TechCrunch+

ChatGPT, the AI that can write poems, emails, spreadsheet formulas and more, has attracted a lot of negative publicity lately. Q&A coding site Stack Overflow temporarily banned users from sharing content generated by ChatGPT, saying that the AI made it too easy for users to flood the site with spammy answers. Then, New York City public schools prohibited students and teachers from using ChatGPT on school-owned devices over fears of cheating and misinformation. .

That’s perhaps why AI21 Labs, an Israeli startup developing text-generating AI systems along the lines of ChatGPT, tried a different tack with its newly-released assistive writing tool, Wordtune Spices. A part of AI21’s expanding suite of generative AI, Wordtune Spices doesn’t compose emails and essays like ChatGPT. Instead, it suggests options that change the voice and style of already-written sentences, also offering up statistics from web-based sources to “strengthen arguments.”

“We see new amazing AI capabilities introduced on a weekly basis — [AI systems] that generate images, audio and text in a convincing human-seeming manner,” Ori Goshen, the co-CEO and co-founder of AI21 Labs, told TechCrunch via email. “With all the excitement, for these systems to become useful, they need to be robust, reliable and explainable.”

To that end, Wordtune Spices cites its work for each tidbit of information it profers, providing users with links back to the original sources. That’s one better than ChatGPT, which doesn’t always — or correctly — name sources and even sometimes points to sources that don’t actually exist.

AI21 says it developed “grounding and attribution” algorithms to search for relevant sources to base Wordtune Spices’ responses on and present the source links alongside info. The tool can help write a thesis statement and main ideas, including explanations and counterarguments, as well as provide analogies and creative expressions like jokes and quotes.

Users can choose from different cues (e.g. “legal,” “health care”) to prompt Wordtune Spices to suggest rewrites appropriate for particular professional documents. But Goshen says that Spices was designed to address a wide array of use cases, from writing essays and working on blog posts to drafting financial reports.

“Spices isn’t a tool that generates a full essay with one click, because we don’t think this helps create strong writers,” Goshen said. “Instead, we equip the user with a toolbox of language and storytelling tools for them to craft their text to perfection by making decisions on their own from a host of possibilities.”

But how well does Wordtune Spices work in practice, and is it better than the other AI-powered assistive writing tools out there?

Writer is among those tools — it employs an AI engine that evaluates key metrics such as plagiarism, sentence complexity, formality and active voice usage. Grammarly offers style and tone suggestions. So does Ginger, ProWritingAid and Slick Write.

This reporter wasn’t able to test Spices prior to its release due to technical issues on AI21’s end. But from a brief demo video, the rewriting feature appeared to work well enough minus the occasional grammatical error. To follow the sentence “It’s no secret that artificial intelligence tools have become one of the latest trends in content writing, blogging and copywriting as a whole,” Spices suggested “Some people, like content writers, may be scared of AI, but it can help you create high-quality content with a minimum of effort [sic].”

Spices cited an article from The New York Times in suggesting the sentence, but Goshen admits that there might be cases where the tool “[misses] information that should have been validated.” He also stressed that, while there are filters and “other measures” in place to try to prevent Spices from outputting toxic or biased text — a problem for text-generating tools like ChatGPT — the model might “make mistakes” all the same.

“We have been working on smart filters and other measures … We also tried to preemptively block any trolling attempts in that regard,” Goshen added. “That being said, Spices isn’t bulletproof, especially because it’s hard to train a [system] to make sensitive decisions that sometimes even humans struggle with.”

Goshen says that AI21 plans to put “a lot of focus” on moderation and “make it more robust” as Spices evolves. But for AI-curious writer holding out for the perfect system, Spices probably isn’t it — even if it works roughly as advertised.

AI21 Labs intros an AI writing assistant that cites its sources by Kyle Wiggers originally published on TechCrunch

Anna Raptis, founder and CEO at Mexico-based venture capital firm Amplifica Capital, was always someone who’s been “interested in facilitating economic development.”

Born in Australia, she followed that lead, holding positions with the United Nations and the World Bank, spending two decades working in the energy sector. That’s what ultimately brought her to Mexico: a focus on providing access to clean, safe and reliable energy as a way to help the country develop and improve people’s lives.

Her background in investing in building and developing investment opportunities, along with a notion to change the narrative around the lack of female tech entrepreneurs in Mexico, was what pushed her into venture capital.

She started out investing as an angel and in a couple of funds to learn more, with a focus on companies looking to solve important problems in Latin America.

“There are so many challenges, which means there are so many opportunities,” Raptis told TechCrunch. “Through investing in tech in Latin America, we can really make a difference in people’s lives.”

Raptis started Amplifica in 2020, the name inspired by the thesis of “celebrating, magnifying and amplifying the success of female founders in the region.”

In fact, her focus is on what she called “gender lens investing,” a concept driven by the notion that you can get better returns through investing in diversity. Amplifica invests mainly in teams where there are female founders or co-founders or goods and services focused on women. If a founder is not a woman, she does look deeper at the leadership team and to see what policies and practices they have in place supporting women.

“I was very inspired by a lot of women in the U.S. building similar funds because I haven’t seen anything like this,” she added. “These women showed me what was possible, and we know that when women see others have success it motivates them to think they can have a try.”

Raptis is a solo general partner working with a staff of five and began raising capital for her first fund, which recently closed with $11 million in capital commitments, exceeding its $10 million target. This fund has nearly 100 investors, including Mexico Ventures (a joint venture between Fondo de Fondos and Sun Mountain Capital); Grupo Herdez; Maria Ariza, CEO of BIVA; Melanie Devlyn, CEO of Devlyn Holdings; Daniel Undurraga and Oskar Hjertonsson, founders of Cornershop; and Loreanne Garcia co-founder of Kavak.

She explained that there are a lot of new funds in Mexico, but believes Amplifica’s success in raising money was because its focus on female founders is unique.

However, that aspect was also what made it a challenge, she recalls.

“People were skeptical because this is something that people had not seen in Mexico before,” Raptis said. There were a lot of questions around that, like, ‘Is Mexico really ready for a fund like this?’ and what would it mean to be investing in women in tech.”

Amplifica has invested in 10 companies so far, including inter-city mobility services startup Kolors, Verqor, focused on agricultural financing for inputs, and Mujer Financiera, an Argentine company providing tools for financial inclusion for women. In addition, Raptis wants to invest in 10 to 15 more companies in the next two years.

“We’ve had a lot of support from the VC ecosystem in Latin America, and we want to be the fund that the best female entrepreneurs in tech, in Latin America, reach out to first,” she added.

Amplifica Capital ‘wants to be fund that Latin America’s female tech founders reach out to first’ by Christine Hall originally published on TechCrunch

Apple’s M2 Mac Mini arrives January 24, starting at $599 by Brian Heater originally published on TechCrunch

Komunal is reaching out to Indonesia’s rural customers by going where they are. The fintech partners with more than 220 banks in tier 2 and tier 3 cities spread through 19 provinces. Founded in 2019, the startup announced today it has raised $8.5 million in funding led by East Ventures Growth Fund, with participation from AlphaTrio Sustainable Technology Fund, Skystar Capital, Sovereign’s Capital, Ozora and Gobi Partners.

Komunal digitizes rural banks, called BPRs (Bank Perkreditan Rakyat) through its DepositoBPR platform, which lets users make deposits and apply for loans digitally without needing to visit their bank’s physical location. In addition to DepositoPR, Komunal also has a peer-to-peer lending platform that connects MSMEs with lenders.

The startup says that in 2022, its platform channeled IDR 3.6 billion, or about $230 million USD, in deposits and loans to BPRs and MSMEs, representing a 350% year-over-year growth from $50 million in 2021. It expects transaction volume to reach more than $500 million USD by this year, and has recorded positive EBITDA since October 2022.

Komunal was founded three years ago as an SME financing platform based in Surabaya, East Java. CEO Hendry Lieviant told TechCrunch that the startup initially benefited from little competition because most lending fintechs were based in Jakarta, but then they realized that lack of data and Indonesian SME culture might limit its scalability.

Another opportunity presented itself in the form of Indonesia’s 1,500 BPRs. “They have strong local connectivity, but their recent performance has been lukewarm, unable to catch up with digitalization,” said Lieviant. “Combined they only made up about 2% of the Indonesian banking market although they have much bigger potential.”

Customers access banks through the DepositoBPR app. From there, they are able to choose their rural bank and deposit product, create a virtual account and transfer money directly into it. Lieviant said they can also get government-guaranteed deposits with higher interest rates.

Komunal’s new funding will be used to market DepositoBPR and build its core banking system for rural banks. The company also announced it has appointed Dr. Peter Jacobs, a former executive director of Bank Indonesia, as its commissioner, and will continue to hire for senior positions.

Komunal raises $8.5M to digitize Indonesia’s rural banks by Catherine Shu originally published on TechCrunch

There’s no big Apple event today, but the company’s got quite a bit of news to share this morning. In addition to the new Mac Mini, a pair of new MacBook Pro just dropped, sporting souped up versions of the M2 chips.

The 14- and 16-inch versions of the company’s high-end laptops now ship with a choice of the M2 Pro and M2 Max chips. The new Mini sports an M2 Pro option (in addition to the default M2), but the new Pros are the first — and thus far only — system to get the M2 Max chip. The company says the top line system is ” up to 6x faster” than the last generation of Intel systems (if you’re able to remember back that far).

“MacBook Pro with Apple silicon has been a game changer, empowering pros to push the limits of their workflows while on the go and do things they never thought possible on a laptop,” Greg Joswiak says in a release. “Today the MacBook Pro gets even better. With faster performance, enhanced connectivity, and the longest battery life ever in a Mac, along with the best display in a laptop, there’s simply nothing else like it.”

Both models are available for pre-order today and are set to start shipping in exactly a week, on January 24.

Apple brings M2 Pro and Max chips to the 14- and 16-inch MacBook Pros by Brian Heater originally published on TechCrunch

Several years ago, co-founders Leena Joshi, Ben Cheung and Erik Buchanan experienced frustration with the amount of manual work they were putting into collecting information about sales and marketing accounts. The tools in the market were focused solely on the collection of information, they say, instead of curating what was actually relevant, actionable and up to date.

Cheung and Buchanan — drawing on their AI and machine learning expertise — saw the potential to boost sales and marketing productivity by applying AI algorithms to workflows. Their work, together with Joshi’s, spawned CloseFactor, a platform that aims to harvest actionable information about companies from disparate sources.

“Built to harness the vast potential in automating manual research by employing machine learning techniques at scale, CloseFactor helps business-to-business sales and marketing teams identify the right target accounts to go after — including the right people at those accounts — so go-to-market teams have the blueprint to execute their strategy and hit their goals,” Joshi said. “Our goal is to become the go-to-marketing operating system for revenue teams.”

Prior to co-launching CloseFactor in 2019, Joshi, Cheung and Buchanan had long been in the business of building business-to-business technologies. Joshi spent several years in senior roles at VMware, Splunk and Redis Labs while Cheung sold his first startup, virtual assistant platform Genee, to Microsoft for an undisclosed amount. Buchanan also held a role at Microsoft before joining Google and then moving to LinkedIn as head of machine learning for LinkedIn Talent Solutions.

CloseFactor identifies different accounts by analyzing attributes across closed won deals (deals in the final stage of the sales cycle), sales strategies and CloseFactor’s own data to identify an ideal customer profile and how many accounts fit that definition. On top of that, CloseFactor segments accounts by those readiest to buy by curating insights including hiring trends and new projects or initiatives a target business is undergoing.

CloseFactor also gives a bird’s-eye view of all accounts, stack-ranked by those readiest to buy.

“CloseFactor automates account research, giving you a deep dive into every one of your accounts — including current or future projects and initiatives, as well as the decision-makers and influencers you should be reaching out to,” Joshi said. “With CloseFactor, you can now engage the right buyers with the right message at the right time. No existing tools provide this type of automation or visibility at scale.”

CloseFactor, whose customers include LaunchDarkly, Chronosphere, Sourcegraph and Zuora, competes both with intent data providers and contact databases like ZoomInfo. But the platform’s momentum evidently has investors impressed. Vertex Ventures led a $15.2 million Series A round in CloseFactor with participation from Sequoia, bringing the startup’s total raised to $20.5 million.

Joshi said that the recent infusion will be put toward scaling up CloseFactor’s product development efforts and creating go-to-market team, as well as building several new workflow integrations.

“We’ve already grown 5,411% since our first revenue quarter and plan to continue to focus on solving problems for our customers,” he said. “The biggest competitor we face in deals today is the status quo. The way revenue teams define their ideal customer profile and target market is broken, leading to massive inefficiencies in their sales and marketing execution.”

Palo Alto, California-based CloseFactor currently has 20 employees and plans to hire this year.

CloseFactor raises $15.2M to automate repetitive sales processes by Kyle Wiggers originally published on TechCrunch

There is very, very little room for error in aerospace and defense (A&D) manufacturing. For companies that build products like missiles, rocket boosters and avionics, each part must not deviate more than a hairsbreadth from its technical specifications.

Despite the precise demands of the industry, however, parts ordering is generally done using systems that are only slightly better than carrier pigeon: generally some combination of Microsoft Excel spreadsheets, long text PDFs attached to email, or an A&D company’s internal portal. Communication between companies and suppliers about these highly technical parts can also be bogged down by similarly low-tech, human-driven errors, like forgetting to copy someone on an email.

“I had this crazy theory that nothing can really move forward in terms of automation until all of that text, all of those decisions, get digitized,” Malory McLemore said in a recent interview with TechCrunch.

To solve this problem, McLemore and Anne Wen founded Stell, a startup that’s building a platform to bring new workflows to parts ordering. The company, which is less than six months old, just closed an oversubscribed $3.1 million pre-seed round led by Wischoff Ventures and Third Prime VC as it gears up expand its team and build out its product. The company is hoping that its platform can reduce error and improve efficiency – two variables that will be key to shoring up America’s industrial base.

Forming a plan

Stell was still just a “crazy theory” when McLemore embarked on an MBA at Harvard Business School. That’s where she met Anne Wen, a fellow graduate who had experience getting space startups off the ground. The idea continued to germinate, but ultimately both McLemore and Wen completed the program unsure of how to move forward.

“We left school feeling like we couldn’t figure out how to make this business work,” McLemore explained. “Selling software to aerospace and defense is hard. There aren’t a lot of solutions competing with the big guys for a reason.”

McLemore has seen how the A&D supply chain functions up close while working for both established aerospace behemoths and disruptive machining startups. Her experience includes stints at Airbus and Raytheon, where she helped build anti-ballistic missiles and airplanes. Most recently, she was machining-parts startup Hadrian’s first product manager.

This experience gave her a look into the parts ordering process from the perspective of the customer, at Raytheon, and the supplier, at Hadrian. On both ends, she was confronted with inefficiencies, missed deadlines, and pointless errors.

Even at a startup like Hadrian, with millions in funding from the likes of Andreessen Horowitz and Lux Capital, the same core issues with parts ordering persisted.

“Hadrian’s customers were still sending PDFs and emails to Hadrian,” she said. “Even if Hadrian builds this totally robotic, autonomous factory, all the inputs to the process are still broken.”

“So I called Anne, and I was like, I think I know how we can do this.”

Toward execution

When it comes to parts ordering in A&D, it is not only the technical specifications that need to be communicated to suppliers. Often, parts need to be inspected in certain ways, or tested; those results need to be communicated back to the customer. McLemore said Stell’s platform would also include room for these key deliverables and provide space for communication with customers directly within the software.

That the A&D industry has such outdated workflows may seem surprising, but up until now there has been very little pressure on the biggest companies to change their systems. McLemore pointed to the outsized power of this very small, very powerful group of primes relative to the suppliers.

But the landscape is changing: more venture capital is pouring into aerospace and defense startups, and there’s increasing pressure to secure these critical supply chains by bringing manufacturing back to American machine shops. Suppliers could have more leverage to demand better workflows, which McLemore says benefits both customers and suppliers.

Other things are changing too: legacy primes have historically been disincentivized from changing any process because it introduces uncertainty and risk. But Stell is hoping competitive pressure from outside the traditional industrial base will also nudge them to rethink their systems while not sacrificing mission assurance.

Right now, Stell is “running full speed,” Wen said; she and McLemore are in the process of onboarding Stell’s first dedicated software engineer and are thinking ahead to a design hire. They’re also planning on registering under the International Traffic in Arms Regulations (ITAR), the series of regulations that govern technologies related to U.S. defense.

The company is aiming to get an alpha prototype to 1-2 customers by April. That test period will last until summer; once concluded, McLemore and Wen plan to launch to the industry at-large.

Stell wants to modernize the “unsexy” workflows slowing down America’s industrial base by Aria Alamalhodaei originally published on TechCrunch

GoMechanic has laid off a significant number of its workforce as the Sequoia India-backed startup struggles to raise funds amid serious concerns of accounting troubles, a source familiar with the matter told TechCrunch.

The Gurgaon-headquartered startup has cut 70% of the workforce and asked the remaining staff to work without pay for three months, Indian news outlet The Morning Context reported Tuesday, citing unnamed sources.

GoMechanic did not respond to a request for comment.

The move comes as GoMechanic struggles to raise funds for over a year despite advanced stages of discussions with several investors. The startup was in talks early last year to raise a round of funding led by Tiger Global at over $1 billion valuation, TechCrunch reported.

The talks did not materialize into a deal after some discrepancy was found during the due diligence process, the source said. The startup later engaged with a number of investors including Malaysia’s Khazanah to raise a large round.

This round is also unlikely to go through as serious discrepancies have been found in its books, the source said, requesting anonymity speaking to the press. A recent probe into the startup, which counts Tiger Global among its backers, found that several of its garages were fictitious among other issues, the source said.

Sequoia India-backed GoMechanic faces severe trouble by Manish Singh originally published on TechCrunch

“This is the next 20 years of my life. This is not like, build the company in four or five years and sell to Google. This is a big, long journey.”

Matt Rogers, who co-founded Nest with Tony Fadell back in 2010, was talking about his new startup, and he thinks it’ll be big. He’s probably right, but it’s also trash, in a way.

Today, Rogers, co-founder Harry Tannenbaum and the rest of the team at Mill Industries are launching their first product. At first glance, it doesn’t look like much, maybe a sleek trash can. That’s not an unfair assessment, but the Mill bin is a lot more than just a receptacle.

Mill’s kitchen bin gobbles up food scraps, dries them out and grinds them to bits overnight. It also neutralizes odors with a charcoal filter. When the bin is full, Mill automatically mails a box to return the waste back to the company, where it’s cleaned, sifted, pasteurized and bagged to be sold as chicken feed to farmers.

The goal is to eliminate the climate impact of food waste.

“This feels like the most tractable of all climate problems,” Rogers said. “We don’t need to invent nuclear fusion. We don’t need to figure out how to build a low weight aviation fuel. It’s just like, don’t put food in the trash.”

Nest co-founder Matt Rogers’ new startup is trash by Tim De Chant originally published on TechCrunch

On the heels of raising a big round of funding at a $1 billion valuation last week, DeepL is taking the wraps off a new language product, the first extension for a startup that made its name from its popular AI-based translation tools.

Write is a new tool that fixes your writing — catching grammar and punctuation mistakes, offering suggestions for clarity and more creative phrasing, and (soon) giving you the option to change your tone. Write is based on the same neural network that powers DeepL’s translator, and significantly, it is another step ahead in how artificial intelligence technologies, specifically those in natural language processing, are being used to alter how humans are communicating with each other, a big theme at the moment.

The functionality of Write may sound a little familiar to you. That’s because DeepL’s new service is going head to head with a popular product already on the market: Grammarly is currently leaned on by more than 30 million daily active users and some 50,000 businesses and teams.

If you don’t use Grammarly, chances are that you at least know about it — its YouTube ads are (in)famously ubiquitous. And at last count, despite much nay-saying from VCs about the pitfalls of features versus platforms, Grammarly, as of 2021, saw a huge spike in its valuation from $1 billion to a whopping $13 million.

DeepL’s mantra is to embrace competition and use it as a motivator to do things better. In its primary (and before today, sole) product, the company has long competed against two of the biggest tech companies in the world, Microsoft and Google, which both offer real-time translators for individuals and as a service used by third parties. (And that’s before considering the myriad other options out there for real-time translation.)

“We are always in race mode,” CEO and founder Jarek Kutylowski said of its flagship translation service. “We are accustomed to big adversaries, and part of our culture is to push forward through that.” Indeed, for many, DeepL’s neural network-based translator works better than these others, capturing certain nuances and meanings that rivals have missed.

That approach, it seems, is now the template for how DeepL will tackle new product frontiers, starting first with Write.

Launching initially in English and German and as a monolinguistic tool (you put in English writing and get English results), the plan is to see how Write is used across these two languages first, both to improve them and to figure out how to develop Write, whether that means new features or new languages. As with its basic translation tools, you can use Write free without needing to register (as you do with Grammarly to use its free tools).

Options, it seems, are at the crux of the service: in addition to snagging basic grammatical and punctuation errors, the focus will be on generating choices for users covering style, tone, phrasing and diction, rather than re-writing everything that’s entered into Write. In doing that, you might ask if Write is showing its limitations, or if its creators are making a conscious choice of where AI can help people do their best work. That’s a debate that definitely has opened up with the release of OpenAI’s GPT service, which takes a basic brief and writes everything for you based on that.

My initial test-drive of Write was a mixed bag and reveals that there is still room for learning and improvement. Write, ironcially, wasn’t very good (yet?) at guessing what I meant below when I wrote “right” instead of “write.”

But it did a good enough job of fixing my misuse of good.

Of all the directions that a AI-based translation startup could choose to go for its next product, Kutylowski told me that DeepL decided to work on Write because of patterns that it started to notice in how its translation product was being used (or misused as the case is here).

“People were misusing the translator to write texts in first in a different language from English, and then plugging them back into English assuming that they cold get the AI to write a better original version,” he said. The team decided to build Write to essentially improve that and cut out the translation middle step, which was skewing the results anyway.

The “AI writing companion,” as Kutylowski describes it, is aimed both at native speakers but perhaps even more at people who write passably if slightly awkwardly in a second language and are hoping to give their words that extra native shine. “The idea is we could help a student improve a grade by one,” in a manner of speaking, he added.

The next level of development, he said, will be to focus on the elusive qualities of tone, “not content but phrasing, creative input,” he said — but critically doing so while continuing to anchor that content in a person’s own words and ideas, not those generated by the AI from scratch.

DeepL takes aim at Grammarly with the launch of Write, to clean up your prose by Ingrid Lunden originally published on TechCrunch