& meet dozens of singles today!
User blogs
From building out Facebook’s first office in Austin to putting together most of Quora’s team, Bain Capital Ventures managing director Sarah Smith has done a bit of everything when it comes to hiring. At TechCrunch Early Stage, she spoke about how to ensure the critical early hires are the right ones to grow a business. As an investor at Bain Capital Ventures, Smith has a broad view into the problems that companies face as they search for the right candidate to spur organizational success.
In our conversation, Smith touched on a number of issues such as who to hire and when, when to fire, and how to ensure diversity from the earliest days.
What to consider when you first think about hiring
When a company is making its first hires — and then evolving into a bigger organization — the processes and needs may change, but the culture should be consistent from the beginning, according to Smith. From there, an emphasis on good early managers is critical.
I would really encourage you to take some time to think about what kind of company you want to make first before you go out and start interviewing people. So that really is going to be about understanding and defining your culture. And then the second thing I’d be thinking about when you’re scaling from, you know, five people up to, you know, 50 and beyond is that managers really are the key to your success as a company. It’s hard to overstate how important managers, great managers, are to the success of your company.
So we’ll talk a little bit about how to think about that, as there’s a lot of questions around helping people grow into management for the first time. You, as a founder, might be managing people for the first time, so how to think about setting up the company for success.
(Timestamp: 4:15)
How do you build culture in the new remote environment?
Facebook has removed 16,000 groups that were trading fake reviews on its platform after another intervention by the UK’s Competition and Markets Authority (CMA), the regulator said today.
The CMA has been leaning on tech giants to prevent their platforms being used as thriving marketplaces for selling fake reviews since it began investigating the issue in 2018 — pressuring both eBay and Facebook to act against fake review sellers back in 2019.
The two companies pledged to do more to tackle the insidious trade last year, after coming under further pressure from the regulator — which found that Facebook-owned Instagram was also a thriving hub of fake review trades.
The latest intervention by the CMA looks considerably more substantial than last year’s action — when Facebook removed a mere 188 groups and disabled 24 user accounts. Although it’s not clear how many accounts the tech giant has banned and/or suspended this time it has removed orders of magnitude more groups. (We’ve asked.)
Update: A spokeswoman for the CMA said the question of how many accounts have been banned/suspended in this wave of group takedowns is one for Facebook to answer, adding that the regulator has focused on the removal of groups trading misleading/fake reviews, rather than individual accounts — “as this is the most effective way of preventing the trade of such content”. “This is because banned or suspended users could create new profiles, whereas removing the group in which they are trading is more effective in disrupting and deterring this activity,” she added.
Facebook was also contacted with questions but it did not answer what we asked directly, sending us this statement instead:
“We have engaged extensively with the CMA to address this issue. Fraudulent and deceptive activity is not allowed on our platforms, including offering or trading fake reviews. Our safety and security teams are continually working to help prevent these practices.”
Since the CMA has been raising the issue of fake review trading, Facebook has been repeatedly criticised for not doing enough to clean up its platforms, plural.
Today the regulator said the social media giant has made further changes to the systems it uses for “identifying, removing and preventing the trading of fake and/or misleading reviews on its platforms to ensure it is fulfilling its previous commitments”.
It’s not clear why it’s taken Facebook well over a year — and a number of high profile interventions — to dial up action against the trade in fake reviews. But the company suggested that the resources it has available to tackle the problem had been strained as a result of the COVID-19 pandemic and associated impacts, such as home working. (Facebook’s full year revenue increased in 2020 but so too did its expenses.)
According to the CMA changes Facebook has made to its system for combating traders of fake reviews include:
- suspending or banning users who are repeatedly creating Facebook groups and Instagram profiles that promote, encourage or facilitate fake and misleading reviews
- introducing new automated processes that will improve the detection and removal of this content
- making it harder for people to use Facebook’s search tools to find fake and misleading review groups and profiles on Facebook and Instagram
- putting in place dedicated processes to make sure that these changes continue to work effectively and stop the problems from reappearing
Again it’s not clear why Facebook would not have already been suspending or banning repeat offenders — at least, not if it was actually taking good faith action to genuinely quash the problem, rather than seeing if it could get away with doing the bare minimum.
Commenting in a statement, Andrea Coscelli, chief executive of the CMA, essentially makes that point, saying: “Facebook has a duty to do all it can to stop the trading of such content on its platforms. After we intervened again, the company made significant changes — but it is disappointing it has taken them over a year to fix these issues.”
“We will continue to keep a close eye on Facebook, including its Instagram business. Should we find it is failing to honour its commitments, we will not hesitate to take further action,” Coscelli added.
A quick search on Facebook’s platform for UK groups trading in fake reviews appears to return fewer obviously dubious results than when we’ve checked in on this problem in 2019 and 2020. Although the results that were returned included a number of private groups so it was not immediately possible to verify what content is being solicited from members.
We did also find a number of Facebook groups offering Amazon reviews intended for other European markets, such as France and Spain (and in one public group aimed at Amazon Spain we found someone offering a “fee” via PayPal for a review; see below screengrab) — suggesting Facebook isn’t applying the same level of attention to tackling fake reviews that are being traded by users in markets where it’s faced fewer regulatory pokes than it has in the UK.
Screengrab: TechCrunch
Efforts to unionize Amazon’s Bessemer, Alabama warehouse were defeated by a wide margin in the second day of vote counting. More than half of the 3,215 votes cast broke in in factor of the retailer. The Retail, Wholesale and Department Store Union, which would have served as the workers’ union, had the vote passed, was quick to challenge the results.
RWDSU President Stuart Appelbaum said in a statement offered to TechCrunch,
Amazon has left no stone unturned in its efforts to gaslight its own employees. We won’t let Amazon’s lies, deception and illegal activities go unchallenged, which is why we are formally filing charges against all of the egregious and blatantly illegal actions taken by Amazon during the union vote. Amazon knew full well that unless they did everything they possibly could, even illegal activity, their workers would have continued supporting the union.
That’s why they required all their employees to attend lecture after lecture, filled with mistruths and lies, where workers had to listen to the company demand they oppose the union. That’s why they flooded the internet, the airwaves and social media with ads spreading misinformation. That’s why they brought in dozens of outsiders and union-busters to walk the floor of the warehouse. That’s why they bombarded people with signs throughout the facility and with text messages and calls at home. And that’s why they have been lying about union dues in a right to work state. Amazon’s conduct has been despicable.
This initial defeat represents a large setback in the biggest unionization push in Amazon’s 27 year history. What might have represented a sea change for both the retail giant and blue collar tech workers has, for now, been fairly soundly defeated.
Amazon has, of course, long insisted that it treats workers fairly, making such union efforts unnecessary. The company cites such standards as a $15 an hour minimum wage, a factor the company initial pushed back on, but ultimately instated after pressure from legislators.
It was a hard fought battle on both sides. A number of legislators threw their weight behind unionization efforts, in an unlikely alliance that ranged from Bernie Sanders to Marco Rubio. The conservative Florida Senator noted the company’s “uniquely malicious corporate behavior.” President Joe Biden also sided with the workers, calling himself, “the most pro-union president you’ve ever seen.”
The company will no doubt tout the results as vindication. It noted in an early statement, “[O]ur employees are smart and know the truth—starting wages of $15 or more, health care from day one, and a safe and inclusive workplace. We encourage all of our employees to vote.” We’ve reached out to the company for a statement following this morning’s news.
Among the expected challenges from the union are lingering questions around ballot boxes reportedly installed by the company in violation of labor board terms.”[E]ven though the NLRB definitively denied Amazon’s request for a drop box on the warehouse property, Amazon felt it was above the law and worked with the postal service anyway to install one,” the RWDSU writes. “They did this because it provided a clear ability to intimidate workers.”
The Bessemer warehouse, which employees around 6,000 workers, was opened at the end of March 2020, as the company looked to expand the operation of its essential workers during the impending lock down. The conversation has surface variously long standing complaints around the company’s treatment of blue collar workers, including numerous reports that employees urinate in water bottles, in order to meet stringent performance standards.
The company initially denied these claims during a social media offensive, but later clarified its stance in an apology of sorts, appearing to shift the blame to wider industry problems. The company also ran anti-union ads on its subsidiary, Twitch, before the streaming platform pulled them, stating that they “should never have been allowed to run.”
All told, 3,215 were cast, representing more than half of the workers at the Alabama warehouse.
TechCrunch is hosting a small virtual event on April 15 for startups in Detroit and we’re still looking for a few startups to pitch at the event. The deadline is today, Friday, April 9th. Apply below. Want to attend the free event? Register here.
TechCrunch just published a feature on Detroit-darling StockX and this meetup will feature those involved in producing that content. The EC-1 can be found here.
Everyone is welcome to attend the event, but we’re looking for startups based in Michigan’s southeast region to pitch at this event. TechCrunch has a long history of hosting small pitch-offs and we’re excited to revive this tradition despite the need to do it virtually.
Not in Michigan? No worries. We’re spinning up similar events in other regions too. Spoiler: Pittsburgh is next.
Qualifications
- Early-stage startups (Series A or earlier)
- Startups based in the Detroit region will be given priority
- Pitch decks are highly recommended
- Apply for the pitch-off here
The event is online and free, but space is limited. Register early. We hope you can make it.
While visual ‘no code‘ tools are helping businesses get more out of computing without the need for armies of in-house techies to configure software on behalf of other staff, access to the most powerful tech tools — at the ‘deep tech’ AI coal face — still requires some expert help (and/or costly in-house expertise).
This is where bootstrapping French startup, NLPCloud.io, is plying a trade in MLOps/AIOps — or ‘compute platform as a service’ (being as it runs the queries on its own servers) — with a focus on natural language processing (NLP), as its name suggests.
Developments in artificial intelligence have, in recent years, led to impressive advances in the field of NLP — a technology that can help businesses scale their capacity to intelligently grapple with all sorts of communications by automating tasks like Named Entity Recognition, sentiment-analysis, text classification, summarization, question answering, and Part-Of-Speech tagging, freeing up (human) staff to focus on more complex/nuanced work. (Although it’s worth emphasizing that the bulk of NLP research has focused on the English language — meaning that’s where this tech is most mature; so associated AI advances are not universally distributed.)
Production ready (pre-trained) NLP models for English are readily available ‘out of the box’. There are also dedicated open source frameworks offering help with training models. But businesses wanting to tap into NLP still need to have the DevOps resource and chops to implement NLP models.
NLPCloud.io is catering to businesses that don’t feel up to the implementation challenge themselves — offering “production-ready NLP API” with the promise of “no DevOps required”.
Its API is based on Hugging Face and spaCy open-source models. Customers can either choose to use ready-to-use pre-trained models (it selects the “best” open source models; it does not build its own); or they can upload custom models developed internally by their own data scientists — which it says is a point of differentiation vs SaaS services such as Google Natural Language (which uses Google’s ML models) or Amazon Comprehend and Monkey Learn.
NLPCloud.io says it wants to democratize NLP by helping developers and data scientists deliver these projects “in no time and at a fair price”. (It has a tiered pricing model based on requests per minute, which starts at $39pm and ranges up to $1,199pm, at the enterprise end, for one custom model running on a GPU. It does also offer a free tier so users can test models at low request velocity without incurring a charge.)
“The idea came from the fact that, as a software engineer, I saw many AI projects fail because of the deployment to production phase,” says sole founder and CTO Julien Salinas. “Companies often focus on building accurate and fast AI models but today more and more excellent open-source models are available and are doing an excellent job… so the toughest challenge now is being able to efficiently use these models in production. It takes AI skills, DevOps skills, programming skill… which is why it’s a challenge for so many companies, and which is why I decided to launch NLPCloud.io.”
The platform launched in January 2021 and now has around 500 users, including 30 who are paying for the service. While the startup, which is based in Grenoble, in the French Alps, is a team of three for now, plus a couple of independent contractors. (Salinas says he plans to hire five people by the end of the year.)
“Most of our users are tech startups but we also start having a couple of bigger companies,” he tells TechCrunch. “The biggest demand I’m seeing is both from software engineers and data scientists. Sometimes it’s from teams who have data science skills but don’t have DevOps skills (or don’t want to spend time on this). Sometimes it’s from tech teams who want to leverage NLP out-of-the-box without hiring a whole data science team.”
“We have very diverse customers, from solo startup founders to bigger companies like BBVA, Mintel, Senuto… in all sorts of sectors (banking, public relations, market research),” he adds.
Use cases of its customers include lead generation from unstructured text (such as web pages), via named entities extraction; and sorting support tickets based on urgency by conducting sentiment analysis.
Content marketers are also using its platform for headline generation (via summarization). While text classification capabilities are being used for economic intelligence and financial data extraction, per Salinas.
He says his own experience as a CTO and software engineer working on NLP projects at a number of tech companies led him to spot an opportunity in the challenge of AI implementation.
“I realized that it was quite easy to build acceptable NLP models thanks to great open-source frameworks like spaCy and Hugging Face Transformers but then I found it quite hard to use these models in production,” he explains. “It takes programming skills in order to develop an API, strong DevOps skills in order to build a robust and fast infrastructure to serve NLP models (AI models in general consume a lot of resources), and also data science skills of course.
“I tried to look for ready-to-use cloud solutions in order to save weeks of work but I couldn’t find anything satisfactory. My intuition was that such a platform would help tech teams save a lot of time, sometimes months of work for the teams who don’t have strong DevOps profiles.”
“NLP has been around for decades but until recently it took whole teams of data scientists to build acceptable NLP models. For a couple of years, we’ve made amazing progress in terms of accuracy and speed of the NLP models. More and more experts who have been working in the NLP field for decades agree that NLP is becoming a ‘commodity’,” he goes on. “Frameworks like spaCy make it extremely simple for developers to leverage NLP models without having advanced data science knowledge. And Hugging Face’s open-source repository for NLP models is also a great step in this direction.
“But having these models run in production is still hard, and maybe even harder than before as these brand new models are very demanding in terms of resources.”
The models NLPCloud.io offers are picked for performance — where “best” means it has “the best compromise between accuracy and speed”. Salinas also says they are paying mind to context, given NLP can be used for diverse user cases — hence proposing number of models so as to be able to adapt to a given use.
“Initially we started with models dedicated to entities extraction only but most of our first customers also asked for other use cases too, so we started adding other models,” he notes, adding that they will continue to add more models from the two chosen frameworks — “in order to cover more use cases, and more languages”.
SpaCy and Hugging Face, meanwhile, were chosen to be the source for the models offered via its API based on their track record as companies, the NLP libraries they offer and their focus on production-ready framework — with the combination allowing NLPCloud.io to offer a selection of models that are fast and accurate, working within the bounds of respective trade-offs, according to Salinas.
“SpaCy is developed by a solid company in Germany called Explosion.ai. This library has become one of the most used NLP libraries among companies who want to leverage NLP in production ‘for real’ (as opposed to academic research only). The reason is that it is very fast, has great accuracy in most scenarios, and is an opinionated” framework which makes it very simple to use by non-data scientists (the tradeoff is that it gives less customization possibilities),” he says.
“Hugging Face is an even more solid company that recently raised $40M for a good reason: They created a disruptive NLP library called ‘transformers’ that improves a lot the accuracy of NLP models (the tradeoff is that it is very resource intensive though). It gives the opportunity to cover more use cases like sentiment analysis, classification, summarization… In addition to that, they created an open-source repository where it is easy to select the best model you need for your use case.”
While AI is advancing at a clip within certain tracks — such as NLP for English — there are still caveats and potential pitfalls attached to automating language processing and analysis, with the risk of getting stuff wrong or worse. AI models trained on human-generated data have, for example, been shown reflecting embedded biases and prejudices of the people who produced the underlying data.
Salinas agrees NLP can sometimes face “concerning bias issues”, such as racism and misogyny. But he expresses confidence in the models they’ve selected.
“Most of the time it seems [bias in NLP] is due to the underlying data used to trained the models. It shows we should be more careful about the origin of this data,” he says. “In my opinion the best solution in order to mitigate this is that the community of NLP users should actively report something inappropriate when using a specific model so that this model can be paused and fixed.”
“Even if we doubt that such a bias exists in the models we’re proposing, we do encourage our users to report such problems to us so we can take measures,” he adds.
