Month: November 2022

Deepgram, a company developing voice-recognition tech for the enterprise, today raised $47 million in new funding led by Madrona Venture Group with participation from Citi Ventures and Alkeon. An extension of Deepgram’s Series B that kicked off in February 2021, led by Tiger Global, it brings the startup’s total raised to $86 million, which CEO Scott Stephenson says is being put toward R&D in areas like emotion detection, intent recognition, summarization, topic detection, translation and redaction.

“We’re pleased that Deepgram achieved its highest-ever pre- and post-money valuation, even despite the challenging market conditions,” Stephenson told TechCrunch in an email interview. (Unfortunately, he wouldn’t reveal what exactly the valuation was.) “We believe that Deepgram is in a strong position to thrive in this tougher macroeconomic environment. Deepgram’s speech AI is the core enabling technology behind many of our customers’ applications, and the demand for speech understanding grows as companies seek greater efficiency.”

Launched in 2015, Deepgram focuses on building custom voice-recognition solutions for customers such as Spotify, Auth0 and even NASA. The company’s data scientists source, create, label and evaluate speech data to produce speech-recognition models that can understand brands and jargon, capture an array of languages and accents, and adapt to challenging audio environments. For example, for NASA, Deepgram built a model to transcribe communications between Mission Control and the International Space Station.

“Audio data is one of the world’s largest untapped data sources. [But] it’s difficult to use in its audio format because audio is an unstructured data type, and, therefore, can’t be mined for insights without further processing,” Stephenson said. “Deepgram takes unstructured audio data and structures it as text and metadata at high speeds and low costs designed for enterprise scale … [W]ith Deepgram, [companies] can send all their customer audio — hundreds of thousands or millions of hours — to be transcribed and analyzed.”

Where does the audio data to train Deepgram’s models come from? Stephenson was a bit coy there, although he didn’t deny that Deepgram uses customer data to improve its systems. He was quick to point out that the company complies with GDPR and lets users request that their data be deleted at any time.

“Deepgram’s models are primarily trained on data collected or generated by our data curation experts, alongside some anonymized data submitted by our users,” Stephenson said. “Training models on real-world data is a cornerstone of our product’s quality; it’s what allows machine learning systems like ours to produce human-like results. That said, we allow our users to opt out of having their anonymized data used for training if they so choose.”

Through Deepgram’s API, companies can build the platform into their tech stacks to enable voice-based automations and customer experiences. For organizations in heavily regulated sectors, like healthcare and government, Deepgram offers an on-premises deployment option that allows customers to manage and process data locally. (Worth noting, In-Q-Tel, the CIA’s strategic investment arm, has backed Deepgram in the past.)

Deepgram — a Y Combinator graduate founded by Stephenson and Noah Shutty, a University of Michigan physics graduate — competes with a number of vendors in a speech-recognition market that could be worth $48.8 billion by 2030, according to one (optimistic?) source. Tech giants like Nuance, Cisco, Google, Microsoft and Amazon offer real-time voice transcription and captioning services, as do startups like Otter, Speechmatics, Voicera and Verbit.

The tech has hurdles to overcome. According to a 2022 report by Speechmatics, 29% of execs have observed AI bias in voice technologies — specifically imbalances in the types of voices that are understood by speech recognition. But the demand is evidently strong enough to prop up the range of vendors out there; Stephenson claims that Deepgram’s gross margins are “in line with top-performing software businesses.”

That’s in contrast to the consumer voice-recognition market, which has taken a turn for the worse as of late. Amazon’s Alexa division is reportedly on pace to lose $10 billion this year. And Google is rumored to be eyeing cuts to Google Assistant development in favor of more profitable projects.

In recent months, Stephenson says that Deepgram’s focus has been on on-the-fly language translation, sentiment analysis and split transcripts of multiway conversations. The company’s also scaling, now reaching over 300 customers and more than 15,000 users.

On the hunt for new business, Deepgram recently launched the Deepgram Startup Program, which offers $10 million in free speech-recognition credits on Deepgram’s platform to startups in education and corporate. Firms participating don’t need to pay any sort of fee and can use the funds in conjunction with existing grant, seed, incubator and accelerator benefits.

“Deepgram’s business continues to grow rapidly. As a foundational AI infrastructure company, we haven’t seen a reduction in demand for Deepgram,” Stephenson said. “In fact, we’ve watched businesses look for ways to cut costs and delegate repetitive, menial tasks to AIs — giving humans more time to pursue interesting, consequential work. Examples of this include reducing large cloud compute costs by switching big cloud transcription to Deepgram’s transcription product, or in new use cases like drive-thru ordering and triaging the first round of customer service responses.”

Deepgram currently has 146 employees distributed across offices in Ann Arbor and San Francisco. When asked about hiring plans for the rest of the year, Stephenson declined to answer — no doubt cognizant of the unpredictability of the current global economy and the optics of committing to a firm number.

Deepgram lands new cash to grow its enterprise voice-recognition business by Kyle Wiggers originally published on TechCrunch

Kenya’s first fully electric taxi service, NopeaRide, is exiting the market after its parent company EkoRent OY failed to raise additional funding to keep it afloat.

NopeaRide said EkoRent Africa, the local subsidiary of the Finnish company, has filed for insolvency in Kenya, bringing to an end the operations of the all-electric vehicle taxi player, which sought to drive a shift to environmentally-friendly transport options, while stepping-up competition for early market entrants Uber and Bolt.

“We have taken our fleet of electric vehicles off the road and have notified our staff and corporate clients. We are now working with relevant authorities to ensure that our operations are wound up in accordance with local legislation,” said NopeaRide in a statement.

“We would like to extend our deepest regret to our dedicated team of staff and drivers. We would also like to thank our loyal NopeaRide customers, corporate clients and other partners who have supported NopeaRide’s vision for electric mobility in Africa,” it said.

Juha Suojanen founded EkoRent Oy in 2014 to develop solutions based on electric vehicles, and solar energy, which later led to the 2018 launch of NopeaRide in Kenya.

NopeaRide provided the charging network and the driver and rider apps, and sourced the electric vehicles. However, the drivers were expected to arrange their own financing frameworks.

The startup grew from three vehicles to 70 by the time of closure, and had also built a charging network across Nairobi after raising undisclosed funding in 2019.

Last year, NopeaRide also received €200,000 funding from EEP Africa, a financing facility for early-stage clean energy in Southern and East Africa, to build more solar charging hubs in Nairobi, and to make it possible for the company to increase its service radius in anticipation of growth.

The startup said it was on a path to recovery this year, after its business was badly hit by the Covid pandemic, which led to a dip in the number of rides as people worked from home.

“In the first half of 2022 our traffic numbers grew to about the same level as before Covid-19. We also started to put more effort in the corporate segment as their employees were returning to office and managed to sign contracts with a few big international companies, some of them potentially reserving the majority of available Nopea capacity,” it said.

“However, EkoRent OY went into insolvency in Finland and was unable to secure additional financing to grow the business in Nairobi to the next level.”

NopeaRide, Kenya’s first EV taxi service, shuts down by Annie Njanja originally published on TechCrunch

The UK government has completed a major revision to controversial but popular online safety legislation that’s been in the works for years — and was finally introduced to parliament earlier this year — but has beenpaused since this summerfollowing turmoil in the governing Conservative Party.

In September, new secretary of state for digital, Michelle Donelan, said the reshuffled government, under newly elected prime minister Liz Truss (who has since been replaced by another new PM, Rishi Sunak) would make certain edits to the bill before bringing it back to parliament.

The draft legislation is now due to return to the House of Commons next week when lawmakers will resume scrutiny of the wide-ranging speech regulation proposals.

The government says the changes its made to the Online Safety Bill are in response to concerns it could lead to platforms overblocking content and chilling freedom of expression online — largely focused on adult safety provisions related to so-called ‘legal but harmful’ content, which included mitigation requirements like transparency obligations but did not actually require such material to be removed.

Nonetheless the controversy and concern over this aspect of the bill has been fierce.

In a press release announcing the latest raft of tweaks, the Department for Digital, Culture, Media and Sport (DCMS) and Secretary of state for digital issues, Michelle Donelan, wrote: “Any incentives for social media firms to over-remove people’s legal online content will be taken out of the Online Safety Bill. Firms will still need to protect children and remove content that is illegal or prohibited in their terms of service, however the Bill will no longer define specific types of legal content that companies must address.

“This removes any influence future governments could have on what private companies do about legal speech on their sites, or any risk that companies are motivated to take down legitimate posts to avoid sanctions. New measures will also be added to make social media platforms more transparent and accountable to their users, as a result of amendments the Government will propose.”

“Parents and the wider public will benefit from new changes to force tech firms to publish more information about the risks their platforms pose to children so people can see what dangers sites really hold. Firms will be made to show how they enforce their user age limits to stop kids circumventing authentication methods and they will have to publish details of when the regulator Ofcom has taken action against them,” DCMS added.

Over the weekend the government revealed another, related amendment to the legislation — saying it would make encouraging self harm a criminal offence, thereby taking that type of problem content out of the ‘legal but harmful’ bucket and meaning platforms will have a legal duty to remove it.

It also recently announced measures to beef up laws against abuse of intimate imagery, including criminalizing the sharing of deepfake porn without consent, among other recent changes.

DCMS is pitching its new approach with the Online Safety Bill as providing what it frames as a “triple shield” of online protection which is most strongly focused on children but still offers measures intended to help general consumers shield themselves from a range of online harms — with social media firms legally required to 1) remove illegal content, 2) take down material in breach of their own terms of service, and 3) provide adults with greater choice over the content they see and engage with.

Provisions in the revised bill could, for example, enable adult users to opt to see a filtered feed if they wish to limit their exposure to content that may be unpleasant to them but which does not meet the bill’s higher bar of being strictly illegal.

The government has also retained measures aimed at empowering adults to be able to block anonymous trolls — via using tools that the biggest platforms will need to offer to let them control whether they can be contacted by unverified social media users.

“To make sure the Bill’s protections for adults online strike the right balance with its protections for free speech, duties relating to ‘legal but harmful’ content accessed by adults will be removed from the legislation and replaced with the consumer-friendly ‘triple shield’,” DCMS wrote. “The Bill will instead give adults greater control over online posts they may not wish to see on platforms.

“If users are likely to encounter certain types of content — such as the glorification of eating disorders, racism, anti-semitism or misogyny not meeting the criminal threshold — internet companies will have to offer adults tools to help them avoid it. These could include human moderation, blocking content flagged by other users or sensitivity and warning screens.”

Donelan mounted an aggressive defence of the changes on BBC Radio 4’s Today program this morning, claiming the government has strengthened provisions to protect children at the same time as adapting it to respond to concerns over the bill’s impact on freedom of expression for adults.

“Nothing is getting watered down or taken out when it comes to children,” she argued. “We’re adding extra in. So there is no change to children.”

Platforms will still be required to prevent children from being exposed to ‘legal but harmful’ speech, she also suggested — arguing that much of the content of greatest concern to child safety campaigners is often prohibited in platforms’ own T&Cs and the problem is they do not enforce them. The legislation will require platforms to live up to their claims, she said.

Earlier in the program, Ian Russell, the father of Molly Russell — the 14-year-old British schoolgirl who killed herself five years ago after viewing social media content promoting self-harm and suicide on algorithmically driven platforms including Instagram and Pinterest — expressed concern that the bill is being watered down, questioning the government’s late stage decision to remove the ‘legal but harmful’ duties clause.

“It’s very hard to understand that something that was important as recently as July — when the bill would have had a third reading in the Commons and [this legal but harmful content was] included in the bill, it’s very hard to understand why that suddenly can’t be there,” he told the BBC.

Discussing why he feels so strongly about risks attached to ‘legal but harmful’ content spreading online, Russell referred to the inquest into his daughter’s death which surfaced evidence from the platforms that showed she had engaged with a lot of such content — giving an example of a pencil-style drawing of a sad girl captioned with the text “who would love a suicidal girl” as one of the pieces of content she had viewed that had particularly stayed with him.

“That in and on its own isn’t necessarily harmful but when the platforms’ algorithms send hundreds if not thousands of those posts or posts like it to someone — particularly if they’re young and vulnerable — then that content had to be regulated against,” he argued. “The algorithms have to be looked into as well. And that’s what the concern is.”

Russell also accused platforms of not taking strong enough measures to prevent minors from accessing their services. “The platforms have not taken seriously the advances in age verification and age assurance that tech now has — they’ve not paid enough attention to that. They’ve sort of turned a blind eye to the age of people on their platforms,” he suggested.

While not embracing the government’s edits to ‘legal but harmful’ duties in the bill, Russell did welcome DCMS’ drive to dial up transparency obligations on platforms as a result of revisions that will require them to publish risk assessments — when previously they may have had to undertaken an assessment but would not have been required to publish it.

Asked by the BBC about Russell’s criticism of the removal of the ‘legal but harmful’ clause, Donelan said: “Content that is harmful or could hurt children but is not illegal — so is legal — will still be removed under this version of the bill. So the content that Molly Russell saw will not be allowed as a result of this bill. And there will no longer be cases like that coming forward because we’re preventing that from happening.”

She also argued the revised bill would force platforms to enforce their own age restrictions — such as by making them explain how they are stopping minors from accessing their services.

“We’ve strengthened the bill,” she reiterated. “We’ve now introduced clauses where companies can’t just say yes we only allow children over 13 to join our platform — then they allow ten year olds and actively promote it to them. We’re stopping that from happening — we’re saying no, you’ve got to enforce that age restriction, you’ve got to tell parents how you’re doing that and everybody else. We’re saying you’ve got to work to the regulator with the children’s commissioner when you’re producing the guidelines and putting them in practice.”

Asked how the government can be sure platforms will really ban underage users, Donelan pointed to what she described as the “very punitive sanctions” still in the bill — including fines of up to 10% of global annual turnover, adding: “If a company breaches any aspect of the bill, including for children, they could face fines… [as large as] billions of pounds. That’s a really big incentive not to breach the bill.”

She said the government has also strengthened this aspect of the bill — saying companies “do have to be assured of the age of their users”.

“Now we’re not saying you have to use ‘X specific tech’ because it will be out of date by next week — this bill has to last the test of time — what we are saying is you could use a range of age assurance technology or age verification technology but whatever you do you’ve got to make sure you know the age of these users to know whether they’re 14 or whether they’re 45 — so you know the protection have got to be in place and I think that’s the right approach.”

This component of the bill is likely to continue to face fierce opposition from digital rights campaigners who are already warning that biased AIs will likely be the tech that gets applied at scale to predict users’ age as platforms seek to meet compliance requirements — and that the legislation therefore risks automating discriminatory outcomes…

Another notable revision to the bill the government confirmed today is the removal of a “harmful communications” offence that free speech campaigners had warned risked having a major speech chilling effect based on a disproportionate weighting on someone taking offence to public speech.

Offences on false and threatening comms have been retained.

“To retain protections for victims of abuse, the government will no longer repeal elements of the Malicious Communications Act and Section 127 of the Communications Act offences, which means the criminal law will continue to protect people from harmful communications, including racist, sexist and misogynistic abuse,” DCMS further notes.

There will also be a requirement for major platforms not to remove content that does not breach the law or suspend or ban users where there has not been a breach of their ToS — as another measure the government claims will help bolster freedom of expression online.

Further amendments are aimed at dialling up protections for women and girls online, with the government saying it will add the criminal offence of controlling or coercive behaviour to the list of priority offences in the Bill.

“This means platforms will have to take proactive steps, such as putting in measures to allow users to manage who can interact with them or their content, instead of only responding when this illegal content is flagged to them through complaints,” per DCMS.

Another change recognizes the Children’s Commissioner to the face of the bill as a “statutory consultee” to the regulator, Ofcom’s, codes of practice, which platforms will be required to cleave to as they seek to demonstrate compliance — casting a key child safety advocate in a core role shaping compliance recommendations.

The government has tabled some of the slew of latest amendments to the Bill in the Commons for Report Stage on December 5, when it returns to parliament — but notes that further amendments will be made at later stages of the Bill’s passage.

Commenting in a statement, Donelan added:

“Unregulated social media has damaged our children for too long and it must end.

I will bring a strengthened Online Safety Bill back to Parliament which will allow parents to see and act on the dangers sites pose to young people. It is also freed from any threat that tech firms or future governments could use the laws as a licence to censor legitimate views.

Young people will be safeguarded, criminality stamped out and adults given control over what they see and engage with online. We now have a binary choice: to get these measures into law and improve things or squabble in the status quo and leave more young lives at risk.”

UK confirms removal of Online Safety Bill’s ‘legal but harmful’ clause by Natasha Lomas originally published on TechCrunch

Earlier today, renowned VC Bill Gurley put together a list of the many “red flags” that VCs should have paid closer attention to when funding FTX, suggesting in a tweet that this summary of warning signs might help keep VCs “out of the investor hurt locker” going forward. Gurley includes such no-nos as “unique financial data presentations,” “aversion to audits,” “large secondary transactions,” and “lack of a legitimate board.”

Yet publishing them now is a little like shouting “fire!” after everyone is already outside the theater, watching its smoldering remains dissolve into the parking lot. Most of the behaviors that Gurley identified today came to a grounding halt when the market abruptly shifted in spring, and by then, the damage was already done. More, if history has shown us anything, it will happen again and not because VCs miss red flags but because they sometimes throw these investing rules out the window.

Gurley asserts, for example, that one reason the startup market cratered was that investors “let the good times roll” (red flag #1). It’s pretty hard to argue with this one. Consider how little VCs really knew about Samuel Bankman-Fried, all while he burnished his image as the crypto industry’s wunderkind. (Weirdly, Sam Bankman-Fried’s smiling visage is still plastered around parts of San Francisco.)

Gurley also cites the “lack of a legitimate board” as a red flag (#2). This was another nod to FTX, which had no board of directors, but barely-there boards have become pervasive. In a story just tonight about Pipe, TechCrunch’s Mary Ann Azevedo writes that the three-year-old marketplace has only one outside board member who is not a cofounder of the company, and that individual has been a VC for three years. (Pipe raised more than $300 million from more than a dozen firms.)

Another issue is dual-class shares (red flag #3), which in many cases give entrepreneurs the power to ignore the wishes of investors. VCs once argued against them but long ago gave into founder demands for them, no matter how ridiculous the ask. Don’t believe us? Lyft’s founders and Snap’s founders have shares designed to keep them in control until they kick the bucket. Adam Neumann had so much control over WeWork that had he not been elbowed out, his children and grandchildren might have been in charge of the company ultimately.

And, Gurley pegs secondary sale transactions (red flag #8) as an obvious danger. Hopin, the virtual events platform, is a prime example. The three-year-old company has been dealing with shrinking market share and layoffs, yet according to a Financial Times piece from earlier this year, its founder was able to take $195 million worth of shares off the table while also retaining nearly 40% of the company and voting control. Bankman-Fried similarly took$300 million off the table last fall in a $420 million round when FTX was barely two years old.

One problem with Gurley’s indictment of his peers is that Gurley himself was complicit in some of these offenses. Remember WeWork, which promised that Adam Neumann’s progeny would rule the company for eternity? Gurley’s firm – Benchmark – had a seat on the company’s board.

The bigger issue ties to how venture firms are structured and paid. VCs can afford to push it to the limit because they know someone else — their own investors — will be around to pick up the pieces.

Unfortunately, the picture isn’t nearly so rosy for everyone else. On the contrary, the consequences of every “red flag” that was waved away is becoming more apparent with every layoff, down round, and executive change-up.

VCs had a good run, and they will again. But right now, if you don’t believe that tens — if not hundreds — of billions of dollars from pension funds, school endowments, hospital systems, and others that provide capital to VCs is about to go up in smoke, you haven’t been paying attention.

It’s not just FTX that’s going down, not by a long shot.

It’s a little late to be talking about red flags in venture investing? by Connie Loizos originally published on TechCrunch