HotBot today is not the HotBot you remember
hotbot.com is live and actively developed as of 19 August 2026 — and it is not a search engine. The site presents itself as "AI made simple", offering free access to what it describes as more than 25 AI models, naming ChatGPT, Claude and Gemini among them, reached through official APIs and without creating an account. There is an accompanying iOS app. This is an AI chat gateway wearing a 1990s search brand.
There is no continuity of ownership, technology or personnel with the search engine that launched in 1996. The current operator is Hotbot Services LLC, whose terms of service specify the laws of Ontario, Canada and the exclusive jurisdiction of Ontario courts, with a footer copyright reading 2024–2026. Lycos sold the domain in October 2016; the brand has been detached from any search technology ever since. Anyone arriving at hotbot.com because they remember the lurid green interface from 1997 has arrived somewhere else entirely.
That is also why there is no clean shutdown date to give. HotBot did not switch off on a particular Tuesday. It faded as Lycos wound down its search operations, and then the name was sold. Writing "HotBot shut down in year X" would be wrong twice over: no such date exists, and the site is live.
Where the answers come from now
HotBot does not produce results. It routes user prompts to third-party large language models through those providers' APIs. The models are somebody else's, the training is somebody else's, and the answers are somebody else's.
The site's terms describe the offering as "search engine services and Artificial Intelligence generated text, search, and chat services" — but no web index, no crawler and no named search partner is disclosed anywhere on the site, checked on 19 August 2026. If any web results are returned at all, the upstream provider of them is undocumented. That is the basis for the Resold results verdict at the top of this page: whatever HotBot serves, it did not gather.
Several things about the current operation are simply not stated, and their absence is the point:
- No revenue model. A free, no-account gateway to paid frontier models has economics that need explaining, and the site does not explain them. A previous operator was a VPN company, which makes a funnel to a paid privacy product a plausible guess — but a guess is what it is.
- No corporate address is published.
- No effective date appears on the terms of service.
- No independent audit, security review or regulatory finding on the current site was located.
None of that means the service is untrustworthy. It means there is no basis on which to verify that it is trustworthy, which for a text-forwarding intermediary is a meaningful difference. Prompts entered at hotbot.com reach OpenAI, Anthropic, Google or others under those companies' terms, mediated by an operator that has published very little about itself. Confidential text does not belong in it.
Where it is run from, and why that changed
Jurisdiction is worth a paragraph of its own here because it has moved. Earlier reporting, dating to 2023, placed the domain under a Seychelles-based VPN company that added AI-powered web search features. The current terms of service place the operating entity under Ontario, Canada law. No public announcement of that transition was located.
The change is in the user's favour on paper. Canada has a general private-sector data-protection statute, PIPEDA, which is a materially stronger baseline than a Seychelles arrangement carrying no data-protection treaty obligations to the EU or the US. But a quiet change of governing jurisdiction with no announcement, on a domain that has changed hands repeatedly, is exactly the kind of detail a reader deserves to be told rather than to discover.
Historically the service was US-based: San Francisco under HotWired, then Waltham, Massachusetts under Lycos.
The original HotBot was a shop window for Inktomi
Here is the fact most often got wrong about HotBot, and getting it the right way round explains everything else. HotBot was a brand and an interface. The engine underneath was Inktomi's. HotWired, the online division of Wired magazine, built the site; Inktomi built and ran the crawler and the index (the engine's stored copy of the web).
Inktomi had been founded in January 1996 by UC Berkeley professor Eric Brewer and graduate student Paul Gauthier, commercialising a search engine developed at the university. HotBot launched in May 1996, four months later — the commonly repeated precise day in May could not be confirmed against a primary source, so this page gives the month. The name Inktomi comes from Iktómi, the Lakota trickster spider who outwits larger opponents, which was not an accident of branding.
Inktomi's contribution was architectural rather than algorithmic, and it was the more consequential kind. Its engine ran as a coupled cluster of commodity workstations operating in parallel rather than on a single large expensive machine. That is the precise counterpoint to AltaVista, which depended on DEC's high-end AlphaServer hardware. Inktomi's bet was that a scale-out cluster of cheap machines beats a scale-up single box for web search, and it is the architecture the entire industry — Google emphatically included — subsequently adopted. HotBot was the shop window for that bet.
The consumer-facing pitch was freshness, which was the right competitive axis to pick against AltaVista. HotBot claimed to index the entire web weekly and marketed itself as the most complete web index online at 54 million documents. In March 1997 Inktomi's Smart Crawl technology was added, allowing 10 million pages to be crawled weekly. Both figures are HotBot's own marketing claims rather than independent measurements, and no audited share-of-searches figure for HotBot's peak exists.
In February 1999 Direct Hit Technologies data was layered on: click-popularity signals recording which results users actually clicked and how long they stayed. That is an early behavioural ranking signal, arriving on a mainstream engine well before such things were common.
Why it lost: it never owned the technology
HotBot's structural weakness was visible from the start. Its differentiator belonged to a supplier, and the supplier's actual business was licensing that same engine to everyone else. Inktomi's real customers became Microsoft, Yahoo and Disney — portals that acquired, on commercial terms, precisely the advantage HotBot was built on. A front end with no technology of its own has no defence when its technology is sold to its competitors.
Ownership then made things worse rather than better. Lycos acquired Wired Digital in October 1998 and with it HotBot, and ran HotBot as a second search brand alongside Lycos search without ever articulating why both should exist. By the end of 2002 HotBot had been relaunched as a multi-engine chooser: a radio button letting the user pick whether to search FAST, Google, Inktomi or Teoma. At that point it was not an engine in any sense. It was a switchboard for four other people's engines.
The endgame was quick. On 19 March 2003 Yahoo acquired Inktomi for US$241 million, or US$1.63 per share, removing HotBot's original technology supplier from the independent market entirely. Lycos relaunched HotBot in July 2011 with a robot mascot and a new design, briefly as a portal, then reverted to a simplified search page in 2012. In October 2016 Lycos sold the domain for US$155,000.
That last number is the most honest valuation available. US$155,000 is what the brand was worth once separated from the technology it never owned — set against the US$241 million Yahoo paid for the technology itself, thirteen years earlier.
Timeline
- January 1996 — Inktomi founded by Eric Brewer and Paul Gauthier at UC Berkeley.
- May 1996 — HotBot launches, built by HotWired on Inktomi's engine.
- March 1997 — Smart Crawl added; 10 million pages crawled weekly, per HotBot's own claims.
- October 1998 — Lycos acquires Wired Digital, taking HotBot, and runs it separately from Lycos search.
- February 1999 — Direct Hit click-popularity data incorporated.
- End of 2002 — relaunched as a chooser across FAST, Google, Inktomi and Teoma.
- 19 March 2003 — Yahoo acquires Inktomi for US$241 million.
- July 2011 — Lycos relaunches HotBot with a robot mascot; reverts to a plain search page in 2012.
- October 2016 — Lycos sells hotbot.com for US$155,000.
- 2023 — the domain is reported as controlled by a Seychelles-based VPN company, which adds AI-powered web search.
- 2024–2026 — operated by Hotbot Services LLC under Ontario law as a free multi-model AI chat gateway.
Who it suits, and who it does not
As a web search engine, HotBot suits nobody, because it is not one. It has no index, no crawler and no disclosed search partner. Directory listings that still categorise hotbot.com as a web search engine — and many do — are describing a product that stopped existing years before the current site was built.
As an AI chat gateway it is harder to assess, and the difficulty is itself the finding. Anyone who wants ChatGPT, Claude or Gemini can reach them directly, where the terms, the data policy and the responsible company are all known and dated. Going through an intermediary that publishes no address, no effective date and no revenue model adds a party to the transaction and subtracts information from it. The convenience on offer is real — several models behind one box, no account required — and the trade for it is not documented.
Three groups should be specific about staying away: anyone entering sensitive or confidential text, since prompts are forwarded to third parties under an undocumented arrangement; anyone who needs to know the provenance of an answer; and anyone who wants a web index, which this is not.
Common misconceptions
"HotBot had its own search index." It did not, at any point. Inktomi built and ran the crawler and index; HotWired built the site. HotBot never had a crawler user-agent of its own — the relevant robot was Inktomi's.
"HotBot was built by Lycos." Lycos bought it in October 1998, more than two years after launch, as part of acquiring Wired Digital.
"HotBot shut down." There is no shutdown date. The brand was separated from search technology when Lycos sold the domain in October 2016, and the site is live today under different ownership.
"Today's HotBot is the old one, modernised." No shared ownership, technology or people. It is a different company's AI product on a purchased domain.
"HotBot is run from the Seychelles." That was the position reported as of 2023. Its current terms specify Ontario, Canada, checked 19 August 2026.
"Inktomi is an Italian word or an acronym." It is from the Lakota trickster spider Iktómi.
"AltaVista was the technically superior engine of that era." On early raw index size, arguably. On freshness and on architecture, Inktomi and HotBot had the better answer, and the commodity-cluster design is the one every engine uses now.
Frequently asked questions
Is HotBot still around?
The domain is, but the search engine is not. As of 19 August 2026, hotbot.com is live and actively developed as an AI chat gateway offering free access to third-party models including ChatGPT, Claude and Gemini, run by Hotbot Services LLC under Ontario, Canada law. It has no crawler and no index. The 1996 search engine and today's site share only a name.
Did HotBot have its own search index?
No, and this is the fact most often got wrong about it. HotBot was a brand and an interface built by HotWired, the online arm of Wired magazine. The crawler and index belonged to Inktomi, a UC Berkeley spin-out founded in January 1996. HotBot was effectively Inktomi's shop window, and Inktomi went on to license the same engine to Microsoft, Yahoo and Disney.
When did HotBot shut down?
It never did, on any identifiable date. HotBot faded as a search engine gradually: by the end of 2002 it had become a chooser between FAST, Google, Inktomi and Teoma rather than an engine, and Lycos sold the domain in October 2016 for US$155,000, severing the brand from any search technology. The site that exists today is a different company's product.
Who owns HotBot now?
Hotbot Services LLC, whose terms of service specify the laws of Ontario, Canada and the exclusive jurisdiction of Ontario courts, with a footer copyright covering 2024 to 2026. The chain runs HotWired and Wired Digital, then Lycos from October 1998, then a private buyer after the October 2016 domain sale, then a Seychelles-based VPN company reported in 2023.
Is HotBot a good Google alternative?
It is not an alternative to Google because it is not a web search engine. It has no index of its own and names no search partner, so there is no independent view of the web behind it. What it offers is a free front end onto third-party AI chat models. Directory listings that still file hotbot.com under web search engines are out of date.
Is HotBot safe to use?
There is no basis on which to verify that either way. The prompts you enter are forwarded to third-party model providers' APIs, so your text reaches those companies under their terms, mediated by an operator that publishes no corporate address, no effective date on its terms and no revenue model. No independent audit or regulatory finding was located. Confidential material does not belong in it.
What was Inktomi?
The search technology company behind HotBot, founded in January 1996 by Eric Brewer and Paul Gauthier at UC Berkeley and named after Iktómi, the Lakota trickster spider. Its innovation was architectural: running web-scale search on a parallel cluster of commodity workstations rather than one large machine. Yahoo acquired Inktomi on 19 March 2003 for US$241 million, or US$1.63 per share.
Why was HotBot green and yellow?
Because it was built by Wired. HotWired, the magazine's online division, gave HotBot the most distinctive visual identity in 1990s search — a lurid green and yellow interface that people who used it still recall decades later. The design was a deliberate contrast with the plain text-and-links look of AltaVista and its contemporaries, and it was the brand's most durable asset.