It’s difficult to convince users to switch search engines. That’s one reason why public search service startups rarely succeed. Another is that it’s expensive to index a huge number of websites (Google has an estimated tens of billions of pages indexed), but one Y Combinator-backed company, Andi, is undeterred — forging ahead to build an AI assistant that provides answers instead of links when searching online.
Andi was founded by Angela Hoover, who registered for YC’s Startup School after dropping out of college and got into YC’s Winter 2022 Batch. After working overseas in construction and with Microsoft as a data center project administrator, Hoover met Andi’s co-founder, Jed White, at the Denver airport upon her return to the U.S.
Hoover and White — who had a background in AI and search, specifically content quality ranking, querying and classification — talked about how bad web search had become for things like travel and what it would take to build a new type of search engine from scratch.
“Gen-Z hates Google. To us, search is broken. We live on our phones in messaging apps with visual feeds like TikTok and Instagram,” Hoover told TechCrunch in an email interview. She’s not conjecturing — Google execs have admitted as much. “I hear my friends say constantly that Google sucks. Search results are overwhelmed with ads, SEO spam and clutter. Gen-Z is so desperate for an alternative that we’re using TikTok as a search engine. We hate the invasive creepy ads, and how Google is Big Brother and surveils everything.”
Hoover offers Andi’s AI-powered assistant as an alternative. The general-purpose system attempts to find and extract answers to questions, combining large language models akin to OpenAI’s GPT-3 with live web data.
Behind the scenes, Andi extracts information from web results ranked for relevance to the question being asked as well as overall quality (although it’s not clear how Andi defines “quality”). Depending on the subject matter, the platform uses different AI systems tailored for specific verticals (e.g. factual knowledge, programming or consumer health) and language models that generate answers by combining knowledge across multiple sources (e.g. Wolfram Alpha, Forbes, The New York Times, etc.).
It’s one step beyond Google’s featured snippets, which pull text from webpages to answer commonly asked questions, and closer to so-called “cognitive search” engines such as Amazon Kendra and Microsoft SharePoint Syntex that draw on knowledge bases to cobble together answers. Startups like Hebbia, Kagi and You.com also leverage AI to return specific content from the web in response to queries as opposed to straightforward lists of results.
So what sets Andi apart? Unlike some of its competitors, Hoover claims it doesn’t charge for its service nor record personally identifiable information. Andi also doesn’t log and store searches or the results people read or click on, only using coarse location data to improve the relevance of search results.
“Even when we add the option for user accounts in future, we will only collect and retain sufficient data to help our customers use the service effectively, when they want to create an account or be remembered between devices and sessions, and to improve the service we provide,” Hoover said. “Users tell us that Andi can save them 15 or 20 minutes searching, and have been asking us to let them use it with their own team and personal data … As we improve the question answering tech and add support to connect to private data sources, we think this has massive potential.”
To filter out info that might be misleading — or patently untrue — Hoover says that Andi uses techniques including blocklists and ranking metrics. Misinformation is an evolving issue, of course — one Google itself has struggled with. But Hoover expressed confidence in the technical steps Andi has taken to mitigate the impact.
“Every other new search startup out there is making yet another weaker copy of Google with the same cluttered page of blue links targeted at a web browser, with more or less variation on ads and privacy practices,” she averred. “The content you see in [Andi’s] results is retrieved from the source live wherever possible, rather than from a stale index. The question answering is improving rapidly, and in many areas is already excellent.”
In a quick experiment, I fed a couple of controversial queries into Andi and found that the search engine handled them quite adeptly, consistently pointing to factual sources. A search for “Who really won the 2020 election?” yielded the answer “Joe Biden,” while the query “Are COVID-19 vaccines fake?” pulled an article from Forbes debunking pandemic conspiracy theories.
Andi is still very much in alpha and intends to stay lean while it iterates based on feedback from early users, Hoover says. The startup will have tough decisions to make. As a New Yorker notes, search algorithms are susceptible to various biases, for example only prioritizing websites that use modern web technologies. They also open the door to bad actors. In 2020, Pinterest took advantage of a quirk of Google’s image search algorithm to surface more of its content in Google Image searches.
As it wrestles with these issues, Andi’s team continues to suss out its business model. While the core service will remain free, Hoover says that Andi will eventually offer paid pro and business plans with premium features and API access, letting customers use Andi’s search and question-answering capabilities with paid content, personal data, and internal company and team data.
Paid features are probably the right route to go, considering Google’s share of the global search market has held steady at more than 90% for most of the past decade. Bing trails with 3.4%, followed by Yahoo! (full disclosure: TechCrunch’s parent company) at 1.34%, according to Statcounter.
To fund the development of these features and potential partnerships with alternative search engines, Andi recently raised $2.5 million, which including backing from YC, Gaingels, GoodWater Capital, K20 Fund, Acacia Venture Capital Partners, Fepo Capital and BBQ Capital, as well as a small family and friends round.
“We’ve kept our burn rate low, working as digital nomads out of Mexico to extend our runway, and staying frugal. Even after we add AI developers and increase our model training costs, we have well over two years of runway,” Hoover said. “We’re using the funds to improve our proprietary generative AI for complex question answering, and the ‘search of vertical searches and APIs’ tech that Andi uses to combine large language models with live data, especially: AI model development and training, adding some more AI developers to our team and hosting and inference costs as we start to grow usage once we get closer to product market fit … At this early stage, we’re focused on making really great search that our users love, ahead of generating revenue.”
Andi doesn’t collect detailed metrics, but Hoover estimates that the search engine has around 5,000 users at present. Andi plans to add a full-time AI developer before the end of the year, which would bring its total headcount to three, including Hoover and White.
Y Combinator-backed Andi taps AI to build a better search engine by Kyle Wiggers originally published on TechCrunch