-language retrieval, and bundles a creation suite that turns findings into slides, pages and mind maps — at the cost of a credit meter on that half of the product.