Y Combinator-backed Andi uses AI to build a better search engine

It’s hard to convince users to switch search engines. This is one of the reasons why public search startups rarely succeed. Another is that it is expensive to index a huge number of websites (Google has appreciated tens of billions of indexed pages), but one Y Combinator-backed company, Me toounfazed – moving forward to build an AI assistant that provides answers instead of links when searching online.

Andi was founded by Angela Hoover, who signed up for YC A startup school after dropping out of college and got into YC winter 2022 batch. After working overseas in construction and with Microsoft as a data center project administrator, Hoover met Andy’s co-founder, Jed White, at the Denver airport after her return to the US

Hoover and White—who have backgrounds in AI and search, specifically content quality ranking, queries, and classification—talked about how bad web search has gotten for things like travel, and what it will take for to create a new type of search engine from scratch.

“Gen-Z hates Google. For us, the 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 doesn’t assume – Google executives have admitted as much as “I hear my friends constantly say that Google sucks. Search results are cluttered with ads, SEO spam, and clutter. Gen-Z is so desperate for an alternative that we use TikTok as a search engine. We hate invasive creepy ads and how Google is Big Brother and watches over everything.

Andy search

Image Credits: Me too

Hoover offers Andi’s AI-powered assistant as an alternative. The general-purpose system attempts to find and extract answers to questions by combining large language models similar to OpenAI’s GPT-3 with live web data.

Behind the scenes, Andy pulls information from web results ranked by the question asked, as well as overall quality (although it’s not clear how Andy defines “quality”). Depending on the subject, the platform uses different AI systems tailored to specific verticals (e.g. factual knowledge, programming or consumer health) and language models that generate answers by combining knowledge from multiple sources (e.g. Wolfram Alpha, Forbes, The New York Times, etc.).

It’s a step up from Google’s presented fragmentswhich extract text from web pages to answer frequently asked questions, and closer to so-called “cognitive search” engines such as Amazon Kendra and Microsoft SharePoint Syntex, which draw on knowledge bases to bring answers together. Startups like HebbiaKagi and You.com also use AI to return specific content from the web in response to queries, as opposed to direct results lists.

So what sets Andy apart? Unlike some of its competitors, Hoover claims it does not charge for its service or record personal information. Andi also does not log or store searches or the results people read or click on, and only uses coarse location data to improve the relevance of search results.

“Even when we add the option of user accounts in the future, we will collect and store only enough 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 we provide,” Hoover said. “Users are telling us that Andi can save them 15 or 20 minutes of searching, and they’re asking us to let them use it with their own team and personal data… As we improve the question-answering technology and add support for connecting to private data sources, we think this has huge potential.”

Andy search

Image Credits: Me too

To filter out information that might be misleading — or patently false — Hoover says Andy uses techniques including blacklists and ranking metrics. Misinformation is a growing problem, of course – and Google itself has one was struggling pp. But Hoover expressed confidence in the technical steps Andy has taken to mitigate the impact.

“Every other new search startup is making another clone of Google with the same cluttered page of blue links targeting a web browser, with more or less variation in ads and privacy practices,” she said. “The content you see in [Andi’s] results are pulled from the live source where possible, not from a stale index. The answers to the questions are improving rapidly and in many areas are already excellent.”

In a quick experiment, I submitted a few conflicting queries to 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?” returned “Joe Biden,” while the question “Are the COVID-19 vaccines fake?” pulled up a Forbes article debunking conspiracy theories about the pandemic.

Andi is still very much in alpha and intends to stay lean as it iterates based on feedback from early users, Hoover says. The startup will have to make tough decisions. As a New Yorker notes, search algorithms are susceptible to various biases, such as prioritizing only websites that use modern web technologies. They also open the door for bad actors. In 2020, Pinterest take advantage of a quirk of Google’s image search algorithm to bring up more of its content in Google Image Search.

While grappling with these issues, Andy’s team continues to explore its business model. While the core service will remain free, Hoover says Andy will eventually offer paid professional and business plans with premium features and API access, allowing customers to use Andy’s search and answer capabilities with paid content, personal data and internal company and team data.

Andy search

Image Credits: Me too

Paid features are probably the way to go, given that Google’s share of the global search market has held steady at more than 90% for most of the past decade. Bing trailed with 3.4%, followed by Yahoo! (full disclosure: parent company of TechCrunch) at 1.34%, According to to Statcounter.

To fund the development of these features and potential partnerships with alternative search engines, Andi recently raised $2.5 million, which includes backing from YC, Gaingels, GoodWater Capital, K20 Fund, Acacia Venture Capital Partners, Fepo Capital and BBQ Capital, as well as some family and friends.

“We’ve kept our burn rate low by working as digital nomads outside of Mexico to expand our runway, and we’re staying lean. Even after we add AI developers and increase our model training costs, we have well over two years of runway,” Hoover said. “We are using the funds to enhance our proprietary generative AI for complex question answers and the ‘vertical search and API’ technology that Andi uses to combine large language models with live data, particularly: developing and training AI models, adding some more AI developers on our team and hosting and inference costs as we start to scale usage once we get closer to product market… At this early stage we’re focused on creating really great search that our users love before we generate revenue.”

Andi doesn’t collect detailed metrics, but Hoover estimates that the search engine currently has about 5,000 users. Andi plans to add a full-time AI developer before the end of the year, bringing the total to three employees, including Hoover and White.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *