Angular Ventures Weekly
Issue #162: For the week ended October 25, 2022
Subscribe here to receive new Angular blogs, data reports, and newsletters directly to your inbox.
Will Google be disrupted?
Like many of you, I’ve been tinkering with a lot of AI-powered tools over the past few weeks. (I’m actually writing this newsletter with Lex, a GPT-3 powered document editor from the makers of Every. Can you guess which sentences were written by me or GPT-3?). In general, I’m incredibly awed and excited, but wanted to share a few quick reactions:
First, we’re clearly witnessing the greatest AI “breakout” moment since the release of Siri/Alexa in 2011. In the past few years, advanced machine learning techniques have been used to achieve increasingly impressive feats (e.g. DeepMind’s AlphaFold breakthrough in 2020). But such achievements have been hard for the average consumer to appreciate (e.g. I doubt the average person has an intuition for how hard it is to determine the 3D shape of protein strands). This is the first time in over a decade when the average consumer is being confronted with AI-powered tools doing work they understand (e.g. writing this newsletter!) well enough that the technology is indistinguishable from magic. That’s worth highlighting (and celebrating).
Second, these sort of breakout moments invite loads of founder interest and investor attention. Founders want to tinker. Investors want to make a “bet” on the space. Which means we’re about to be inundated with AI-powered tools that do a better job of highlighting the latest and greatest AI models than they do of solving real customer problems. For technologists hacking away on side projects, this is an amazing time to build (especially given that most models are open source by default). But for investors and founders trying to build long-lasting companies, this is a dangerous moment. For some actionable advice, I’d recommend this piece by Michael Dempsey from Compound.
Third, while it’s difficult to forecast how new technological capabilities will impact existing industries, my wild prediction is that Google’s core search business might finally be disrupted, for two main reasons:
- SEO spam just got super-powered. Writing assistant tools, like Jasper.ai, Playground (and many others), have made it dead simple to craft compelling-enough copy. Can Google’s SEO team keep up? Or will Google’s search results slowly devolve into a sea of AI-generated SEO-optimized spam?
- Generative AI might be able to provide a categorically better search experience than PageRank for certain types of search queries…namely, those that are looking for answers to questions. As an example, take a look at this prompt I just ran using the AI-powered writing assistant Lex:
Those results are actually…pretty good! (I Googled to make sure). Lex is just using GPT-3, not some model fine-tuned for question answering, but despite that, the experience is like the “I’m Feeling Lucky” button, except it works! I’m not sent off to some random site where I hope I’ll find the answer. The answer is served back to me directly. Why visit Google if any old application leveraging an open model like GPT-3 can provide relatively good answers to questions with factual answers? Search is being decentralized.
As you might imagine, Google AI is working on this problem as well. Flan-T5 is a new model that was fine-tuned on language tasks, making it much better at answering questions than a standard LLM. Perhaps Google can disrupt itself before somebody else does…
Who knows! But overall I just feel incredibly lucky to be here for another breakout moment in artificial intelligence. We’re on the cusp of so many amazing technological shifts and advancements, and while Angular has been a big believer in the future of machine learning and artificial intelligence for awhile (having made investments in the space consistently over the past four years), we’re always looking for more founders aiming to build generational companies in the AI realm. If that sounds like what you’re building, let us know.
Jan 25 / Lessons Learned From Investing Early in Over a Dozen SaaS Unicorns Including Salesforce, SuccessFactors, Box, Gusto, SalesLoft, ServiceMax, Veeva, Bill.com, Doximity, Yammer and Zoom Among Others
Jason Green, Founder & General Partner, Emergence Capital
Feb 15 / The Evolution of Collibra’s Product Positioning & How They Created a Category
Stan Christiaens, Co-Founder & Chief Data Citizen, Collibra
FROM THE BLOG
It’s Never too Early to Build your Growth Model
What are the specific mechanisms by which one user turns into many, and an initial investment turns into revenue?
How to Think About Revenue Quality as an Early Stage Founder
What does “quality revenue” mean when you don’t have much revenue at all?
It’s Not All About Bottoms-up
Two recent trends indicate that we may finally be past the mistaken belief that bottoms-up is the only “fundable” business model in town.
Don’t be Fooled by the PLG Mullet
How to know if you should be building a PLG Now, PLG Later or PLG Never company.
EUROPE & ISRAEL FUNDING NEWS
UK/ML Tooling. Stability AI raised $101M for its AI-based open-source software for generating music and images.
UK/Financial. Enable closed $94M for its platform that helps business-to-business (B2B) companies manage their rebate programs.
UK/Space Systems. Orbex raised $45M for its small satellite launch system.
Belgium/Biology Services. Univercells raised $43M for its biomanufacturing technology services, scaling up production for the biopharma sector.
Germany/Industrial System. Quantum Systems raised $17.5M to continue to develop, design, and manufacture advanced unmanned aerial systems.
UK/Security. OutThink closed $10M for its cybersecurity human risk management platform.
Israel/ML Tooling. TensorLeap (an Angular portfolio company) closed $5.2M for its platform that helps data scientists understand how a neural network interprets data, how it makes decisions, and where and why it failed.
The chip industry in transition. The Economist took a look at the US government’s efforts to secure domestic semi-conductor supply, concluding that we are still in the very early innings of a dramatic global battle in this key industry. “Whether or not it makes strategic sense for America to bring more chipmaking home and to hamstring its geopolitical rival with export bans, the combination of more supply and less demand is a recipe for trouble. And if America’s policies speed up China’s efforts to “resolutely win the battle in key core technologies”, as President Xi Jinping affirmed in a speech to the Communist Party congress on October 16th, they may give rise to powerful Chinese competitors. Field of dreams? Enough to keep you awake in terror at night.”
Where have all the autoshops gone? Wired reported on the decline of the local autoshop as more advanced (electronic) automotive technologies have become impossible for most autoshops to repair. “If you want to understand the rising complication in the auto repair industry, try to get a realignment on a new Audi. A car needs realignment when it’s drifting to one side or the steering wheel is vibrating, a procedure that involves adjusting the suspension, which connects a car to its wheels. A decade or so ago, that took about an hour and a half, auto repairers who spoke to WIRED say. Today, that same procedure is usually closer to three or four hours, and it can take up to nine. That’s because newer cars have advanced driver-assistance systems, which can keep cars in their lane, detect blind spots, and avoid collisions — functions that require a car to have a firm grasp of where it is in space. That requires repairers to calibrate the sensors and cameras in a car underpinning those advanced systems. Some brands of vehicle can only be calibrated with specialized and expensive tools. To start with, the equipment needed to assure a car’s wheels are in alignment costs in the $70,000 range, says Lucas Underwood, the president of L&N Performance Auto Repair in Blowing Rock, North Carolina. Then you’ll need targets, which help a car’s sensors and camera systems orient themselves. These can vary by automaker and cost around $30,000 per set.”
A consolidating IT vendor landscape. Gartner released six predictions for the near future of IT infrastructure, aimed at CIOs. Among them is a prediction of dramatic consolidation of cloud infrastructure stacks, leaving CIOs with less choice. “M&A activity in the cloud software market has been on the rise, a trend that is likely to continue, according to Gartner. As consolidation mounts, cloud customers will be left with fewer vendor choices. To prepare, CIOs should choose a default cloud vendor using ecosystem coherence as a primary partnering criteria…Leveraging open source options and joining a customer lobby are other measures that may also prove helpful, but cloud customers will have to make tough choices between best-of-breed technologies and assets that are good enough for specific use cases.”
Cybersecurity still tops the CIO agenda. Gartner also predicted that cybersecurity will continue to be a key area of focus for CIO spending. ““At packaged-food manufacturer Kellogg Co. safety and security account for roughly 15% of its total corporate information and technology spend, according to global CIO Lesley Salmon. “From an expense perspective, not a capital investment perspective, it will be going up proportionately more than any of my other areas next year,” said Ms. Salmon, who is based in the U.K. Moreover, should the need to cut expenses arise, Ms. Salmon said she leaves that part of the budget intact. “I don’t ever go to that hunting ground to save money. If I get a budget challenge, it doesn’t come out of cyber,” said Ms. Salmon.”
HOW TO STARTUP
How should startup CTOs think about cloud spend? A few great VCs, including our friend Shomik Ghosh from Boldstart, offered advice to founders on how to think about cloud spend in the early days. Shomik addressed the question of when startups should offer on-prem solutions: “The most common reason startups should consider modern on-prem is for dealing with sensitive data, which especially occurs in regulated industries (healthcare, financial services or pharma). The scope of what is considered sensitive is growing over time with regulations though, so it’s something more startups need to be aware of. A lot of ML tooling does need to be deployed across any environment, as the large enterprises keep some of this data in strictly controlled environments. In the end, startups need to meet the customer where they are — if you are designing cloud-first and dealing with customers who have sensitive data, then you should consider what your “any environment” deployment strategy would be, whether using Replicated, building your own or choosing not to work with those customers.”
Scaling SimilarWeb with PLG. Openview provided a look at how SimilarWeb scaled to $190M in ARR (!). “Similarweb’s PLG strategy starts with its enviable top-of-funnel acquisition, which Similarweb calls its “inbound machine.” In fact, the company attracted 9.4 million website visitors in July 2022 according to the company’s traffic analysis tool. Here’s how: free products built with a marketing intent. And Similarweb has a lot of them, including a traffic analysis tool, top website rankings, browser extensions, and even a free API. “We always lead with our product. We build very effective free tools for the top of the funnel,” said Maoz. “Some are web tools, some are plug-ins or Chrome extensions. These get more and more demand over the years.” Similarweb sees these free tools as both brand and SEO plays. While traditional enterprise-focused software companies normally have a “contact us” barrier, Similarweb has a “try now” option where folks can immediately engage with the product. One concrete example is the Similarweb ranking tool, which brings up the top websites, search engines, iOS apps, browsers, and more. While this is an ungated tool available completely free of charge, it’s also an effective vehicle for generating leads. Similarweb’s goal is to understand who these users are and what they’re trying to accomplish, and then route them into the right offering for that specific use case.”
HOW TO VENTURE
Lets get realistic about VC returns. Samir Kaji put out a thoughtful and wise Twitter thread about LP and GP expectations for VC returns over a long time horizon, arguing that 5x net is an unrealistically optimistic target. He walks through the math, concluding that “it’s really really hard to get a 5x, let alone a 3x consistently. It’s important to really internalize this math when either raising or allocating. Power law outcomes and the unbounded scale of companies can allow this, but this is not realistic as a base benchmark.”
Tensorleap announced a $5.2M seed round for its debugging and explainability platform.
groundcover launched Murre, an OSS metrics monitoring tool that helps developers get quick access to metrics without the need to install any 3rd party components on the cluster.
Forter’s CEO, Michael Reitblat, was interviewed by KrAsia and shared his insights on e-commerce fraud.
JFrog unveiled JFrog Advanced Security, the first DevOps-centric security solution to control the entire software supply chain.