Technology

Four years ago, mathematician Vlad Voroninski saw an opportunity to remove some of the bottlenecks in the development of autonomous vehicle technology thanks to breakthroughs in deep learning.

Now, Helm.ai, the startup he co-founded in 2016 with Tudor Achim, is coming out of stealth with an announcement that it has raised $13 million in a seed round that includes investment from A.Capital Ventures, Amplo, Binnacle Partners, Sound Ventures, Fontinalis Partners and SV Angel. More than a dozen angel investors also participated, including Berggruen Holdings founder Nicolas Berggruen, Quora co-founders Charlie Cheever and Adam D&Angelo, professional NBA player Kevin Durant, Gen. David Petraeus, Matician co-founder and CEO Navneet Dalal, Quiet Capital managing partner Lee Linden and Robinhood co-founder Vladimir Tenev, among others.

Helm.ai will put the $13 million in seed funding toward advanced engineering and R-D and hiring more employees, as well as locking in and fulfilling deals with customers.

Helm.ai is focused solely on the software. It isn&t building the compute platform or sensors that are also required in a self-driving vehicle. Instead, it is agnostic to those variables. In the most basic terms, Helm.ai is creating software that tries to understand sensor data as well as a human would, in order to be able to drive, Voroninski said.

That aim doesn&t sound different from other companies. ItHelm.aiapproach to software that is noteworthy. Autonomous vehicle developers often rely on a combination of simulation and on-road testing, along with reams of data sets that have been annotated by humans, to train and improve the so-called &brain& of the self-driving vehicle.

Helm.ai says it has developed software that can skip those steps, which expedites the timeline and reduces costs. The startup uses an unsupervised learning approach to develop software that can train neural networks without the need for large-scale fleet data, simulation or annotation.

&Therethis very long tail end and an endless sea of corner cases to go through when developing AI software for autonomous vehicles, Voroninski explained. &What really matters is the unit of efficiency of how much does it cost to solve any given corner case, and how quickly can you do it? And so thatthe part that we really innovated on.&

Voroninski first became interested in autonomous driving at UCLA, where he learned about the technology from his undergrad adviser who had participated in the DARPA Grand Challenge, a driverless car competition in the U.S. funded by the Defense Advanced Research Projects Agency. And while Voroninski turned his attention to applied mathematics for the next decade — earning a PhD in math at UC Berkeley and then joining the faculty in the MIT mathematics department — he knew he&d eventually come back to autonomous vehicles.

By 2016, Voroninski said breakthroughs in deep learning created opportunities to jump in. Voroninski left MIT and Sift Security, a cybersecurity startup later acquired by Netskope, to start Helm.ai with Achim in November 2016.

&We identified some key challenges that we felt like weren&t being addressed with the traditional approaches,& Voroninski said. &We built some prototypes early on that made us believe that we can actually take this all the way.&

Helm.ai is still a small team of about 15 people. Its business aim is to license its software for two use cases — Level 2 (and a newer term called Level 2+) advanced driver assistance systems found in passenger vehicles and Level 4 autonomous vehicle fleets.

Helm.ai does have customers, some of which have gone beyond the pilot phase, Voroninski said, adding that he couldn&t name them.

With limited prospects for growth, one of the iron laws of economic downturns is that advertising is among the first budgets to be cut.

Advertising revenues have already cratered at many alt-weekly newspapers, which heavily rely on local events and restaurants that have been shuttered in the wake of the COVID-19 outbreak. BuzzFeed even went so far (as they do) to label it a &media extinction event.&

Clearly itbad times, but I wanted to get a lot more granular around the data for ad rates, particularly around top startups. So I compiled a list of a little more than 100 unicorns across a variety of sectors and explored how the prices of their search engine keywords have changed with the global pandemic that is sparking a global recession.

The results aren&t surprising — there has been a collapse in prices for almost all ads (with some very interesting exceptions we will get to in a bit). But the variations across startups in their online ad performance says a lot about industries like food delivery and enterprise software, and also the long-term revenue performance of Google, Facebook and other digital advertising networks.

A quick overview of the data

Itcommon for startups to buy their own keywords on search engines like Google and the App Store. Owning that top rank guarantees that their own companypage is the first result a user sees and prevents competitors from buying their name, potentially intercepting customers.

Day after day, the burden of COVID-19 caused by SARS-CoV-2 mounts further. As of this writing, nearly 400,000 patients worldwide were confirmed for the disease, including over 46,000 cases spanning every state in this country.

In tandem with this mounting burden, due to numerous fumbles over the past 10 weeks, the U.S. has faced significant bottlenecks in the production of diagnostic testing and imposed substantive red tape to deter testing. As more diagnostic tests have come online, the rate of &confirmed cases& — the key performance indicator (KPI) being monitored by decision-makers across the country — continues to accelerate.

Why businesses love KPIs

Leaders of startups are familiar with the attention paid towards KPIs such as profit margins, burn rates, net dollar retention rate, and customer acquisition costs. These metrics, when chosen appropriately, allow leaders to continuously take the pulse of their companies and take action in response.

Along these lines, Phil Nadel has writtenon TechCrunch that founders &cannot hope to grow a company in any meaningful way without…KPIs…[b]ecause KPIs, if constructed correctly, give management and potential investors a cold, analytical snapshot of the state of the company, untainted by emotion or rhetoric& (emphasis added).

Conversely, when misconstructed, misconstrued or overlooked, KPIs can cause organizations to crash and burn. This phenomenon has been deemed &surrogation,& or reflecting the potential for critical thinking around strategy to be subverted by elevation or suppression of a single number. For example, maximizing profit margin can hurt the quality of goods/services, and maximizing lifetime value (LTV) can negatively impact customer experience by, say, encouraging the use of shady cross-selling or up-selling techniques that are harmful in the long-term to customers (aswas the caseat Wells Fargo after the financial crisis).

Taken too far, wrongly-selected KPIs can cause organizations to suffer profoundly, such as with Uberpatchy quality standards for driver recruitment leading to innumerable controversies around customer experience. And while consistently poor quarterly failures in the boardroom can lead to organization crises, they do not bear remotely the same stakes as public health crises.

The primary KPI for COVID-19 has been an unreliable figure in the U.S.

During the COVID-19 pandemic, the primary KPI tracked by U.S. leaders has been the number of confirmed cases of the disease. Yet epidemiologists monitoring the outbreaks have become increasingly frustrated with the meandering pace of actions taken by U.S. decision-makers. The core of the mismatch between actions desired by scientists and those (not) taken by policymakers lies in misinterpreting the &denominator.&In other words, misconstruing the pandemicprimary KPI.

The denominator refers to the formula from which mortality rates are calculated — the number of deaths divided by the number of cases. While this seems to be straightforward algebra, the devil is in the details. Since the beginning of the outbreak, the Chinese government has waffled on their definition of &confirmed cases&: deciding ultimately to go with a definition (positive only if laboratory-confirmed, regardless of symptoms or other tests) that may suppress the real number. Analogously, diagnostic testing snafus in the U.S. suggest that the number of &confirmed cases& here is simply not reliable. Both of these stand in stark contrast to countries such as Taiwan and South Korea, both of which swiftly diverted resources to scale up, broaden, and fully report testing.

Predictably, these snafus have led to diagnosis of disproportionately severe cases thus far in the U.S.: with the exception of NBA players and movie stars, only highly symptomatic individuals have been eligible to receive testing (due to clinical red tape), able to receive testing (due to shortages), and subsequently confirmed. This selection bias of severe cases implies that overall case number — the true denominator — is markedly underestimated. Accordingly, since the beginning of February, epidemiologists have been vocal that &simple counts of the number of confirmed cases can be misleading indicators of the epidemictrajectory.&

Relying on confirmed cases as the primary KPI may have delayed and misled critical action

A timeline of COVID-19progression in the U.S. illustrates the impact of using confirmed cases as the countryprimary KPI for public health response.

On January 22, President Trump stated that &we have [SARS-CoV-2] totally under control, itone person [confirmed].& On February 25, with 53 confirmed, he claimed that &the coronavirus…is very well under control in our country. We have very few people with it.& The following day, the president declared that &the risk to the American people remains very low.& On March 6, with 227 confirmed, he shared that &I think we&ve done a tremendous job of keeping [the number of confirmed cases] down.& And just one week ago on March 17, the president praised West Virginia as the single state without any cases, hailing &Big Jim, the governor…must be doing a good job.&

By the end of that day, the praise no longer held. Reports later emerged that the reason West Virginia had no cases for so long was related to the fact that the state was ill-equipped, and perhaps resistant, to performing testing that might come back positive.

With confirmed cases on U.S. soil as the North Star for decision-making — despite &the system blinking red& on other dimensions — early decisions to take vital preventive actions were punted. It wasn&t until March 13 (over seven weeks after the first confirmed U.S. case) that the president declared a national emergency. And even as 70 million Americans were under lockdown as of March 20, 47 states maintained few restrictions for social distancing (which remains the only intervention currently with proven effectiveness against COVID-19). Those under lockdown permit leaving for &essential& tasks, which itself is fuzzily-defined.

Moreover, emerging evidence suggests that mild (&subclinical&) cases — exactly the ones that have not been tested — may be driving community spread of the virus. Nonetheless, focus on confirmed cases as principal KPI has elicited few interventions against these &below the surface& cases. Domestic travel restrictions by foot, bike, car, bus, train, and plane remain mostly limited. Spring breakers continue to party at Miami Beachhookah shops and Nashvillehonky-tonks.

Simultaneously, numerous universities — sensitive to the fact that college dormitories are amongst the highest risk settings for contracting an infection — suspended classes and evacuated their students. However, these moves overlook the fact that college students are amongst the least likely to manifest symptoms. As such, by seeking to prevent any confirmed cases on their campuses, these universities may have ushered the virus into the homes of parents and grandparents in innumerable local communities across the U.S.

Alternative KPIs for COVID-19 can help get the U.S. back on track

If the wrong KPI created this mess, then the right KPI(s) are urgently needed to begin fixing it.

For starters, risk stratification is desperately required. Rather than resorting to black-and-white measures of infected versus uninfected, risk spectra can better characterize the threat faced by specific individuals, communities, states, and countries.

For example, mortality risk can be quantified and monitored. Elderly individuals and those with pre-existing medical conditions are at the highest risk of severe infection. By deriving and pooling individual risks, the relative threat to communities can be evaluated: encouraging high-risk communities to take more immediate and more proactive preventive action. Florida, for instance, could benefit from this kind of measurement: 27% of the stateresidents population are elderly, but thus far shockingly few individuals have been tested for coronavirus in retirement communities like the Florida Keys (where 75 people had been assessed as of March 24 amidst the height of Spring Break).This could lead to protective policies like visitation restrictions and hygiene guidelines.

Additionally, transmission risk can be a useful KPI. Certain demographic groups, like students and healthcare workers, are at much higher risk of propagating the virus — with or without the presence of symptoms. Individuals with exposure to the virus in confined spaces (such as cruise ships or airplanes) also have much higher transmission risk. Scoring metrics could be adapted from other infectious diseases to help policymakers better visualize and prevent transmission in their communities. This could encourage prospective procedures like contact tracing and symptom monitoring.

Designating mortality risk (for vulnerable patients) and transmission risk (to vulnerable patients) as KPIs could provide much more granularity to decision-makers than can confirmed cases alone. Including these collectively on a dashboard will generate insights and stimulate further actions than relying on a single narrow, fallible KPI.

And of course, these are only two (reductive) examples of potentially useful metrics for public health responders. Far more diverse forms of metrics are possible. These are often pioneered, unsurprisingly, by startups — such as in the case of Kinsa Health, a producer of smart thermometers. The companytemperature geo-maps are already providing a leading COVID-19 indicator for local decisionmakers, as they have previously for the seasonal flu (compared to CDC models).

In healthcare and beyond, new KPIs will prove critical moving forward

As far as COVID-19 goes, the steps our country needs to take are fairly clear. An analysisby researchers at Imperial College LondonCOVID-19 Response Team suggested that near-universal social distancing measures are the only tool available to prevent &hundreds of thousands of deaths and health systems (most notably intensive care units) being overwhelmed many times over.& Countries that continue to be ravaged by the virus (such as Iran and Italy) enacted social distancing interventions excessively late. As the trajectory of the infection curve in the U.S. closely mirrors these countries (despite probable underestimates of case burden), more universal measures are sorely needed. Better healthcare KPIs can help us understand the next steps we need to take.

As far as the startup community goes, there is much to learn from this situation. A proverb in medicine advises: &during a cardiac arrest, the first procedure is to take your own pulse.& Since startups may face numerous crises (in the business sense) amidst the economic convulsions of COVID-19, composure will be essential to guide decisions under pressure. Meaningful KPIs are the stethoscopes, tourniquets, and barometers that enable startups to take their own pulse—and to rapidly commence any resuscitation that may be required.

Another medical proverb dictates that &an ounce of prevention is worth a pound of cure.& Prescient leaders of startups would be well-served by seeking the tools for prevention sooner rather than later.

The world is changing fast. With the spread of the coronavirus keeping folks in their homes, thereno time like the present to connect with people online.

Tomorrow at 3pm EDT, betaworks& John Borthwick and Matt Hartman will be joining us for a live conference call via Zoom (herethe dial-in link) for everyone on TechCrunch. Folks will be able to join and get characteristically insightful answers from these two tech scene veterans.

John Borthwick is co-founder and CEO of betaworks, the startup studio that has fostered companies like Digg, Dots, Giphy, Chartbeat, TweetDeck, Gimlet Media, Bitly and Quibb. Borthwick is one of New Yorkmost prestigious investors and, on a personal note, a delightful conversationalist.

Matt Hartman is betaworks& first partner, leading the companyexternal investment arm. Heinvested in companies like Giphy, Anchor, Gimlet Media, Twitter, Venmo, Kickstarter and Tumblr.

Both Borthwick and Hartman are experts in the field of new media, among other things, and will have some interesting insights into the changing landscape of media, both broadly and within the scope of the coronavirus.

We&ll be talking to them about how they&re advising their portfolio companies during this tumultuous time, which of the coronavirus-caused evolutions are here to stay and which are temporary, and what tech they&re most interested in right now.

Folks who have questions should come prepared. I&m going to try to prioritize audience questions over my own, so jot &em down now!

Join us tomorrow (Thursday, March 26) at 3pm EDT/noon PDT using this Zoom link.If you can&t make it to the call, we&ll publish the transcript on Extra Crunch shortly after the call.See you there!

Earlier this week, the U.S. Food and Drug Administration (FDA) announced that it would be updating its guidance to allow self-swab tests for COVID-19, in which a patient collects a sample from their own nose for a health professional to test. On Wednesday, UnitedHealth Group revealed the results of a peer-reviewed large-scale study that provided the science behind the decision to switch to the less-invasive sample collection method.

The self-swab process doesn&t change where FDA-approved testing can happen — this expanded guidance only applies to the method of collection, meaning at-home swab-based PCR tests that many startups had hoped to bring to market are still on hold. But even though people still have to go to either clinics or drive-through testing sites to get a COVID-19 test done, the ability to self-swab offers more comfort, as well as real advantages when it comes to the health and safety of the clinicians and front-line healthcare workers staffing the sites.

This new study shows that not only does self-swabbing lessen the chance of someone with COVID-19 passing on their infection to a healthcare worker, italso just as effective as a test where clinicians collected the sample from much deeper inside a personnasal cavity. UnitedHeatlh worked with the Bill - Melinda Gates Foundation, as well as Quest Diagnostics and the University of Washington to conduct the study, which covered almost 500 patients who received tests at OptumCare diagnostic facilities in the state of Washington.

There are other benefits to the self-swab method as well, including eliminating the need for specifically trained medical professionals who have to administer the tests at point-of-care. This should help with clearing up backlogs owing to staffing, at least, though supplies and bottlenecks due to demand are going to persist as more people seek diagnosis.

Victoria Stafford, a third-year student at UC Berkeley, was set to begin working at Yelp in June as a sales intern — the only internship she applied to. And then it was canceled because of the COVID-19 pandemic.

&When I first read the cancellation email, I didn&t believe it. I refreshed my inbox; I rubbed my eyes as if I were waking up from a dream. It was clear that COVID-19 was becoming a mounting concern, but it never occurred to me that my internship was in jeopardy,& Stafford said.

Internship cancellations hurt more than just summer plans. The programs are often pipelines into future jobs and access to valuable work experience.

For Stafford, a business and domestic environmental major from a small town in rural Utah, there are very few business and policy-related opportunities.

&I ask that employers do everything they can to make their internship opportunities more accessible in these upcoming months, and come next year and the year after, show understanding and compassion for employment gaps,& she said.

Dozens of other students from across the country flooded my inbox, sharing stories about the impact on internship cancellations on their paths toward employment.

One student turned down offers and interviews from Google, JP Morgan and Goldman Sachs to pursue a software engineering position. The offer they accepted was yanked weeks later. Another student lost their chance at a post-graduate job at their dream company because their offer was revoked. One only had an offer in their hands for three weeks before it was rescinded.

A number of companies across the country, including Glassdoor, StubHub, Funding Circle, Yelp, Checkr and even the National Institutes of Health, have canceled their internship programs due to COVID-19, TechCrunch has learned. The cancellations, which will likely increase in the days and weeks to come, are unsurprising, due to the uncertainty the pandemic has caused. Still, fewer internships jeopardize the postgraduate job prospects for thousands of college students, and, beyond that, limit the talent pipeline on which tech companies so often are dependent.

Thereeven a Twitter account that tracks the status of 2020 internships.

From Denver to San Francisco

Like the concerts, conferences, universities and schools, these cancellations are because of the COVID-19 pandemic ravaging the world right now. While some companies cited health concerns, others pointed to the uncertain economic landscape.

In a statement, job search and review platform Glassdoor said the rapid spread of COVID-19 has grown &beyond a health concern into an economic one.& As a result, it has &decided to pause hiring and reprioritize some initiatives internally to ensure we are well positioned for both the downturn and recovery.&

A Funding Circle spokesperson confirmed that the company halted its internship program, &given the travel and relocation& for the upcoming intern cohort to San Francisco. In an email obtained by TechCrunch, the National Institute of Health canceled its prestigious internship — which has a 20% acceptance rate — to &stop community spread of Sars-Cov-2 through social distancing.&

&Therefore, hosting 1000+ early career scientists who deserve close supervision and intense mentoring is not appropriate at this time,& the email reads. &The cancellation of the NIH SIP applies to all students, whether you were planning to volunteer or were offered a fellowship position. It also applies, even if you were planning to do computational work that could be done remotely.&

In a statement to TechCrunch, NIH said its program has been reduced to &maintenance-only and mission critical (including research on COVID-19) operation due to spread of the novel coronavirus.&

&Regrettably, as part of this effort to keep people safe and limit the spread through social/physical distancing, it has been necessary to cancel the Summer Internship Program for young trainees at NIH for 2020, but those students already selected for the program will be given priority for summer internship positions in 2021.&

Checkr, based in Denver and San Francisco, put its summer internship program on hold due to &the challenges of onboarding interns while everyone is remote.&

Google has rescinded some internship offers for its UX design internship, per a LinkedIn post. After the publication of this article, Google announced that it is making its internship program virtual this year, but it is unclear whether or not that impacts interns who have already had their internships rescinded.

&If you log on to a laptop, you can access an opportunity&

While a number of tech companies have put their internship programs on hold, others are piecing together experimental remote internship programs for their students.

Quizlet is preparing for its annual internship program and is preparing a &contingency plan for an internship that will be virtual if necessary.& Uber has formed a dedicated team to start working on an online internship program &should the situation remain unchanged.& Lyft and Twitter, depending on the state of the pandemic, plan to onboard San Francisco interns virtually.

The pandemic has certainly put remote internship management services in high demand. That said, a handful of startups have been working on the sector for years.

San Francisco-based Symba, which helps companies offer virtual internship programs, was founded in 2017. Co-founder Ahva Sadeghi said that last week more than 100 companies and 1,000 students reached out to Symba in regards to internship cancellations because of COVID-19.

&The companies we reached out to in the beginning who said, ‘This is great but not top of mind for us,& are now calling us back asking us to jump on the phone today or tomorrow to get something implemented,& Sadeghi said in a phone call. &We thought we didn&t have product-market fit and now the conversation has completely changed.&

Sadeghi noted how internships assume a certain level of privilege in applicants, prioritizing those who can afford to move to a highly populated city with little to no pay. A remote internship, even in a time of health and prosperity, is important, she said.

&If you can log on to a laptop, you can access an opportunity,& she said. Another program, Chicago-based Sage Corps, founded in 2013, is pushing companies to sponsor the students impacted by internship cancellations. If sponsored, students can still participate in career growth development workshops virtually from Sage Corps, at $1,250 per student.

Thomas Brunskill, the founder of InsideSherpa, which helps companies host virtual internships, said heseen nearly 1,000 students a day sign up for the platform, from Northern Italy, to South-East Asia, to the United States. He started the company, which went through Y Combinator last year, to give students courses and online simulations of jobs through the comfort of their own homes.

He said his customers are mainly larger companies that employ upwards of 1,000 students, like JPMorgan Chase, Deloitte, Citi, BCG and GE.

On one end, Brunskill said the interest makes sense, as larger companies have to meet significant hiring demands. Per the National Association of Colleges and Employers, 70.4% of interns get return offers from the company where they intern.

On the other end, this concentration further showcases how smaller businesses will be impacted disproportionately from this pandemic. Many will freeze hiring altogether.

&Obviously [this] matters for students, but it also matters for companies who are now going to have this blackhole of talent,& Brunskill said. &Nobody wins in that situation — companies end up with less work-ready students who don&t really know what they&re getting into and students end up in full-time jobs that might not be aligned to their interest or skills because they never had an opportunity to test it out first.&

While layoffs are devastating, and obviously well upon us in the tech world, internship cancellations offer a harsh window into how COVID-19 doesn&t just impact our current workforce, but our future one as well.

Update 3/25/2020 3:35 p.m. PST: After asking Google on Monday to comment on internship cancellations it denied any changes. After TC published this article, however, Google provided a statement that it will moving its program to a &virtual model& this summer. The story has been updated to reflect this development.