Former tech journalist Jed (Jeremy) White is co-founder of Andi, self-described as “the next generation of search using the power of AI”. Others have called it a mashup between Google and chat-emulating challengers like ChatGPT.
White dropped out of the University of Sydney, in Australia, to become a journalist. Soon after he should have been graduating in english and computer science, he became the endlessly inventive editor of Australian Personal Computer, the magazine he had been reading since schooldays. Even in the last years of the 20th century, before the explosion of the web, he pioneered the simultaneous publishing of content in print and online. But he always wanted to be a tech entrepreneur.
After some startup struggles in Australia, White decamped to the US. With co-founder Angela Hoover, he has secured the support of Y Combinator (YC) in San Francisco, the 19-year-old tech startup accelerator and venture funder whose 5,000 startups have included Airbnb, Stripe, Reddit and Dropbox.
Andi has so far fundraised $2.5mn including $500k from YC. The co-founders own some 70% of the shares, although this will be subject to dilution by existing share pledges and future funding. YC has its standard initial 7% shareholding but this too will be subject to increases as the company fundraises. In a nod to the egregiously over-funded failures of BuzzFeed, Vice and others, the YC dictum is: “too much money will kill you”. It urges startups to stay lean and raise only the minimum they need.
The three-year-old Andi’s most recent valuation is believed to be some $50mn. But the potential for valuation – as well as for competition – may be guaged by the fact that the four-year-old You. Com (in which Salesforce founder Marc Benioff is an investor) has raised $45mn and has an estimated valuation of $215mn. The 20-month-old Perplexity (backed by Jeff Bezos) is rumoured to be valued at $1bn in a current fundraising – doubled in just four months on the back of claimed growth to more than 10mn active users.
Buckle up.
What were your earliest ambitions?
When I was a kid, we lived in a little town called Bli Bli on the Sunshine Coast in Queensland, Australia, a land of sugar cane fields and beaches. I was always more interested in silicon than sand. Thanks to my Dad, I became obsessed with computers early, and started programming when I was eight. By the time I was 12, I was selling computer games I’d written on floppy disks at hobby meetups. I was addicted to Australian Personal Computer magazine (APC – now the longest-running computer magazine in the world, and published by Future). I inhaled each issue when it came out. Instead of The Clash or Joy Division, I had a poster pinned to my teenage wall of Silicon Valley, marked out with the names of startups stretching up the Bay Area, from San Jose to San Francisco. Some names are icons now, like Apple, with a big pin in Cupertino. Most are long gone. I was hooked. I wanted to build tech startups.
How did you get into media?
I’ve always had a dual love for media as well as tech. I was editor of my high school magazine, wrote for street media and tech publications freelance (well, just free really), and studied English and Computer Science at the University of Sydney, where I was an editor and then Director of Student Publications. I computerized the student council with desktop publishing, and connected them to the Internet for the first time, before there was a Web.
I spent very little time on studies. I paid my way through uni working four part-time jobs, including doing IT admin every night. I wrote code for my startup idea. And I wrote as much as I could for the student publications. I was also acting president of the student council. I did everything but go to class.
When I was offered a journalism cadetship at Computer Publications, I jumped. It published Australian Personal Computer magazine, and had just been bought from the founder Sean Howard by the late Kerry Packer’s ACP Magazines (then Australia’s biggest publisher).
I loved APC. The idea of working on it was thrilling. I had planned to drop out of uni to do a startup like my heroes Gates and Jobs. But, instead, I dropped out to work on computer mags.
How was your journalistic career?
I didn’t plan to stay at a big company for long. Startups were the goal. But I loved Australian Personal Computer and magazines, and I thought it would be useful to learn how big media companies work. I was about to learn a first lesson about the counter-intuitiveness of startups: You don’t learn anything about startups from big companies.
When I started on PC Week, it hadn’t yet been tamed by ACP Magazines. They’d recently been moved into an almost hidden corner on Level 6 of Packer’s 54 Park Street, central Sydney office building. Luckily for me, our unit was a bit of a law unto itself and still had a scrappy startup vibe, quite separate from the big-selling women’s magazines that dominated the mighty ACP. My first desk was a packing crate. I had a kitchen chair. I had to borrow a computer to write on from a PC company. My clever and gentle and (to me) quietly terrifying editor chain smoked in the office. I loved it.
I could write. I knew tech. I loved magazines. But my plan was a startup. Mike Udabage, the managing director brought in to tame the computer mags was a 20-year ACP veteran who convinced me to stay. At the age of 23, I became Editor of APC, the magazine I’d been reading since I was 12.
Udabage saw something promising in the wild bunch of tech cowboys and girls on level 6 and, rather than taming us, he gave us enough cover with ACP management to create an “intrapreneurial” new media skunkworks, and evangelized and took the heat for us on costs with senior management.
We went all-in on the Internet early and became one of the first magazines on the web. I wrote content management systems and coded websites and search engines. We built wildly popular user forums. And we launched a tripartite model where the magazine was equal parts print, web and cover-mounted CDs. We saw that the era of licensing international content was over with the web: we built local testing labs and switched to 100% local Australian content. We tripled paid print circulation, sold record ads, and launched a fleet of online titles, eventually gaining more than 3mn users. We started helping the other magazines in ACP to move online. We even sent over a young journo as a Silicon Valley Correspondent to cover startups in detail.
But, all the while, startups called.
I’d been appointed Publishing Director of a new group called ACP Tech, but I couldn’t resist the call any longer. In the meantime, ACP had become part of PBL, a public company across TV as well as magazines. But it memorably divested its online rights to the Ninemsn joint venture with Microsoft, which was then spun-off into the listed company Ecorp. It was the first dotcom boom and the spin-off made another fortune for the Packer family – but essentially killed the publishing business’s online future. (As was also lamented by another young ACP Magazines editor of the time, Mia Freedman, of Cosmopolitan.)
What was your startup journey?
Startups are hard. The world’s most successful startup investor and accelerator, Y Combinator (YC) – which has backed a portfolio of startups worth more than US$600bn including household names like Airbnb, Doordash, Coinbase and more – says that startups are like “chewing glass, and learning to like the taste of your own blood”.
They also say that if you knew how hard startups were, you would never start one. They also have the insight that startups are counter-intuitive. All the regular rules of business that you learn in college or in a large business do not work when you’re launching a business that scales from two coders in a dorm room to a trillion dollar company in 10 years.
I did not know. I did not have Y Combinator to learn from. I had to learn the hard way.
Australia now has successful startups that have blazed the trail, like Canva and Atlassian. That has helped to start a change in the culture. People grow startups better now. And it created some venture capital and startup infrastructure in the country. These things did not exist when I left ACP and started a startup. I had no savings, no funding, and no idea.
Startups are critical to Australia’s future. We can’t go on forever digging things out of the ground and flogging them overseas with minimal value-add as the lucky country blessed by natural resources. Australia has a well-educated, technically-literate population and decent infrastructure. But, culturally, it struggles with innovation and risk. And startups are high risk. Most fail. One in 100 really work. The one pays for the 100 many times over. My friends at ACP Magazines told me I had rocks in my head.
I tried building what would now be called a Media B2B SaaS business – essentially software and machine learning for media businesses. It endured the ultimate startup failure – where a software startup turns into a zombie client services business. You start charging for performing custom services work rather than selling a pure software product.
We broke all of YC’s cardinal rules without even knowing them. Never do services or custom paid work because attention goes to where the money comes from. Tick. Focus on just one thing rather than lots of projects. Tick. Hire too fast rather than stay lean. Tick. Do fake work (like conferences or speaking) rather than just write code and talking to users. Tick. Company building rather than product building. Tick. Marketing before product market fit. Tick. Tick. Tick.
At the same time, Google and Facebook between them have laid waste to the media industry’s economic model, taking more than 90% of all digital ad revenue and leaving publishers with the scraps. I watched as our media industry clients shrivelled or went into liquidation (with huge unpaid bills), my journalist friends all lost their jobs and “content marketing” took over.
I started reading Paul Graham’s essays on Startups around 2006. He was a computer scientist and programmer who created the first SaaS startup Viaweb, which he sold to Yahoo. He used the funds to start Y Combinator as an experiment in angel investing. The very first batch, incidentally, included Reddit which was bought by Condé Nast and IPO’d just last month.
I realized I had to get to the US. It’s obvious today that YC has deep insights into what it takes to scale startups, after backing more than 4,000 founders and a string of IPOs. But I went to the wrong place (Colorado rather than the San Francisco Bay Area). And the startup network I engaged with was old school rather than YC. I also continued the same mistake of trying to bootstrap, continually returning to doing services work to pay the bills. The result was two more zombie startups with interesting AI and search tech, but no path to product market fit – the YC startup term for when customers are begging for what you’re making.
By the end of 2018, I was able to put together enough cash to leave most of the client work behind, live off ramen, and focus nearly 100% on code. I started building Orac, a content quality ranking and topic detection AI. Orac was able to detect good and bad content – including content marketing spam, misinformation, hate speech and clickbait.
I spent most of my time in Silicon Valley, and attended YC’s Startup School, which is a feeder for its investment program. It was transformational. With encouragement from YC, I launched an early alpha of Orac and got great feedback. I didn’t have product market fit, and didn’t know exactly what it could be used for. But the AI models were clearly powerful and there was something there.
How did Andi come about?
Andi happened by serendipity. I met my co-founder Angela Hoover at Denver airport. We talked about what it would take to build a new type of search engine. Angela had just returned to the US from a year in Australia where she had worked with Microsoft on Azure. She had the idea for a search engine for Gen-Z with a messaging interface and a visual feed like Instagram or TikTok. Gen-Z hates Google because it’s old and overwhelmed with SEO spam and advertising. Gen-Z spend their lives on their phones in messaging apps.
We realized that AI and the models I’d been coding offered a new way to build a search engine that was based on understanding content and the quality of sources, rather than on page rank signals that could be easily gamed by spammers with backlinks and keyword stuffing. We built an initial “minimal viable product” and took Andi (then called LazyWeb) through YC’s Startup School program. Angela won the pitch comp, and we did office hours in front of the entire Startup School community with YC Group Partner Jared Friedman. We kept them updated, and based on fast progress and enthusiastic engagement from early users, applied and were accepted into YC for their Winter 2022 batch.
YC is incredibly competitive, with more than 40k applications a year from startups, and an acceptance rate of around 1% – lower than Harvard or Stanford. They invest $500k, and put startups through an intensive three-month program that culminates in Demo Day which acts like an auction to maximize fundraising from Silicon Valley investors. They then mentor and support the company directly through all its stages to IPO and beyond.
Google is an ad tech company with a browser distribution monopoly. It steals content with minimal attribution, competes for advertisers, promotes low-quality content farms ahead of high quality media, and holds publishers to ransom. AI makes this worse.
Our mission with Andi is to un-break the Internet, fight spam and ad tech, and create a new economic model for search that heavily promotes sources and shares search revenue with publishers.
How’s it going?
Andi is still pre-product market fit, but we’ve definitely made something people love. We hit 1mn users for the first time in December, and use is growing at around 30% a month, purely by word of mouth with no paid marketing.
Our user base is technical Gen-Z – 46% under 25 years old (70% under 35). They’re highly educated (19% postgrad), high income (20%), 69% male and and 94% highly technical. More than 88% are on mobile. Like computer magazine readers, these are the people other people turn to for tech advice and help. They’re also the smart users who give us great feedback to keep Andi improving, and who evangelize us – so we’re spending our early funds on building rather than marketing.
With YC’s backing, we’ve raised $2.5mn in venture capital funding from high profile investors like Goodwater Capital, Gaingels, K20 Fund and Acacia, as well as prominent angels. We’re blessed to have some early investors and advisors from the Australian media, including some of my former bosses at PBL/ACP: Nick Chan (now COO of Andi), Tim Trumper, David Gyngell and Peter Zavecz.
We’ve been covered by the New York Times, Wall Street Journal, Forbes, Fortune, Bloomberg, TechCrunch, Fast Company and more. And, as CEO, Angela has spoken at the White House, and in the US Congress and Senate.
What’s so special about Andi?
Andi was the first search engine to use a conversational interface with AI summaries and answers, connecting AI to the Internet. But it has more than a chat UX – showing search results in a rich visual feed like Instagram or TikTok, with full branding promoting the sources. Instead of a list of blue links with truncated web text snippets from publishers, Andi gives you answers and quotes, while also letting users go deeper to explore the rich variety of great sources online.
Our competitors are 1st generation ad tech-based search like Google, full of spam and sponsored results. And 2nd generation AI answer bots like ChatGPT and Perplexity, that are “good until you use it”, rife with wrong answers, and which hide or make-up sources.
What’s your vision?
The big problem with AI and large language models (LLMs) is that they “lie like truth” – they’re not only wrong, but they are convincingly wrong. They make up facts. They make up sources. 94% of AI answers from chatbots like ChatGPT contain information that is inaccurate, misleading or incomplete. And they make it cheap and easy to produce misinformation – 40mn+ items of AI spam are being posted to the Internet a day. There is a tidal wave of AI rubbish coming. And 56% of people say they do not think they can identify when an AI is lying to them.
We are obsessively focused on accuracy and content quality – finding high-quality material from the best sources and publishers, identifying accurate information and answers fast, and heavily promoting interesting sources.
To do this, we’re building a content engine (named Trantora) that deeply understands content and sources using AI. This is how we can build a much better type of search engine that helps users to “find and explore the best of what the web has to offer”, as Search Engine Journal put it.
What’s the business model?
We plan a freemium model where searching anonymously will always be free – but without ads or tracking. We will offer paid plans with premium features, such as the ability to connect to your personal data (like email and documents) for search, question answering and text generation, or access to premium content through partnerships.
Our goal at the moment is to keep working on nailing factual accuracy and speed, and improving Andi’s AI models and content understanding week after week. We will keep iterating based on user feedback. We’re not there yet, but we have some very strong indicators that we’re approaching it – including high user scores and strong retention among “activated” users (people who use Andi for at least a week and do at least 30 searches).
The real test will be trying paid plans to see if we’ve built something people will pay for; we will test that soon.
YC describes product market fit as being like pushing a boulder up a hill. You can’t see in front of you, and you have to just keep inching upwards. Once you reach PMF, you’re chasing a boulder down the hill. That’s when things are taking off like wildfire and your servers are burning down. So the right time to switch to marketing and promotion is when that change happens – so you’re pouring fuel on to the fire. We’re somewhere on that uphill push right now.
What’s it like being a startup in San Francisco?
AI is the next major revolution in tech – as big or bigger than the rise of the PC, the Internet or the smartphone. And San Francisco is the heart of the AI revolution.
We’re based in Cerebral Valley <aka Hayes Valley, described by a local paper as “San Francisco’s nerdiest neighborhood”>. There is an intense concentration of talent, ideas and energy around AI here that is unlike anything I’ve ever seen. Every week you can hop in a self-driving car to get help from a YC group partner, or go to one of the dozens of hackathons and open source community events and co-working sessions on every day. There are a thousand opportunities to connect, learn, engage, and “make luck happen” as a startup. With interested new users. With technology partners. With investors. And with potential recruits who could become part of the Andi family as we grow, and who are excited by the promise of building new tech with AI that can help people and make a real impact.
Which companies do you most admire?
We’re in awe of Y Combinator and our group partners there. They have built high-scale startups themselves, are insanely incisive and blunt on the reality of things, while also the warmest and most encouraging people you could ever work with.
We’re fans of some of YC’s biggest hits, who we’ve been lucky to learn from. Brian Chesky and the team at Airbnb are inspiring because they are relentlessly resourceful, and absolutely obsessed with making users happy. Coinbase has been inspiring for maintaining focus through ups and downs in the crypto market.
I’m also a fan of DuckDuckGo, which pioneered caring about user privacy while building a popular search engine. And – although it’s contrarian and I wish he wasn’t distracted with right wing politics – I think Elon Musk’s first principles thinking and relentless execution with SpaceX and Tesla and Neuralink are genuinely inspiring.
What are the lessons?
I can’t take credit for these lessons. They’re all YC. If I had a time machine, I would go back to Bli Bli and give myself the YC User Manual which would (more or less) tell me:
1. Make something people want.
2. Write code, talk to users. Repeat.
3. Attention goes to where the money comes from. Don’t do services work.
4. Be relentlessly resourceful.
5. Avoid fake work. Stay lean.
6. Startups die by suicide not homicide.
7. Be nice.
