Microsoft partners with AiFi to launch Smart Store Analytics, a tracking service for cashierless outlets

In partnership with AiFi, a startup that aims to enable retailers to deploy autonomous shopping tech cost-effectively, Microsoft today launched a preview of a cloud service called Smart Store Analytics. A part of Microsoft’s Cloud for Retail product suite, Smart Store Analytics provides retailers using AiFi’s technology with shopper and operational analytics for their fleets of “smart stores.”

AiFi’s “smart stores” are grocery stores, sport stadiums and convenience stores equipped with cameras that track what customers pick up and place in their carts, automatically charging their payment cards when they’re ready to check out. It’s similar to Amazon’s Just Walk Out tech; digital video captured by AiFi’s cameras feed a computer vision algorithm that recognizes shoppers’ behaviors, including when they pluck individual products from the shelves.

Once installed, Smart Store Analytics — which was also co-developed by Zabka Polska, a chain of Polish convenience stores — pulls store data from the AiFi platform to deliver insights that let retail managers better adjust store layouts and inventory. Smart Store Analytics shows how much customers shop, interact with products and move through aisles, displaying foot traffic as a heat map and tracking how much money each customer spends on average. The service also plots how long customers dwell in front of certain displays and the “unit sales-to-shopper height” ratio to help dial in on optimal shelf placements.

It might sound like a lot of personal data Smart Store Analytics is collecting. But Microsoft and AiFi claim that the service doesn’t use facial recognition or biometrics, only creating a virtual avatar of customers as they enter retail stores.

With Smart Store Analytics, AiFi will handle store setup, logistics and support while Microsoft will deliver models for optimizing store payout, product recommendations, inventory and so on. Żabka plans to use Smart Store Analytics in its 50-plus “Nano” stores to predict individual store demands, build ordering schedules for replenishment and react faster when a product is out of stock.

AiFi’s autonomous in-store tracking tech in action, with avatar stand-ins for real shoppers.

“Digital technology is what will make the difference between retailers that thrive during this period of economic, societal and technological change, and those that get left behind,” Shelley Bransten, CVP of worldwide retail and consumer goods industries at Microsoft, said in a statement. “Honestly, there’s not a retailer out there who doesn’t want to reimagine the store experience, but until now it’s not really been possible.”

Why’s Microsoft working with AiFi as opposed to rival startups like Sensei,Standard Cognition,Zippin,GrabangoandTrigo? Besides being an Azure customer, AiFi’s market share could well have something to do with it. The california-based company claims to be the world’s largest provider of autonomous shopping solutions with a platform that’s deployed in 100 retail spaces across North America, Europe, the Middle East, Asia and Australia. In addition to having autonomous shopping tech in grocery stores, AiFi’s brought it to college campuses, sports venues and workplaces including the Miami Dolphins stadium and the University of Denver; over 800,000 shoppers have used it to make 1.5 million purchases to date.

“Our platform is way more flexible than a lot of our competition because we don’t have to touch the shelves at all,” AiFi CEO Steve Carlin said in a press release. “You can take a space that already exists and retrofit it. And you can move the shelves. You can’t do that with weight sensors, which require digging a trench in the floor, running cable to the shelf and electrifying the shelf. And once you’ve done that, that shelf isn’t moving.”

Interestingly, Microsoft was reportedly once developing a Just Walk Out-like system for tracking what people place in their shopping carts. The effort seemingly bore fruit in Dynamics 365 Connected Store, Microsoft’s service that uses a combination of computer vision, cameras and internet of things sensors to spotlight customers inside stores and personalize recommendations based on their behaviors.

Microsoft’s explorations in the space make sense given that so-called cashierless checkout systems stand to become quite a lucrative venture in the coming years.

A 2017 report from Juniper Research predicted that automated checkout systems would process more than $78 billion in transactions by 2022 across as many as 5,000 retail outlets. Juniper estimated that these technologies would drive an average increase in revenue of more than $300 per shopper per year, in part because they’d allow store associates to provide more personal service.

In a separate survey published in 2019, RBC Capital Markets estimated that Amazon’s cashierless Amazon Go stores bring in about 50% more revenue on average than typical convenience stores. The average Go store generates an estimated $1.5 million in revenue a year (excluding days when they’re closed), according to RBC, while a regular convenience store of the same size brings in just over $1 million a year in sales.

Microsoft partners with AiFi to launch Smart Store Analytics, a tracking service for cashierless outlets by Kyle Wiggers originally published on TechCrunch

Coinbase to cut 20% jobs, abandon ‘several’ projects to weather downturns in crypto market

Coinbase plans to cut 950 jobs, or about 20% of its workforce, and shut down “several” projects as the U.S. crypto exchange giant looks to reduce its expenses to increase its “chances of doing well in every scenario.”

This is the second round of major layoff at the crypto exchange, which eliminated 18% of its workforce, or nearly 1,100 jobs last June, but there was “no way to reduce our expenses significantly enough, without considering changes to headcount,” Coinbase co-founder and chief executive Brian Armstrong wrote in a blog post Tuesday.

The moves are part of the company’s efforts to cut its operating expenses by about 25% quarter over quarter, he said. The company estimates that it will incur approximately $149 million to $163 million in total restructuring expenses, consisting of approximately $58 million to $68 million in cash charges related to employee severance and other termination benefits, it disclosed(PDF) in a 8K filing with SEC Tuesday.

In the filing, Coinbase also disclosed that it expects its adjusted EBIDTA losses for the year ending in December 31, 2022 to be within “the negative $500 million loss guardrail” it set last year.

As with firms in other industries, crypto firms are aggressively undertaking major decisions to survive the downturn in the broader market, which has reversed much of the gains from the 13-year long bull run. Kraken said in November that it plans to lay off 1,100 people, or 30% of its workforce.

Armstrong said the crypto industry is reeling from the fallout from “unscrupulous actors,” likely referring to Sam Bankman-Fried, the founder of the collapsed crypto exchange FTX (which stole billions of dollars of customer funds), and disgraced crypto hedge fund Three Arrows Capital founders Kyle Davies and Su Zhu, and warned that “there could still be further contagion.”

“As we examined our 2023 scenarios, it became clear that we would need to reduce expenses to increase our chances of doing well in every scenario. While it is always painful to part ways with our fellow colleagues, there was no way to reduce our expenses significantly enough, without considering changes to headcount,” he wrote.

“Dark times also weed out bad companies, as we’re seeing right now. But those of us who believe in crypto will keep building great products and increasing economic freedom in the world. Better days are ahead, and when they arrive, we’ll be ready. Thank you for everything you’ve done to get us this far, and everything you will do to carry us forward.”

Shares of Coinbase are down 83% in the past one year.

Armstrong did not specify which all projects the firm plans to shut down, but said those projects had a “lower probability of success.”

Coinbase to cut 20% jobs, abandon ‘several’ projects to weather downturns in crypto market by Manish Singh originally published on TechCrunch

Google’s Extension SDK aims to bring latest features to older Android versions

Fragmentation has been a longstanding complaint about the Android ecosystem. Often users miss out on features of the latest Android version because they are using older devices that are no longer updated. To reduce the gap, Google has released the first public version of the Extension SDK, which aims to bring features of the latest Android version to older iterations.

As a part of this announcement, Google is opening up Photo Picker API support to Android 11 and Android 12. Photo picker lets users give access to select photos from their library to an app — right now, only Android 13 users can use the latest photo picker interface.

Notably, as this is an SDK app developers will have to include this functionality in their apps. But users won’t need a system update to use the photo picker.

Google Photo Picker

The Extension SDK also brings support to implement AdServices APIs to prepare for the launch of the Android Privacy Sandbox. The Android Privacy Sandbox is Google’s new ad stack to replace cookies in Chrome. Currently, Google is testing Privacy Sandbox on a limited number of devices running Android 13 with a wider beta launch expected this year. With components like the Extension SDK, Google will aim to roll out Privacy Sandbox support to devices running older versions of Android.

Google has tried to compress feature rollout times through programs like Project Mainline, which was introduced with Android 10 to make the operating system more modular and push features through Google Play. So users don’t have to wait for phone makers to issue a software update. The Extension SDK is another step in that direction.

Google’s Extension SDK aims to bring latest features to older Android versions by Ivan Mehta originally published on TechCrunch

Indian edtech Unacademy cuts upskilling service to double down on tests product and LinkedIn rival

Unacademy said on Tuesday that Relevel, its upskilling product, is shifting focus to tests product and the newly launched LinkedIn-rival NextLevel, the latest in a series of changes from the Indian edtech unicorn as it scrambles to aggressively find the next breakthrough.

As part of the move, about 40 people at Relevel will be let go “because of lack of availability of roles,” Unacademy co-founder and chief executive Gaurav Munjal wrote to employees on Tuesday. About 80% of Relevel’s team will be absorbed by other businesses within the Unacademy Group, he said in the note, reviewed by TechCrunch.

Those enrolled in Relevel’s cohorts won’t be impacted and the parting team members will be offered severance pay for their notice period and two additional months, said the Indian edtech, which is backed by SoftBank, Tiger Global and Sequoia India.

“We are very grateful for the hard work and contributions of the Relevel team. Their hard work and hustle allowed us to scale revenues quickly but unit economics proved challenging,” wrote Munjal.

“Our culture is to pursue novel and creative ideas but we are also disciplined about holding ourselves to a high bar to continue the projects we start. Again, this decision doesn’t take away from them their positive contributions especially towards improving the learning trajectory and job prospects of our customers.”

The company did not immediately respond to a request for comment.

Unacademy launched Relevel in 2021, hoping to bank on the rising popularity of creators by giving them a platform to launch cohort-based live courses. The product, in which the edtech giant invested over $20 million, crossed $10 million in annualized recurring revenue last year.

The startup launched NextLevel, its take on LinkedIn last month.

Unacademy has eliminated about 1,400 full-time and contract positions since last year as the startup has sought to cut costs and become more disciplined.

The Bengaluru-headquartered startup “always raised more money than what was needed” to “continuously experiment and grow our platform without worrying about when we will run out of money,” Munjal told employees last year. “But now we must change our ways. […] Winter is here.”

Indian edtech Unacademy cuts upskilling service to double down on tests product and LinkedIn rival by Manish Singh originally published on TechCrunch

App Store and Play Store are flooded with dubious ChatGPT apps

ChatGPT is the hottest topic of discussion in the tech industry. OpenAI’s chatbot that answers questions in natural language has attracted interest from users and developers. Some developers are trying to take advantage of the trend by making dubious apps — both on the App Store and the Play Store — that aim to make money in the name of pro versions or extra credits to get more answers from AI.

It’s important to remember that ChatGPT is free to use for anyone on the web and OpenAI hasn’t released any official mobile app. While there are plently of apps that take advantage of GPT-3, there is no official ChatGPT API.

As Macrumors noted, an app named “ChatGPT Chat GPT AI With GPT-3,” has managed to reach the top charts in the productivity category in multiple countries.

While the app is free, it offers weekly ($7.99) and monthly ($49.99) packs to have unlimited chats with the AI bot. Some of the reviews of the app suggest that the subscription doesn’t add any value and the app appears to be fake. There is no way to tell if the app produces any tangible results because I couldn’t get past the loading screen.

Notably, the developer had a similar app on the Play Store with more than 100,000 downloads (archived link), but it has been removed now.

Over the weekend, CEO of edtech startup Bloomtech Austen Allred tweeted that the App Store is full of apps — with no association with OpenAI — that are trying to charge money for using ChatGPT.

The iOS App Store is full of folks putting ChatGPT into a paid wrapper with ambiguous language that would let you believe you’re paying for ChatGPT pic.twitter.com/3w0rK14E5I

— Austen Allred (@Austen) January 7, 2023

Google Play Store is also filled with sketchy ChatGPT apps with thousands of downloads that don’t offer any usable functionality.

Meanwhile on Google Play…fake chatGPT 1star app with >100k downloads. pic.twitter.com/5fj5SEITwp

— nisten (@nisten) January 6, 2023

The playbook of these apps is to include ChatGPT in the app name and appear favorably in search results by bolstering their own ratings. Some folks are also developing multiple apps with similar names in hopes that one of them catches users’ attention.

It’s not clear if Apple and Google are actively taking any action on these apps. We have asked the companies for a comment, and we will update the story if we hear back.

App developers cloning websites or other popular apps is not a new trend. Previously we have seen both Apple’s and Google’s app stores being flooded with clones of Wordle, Threes, and Flappy Bird. The question is about how swiftly and stricly these platform gatekeepers act on them, if at all.

App Store and Play Store are flooded with dubious ChatGPT apps by Ivan Mehta originally published on TechCrunch

OpenAI in talks to back Zeloof and chip legend Keller’s startup at $100 million valuation

Sam Zeloof is a popular name on YouTube and tech Twitter. For years, he has been documenting his impressive journey of building silicon chips at his garage. What most people don’t know is that Zeloof has been working to take this expertise to its next logical level.

Zeloof has partnered with engineering veteran Jim Keller to found Atomic Semi, a startup that seeks to manufacture chips, two sources familiar with the matter told TechCrunch. The startup is using “radically” simplified and miniaturized semiconductor fabs and prototyping integrated circuits to produce “much more affordable” chips in hours, instead of the typical months-long timeframe, the sources said.

The duo’s startup has received attention from investors, too. OpenAI is in advanced stages of deliberations to invest in Atomic Semi, participating in a seed round of about $15 million at a valuation of $100 million, the people said, requesting anonymity sharing non-public information.

The deliberations have not been finalized and the terms of the deal may change, they said. OpenAI declined to comment last week. Zeloof did not respond to a request for comment. Keller could not be reached.

The startup’s existence, the duo’s partnership and their fundraising efforts have not been previously reported.

OpenAI in talks to back Zeloof and chip legend Keller’s startup at $100 million valuation by Manish Singh originally published on TechCrunch

Jetstream, a Ghanaian e-logistics platform for Africa’s B2B importers and exporters, takes in $13M equity, debt

The market for cross-border logistics services is said to hit revenues of $32 billion by 2025, with several companies vying for market share in the ever-growing competitive industry. Ghanaian e-logistics startup Jetstream Africa is on the list, and today, it’s announcing that it has secured $13 million in equity and debt pre-Series A financing.

Fintech lender and private equity firm Cauris was the sole provider of the debt financing while the equity investors include Octerra, Wuri Ventures, Seed9, The MBA Fund, French development institution Proparco and ASCVC, a venture fund founded by executives of the supply chain visibility platform Project44. Existing investors Alitheia IDF and Golden Palm participated as well.

The round is coming about 18 months after the Tema-based cross-border logistics platform announced a $3 million seed round (including $1 million in debt). Jetstream says this new investment will allow it to expand into new countries — it’s currently in 29 (12 in Africa) countries — and continue to develop its technology platform, which vertically aggregates fragmented logistics and financing vendors in the world of African trade.

At the time of its seed round, Jetstream Africa had two business lines: one providing logistics services to cargo owners dealing with import and export and another distributing financing to freight forwarders. However, Jetstream has bundled both products over the past couple of months to serve only cargo owners. According to the startup’s chief executive Miishe Addy, Jetstream achieved product-market fit correspondingly.

“Running those two lines side by side, we observed that the import or export business controls the supply chain,” she said on the pivot. “Although the cargo owners and freight forwarders have a lot of information asymmetry, the importer and exporter can put pressure on the freight forwarder to digitize the supply chain. We simplified our business into just the import-export product line by directly them with a combination of trade financing and logistics.”

Jetstream’s new business model has shifted to that of a freight forwarder. The company now involves itself in the end-to-end movement of shippers’ cargo (both import and export), charges a fee and, most importantly, supplies finance to those who need it. Typically, the traditional method for most cargo owners when they want to take out a loan to run their businesses is to go to banks to secure a letter of credit. Whether they get it or not depends on the bank of their counterparty. To elucidate: Say a Ghanaian importer is making a transaction with a Chinese exporter — the bank in Ghana collects cedi and interacts with the exporter’s bank in China, which, upon vouching for the cargo owner, dispenses the yuan.

It’s a time-consuming process that can take several weeks. And for cargo owners on both sides of the transaction who want access to faster credit, the letter of credit system isn’t efficient, leaving them to find other sources of capital that require some form of collateral for their loans. Jetstream essentially provides them with working capital backed by actual shipment. According to Addy, the three-year-old startup takes a security interest in the cargo. Rather than handling the letter of credit itself, Jetstream underwrites loans — to be paid back within 15-90 days — through its banking partners and disburses the loan proceeds to every vendor in the supply chain.

“If you’re importing 10 containers, in addition to paying for the actual good, importers have to pay the shipping line, customs broker on both sides, truck drivers on both sides, you have to pay a warehouse operator in some cases, or container terminal. There’s a minimum of nine different vendors you have to pay,” noted Addy, who co-founded Jetstream with COO Solomon Torgbor in 2018.

“And when someone applies for a Jetstream loan, they’re not just saying give me $50,000 but enough money to fund this entire shipment and pay these nine vendors. Also, we don’t give the money to the cargo owners but to the nine vendors directly.”

R: Miishe Addy (Jetstream Africa CEO)

Jetstream has grown its trade finance product from the $1 million debt it secured in mid-2021 to about $9 million in total loans disbursed so far. Its projection is to increase that amount fivefold by the end of this year, Addy said. The chief executive also mentioned that Jetstream has scaled from disbursing one loan per month to up to 50 loans per month after switching its business model, thus becoming EBITDA positive. Also, revenue has grown by 48% and active customers by 102% within the past year, according to a statement shared by the e-logistics startup, which handles shipments consisting of 47% air freight, 44% ocean freight and 9% ground transport.

The 44-man team, which competes with the likes of Sote, SEND, One35 Port and MVX among others, has been able to strike several essential partnerships for its next growth phase, including multinational banks like Societe Generale and startups such as Lami and MFS Africa. Tokunboh Ishmael, co-founder and principal partner at Alitheia IDF, one of Jetstream’s investors, says this round of funding, which supports the startup’s expansion to new markets, will see it capitalize on trade policies like AfCFTA, “enabling richer inter-continental trade which is needed to support inclusive economic development and unleash the continent’s full potential.”

Jetstream, a Ghanaian e-logistics platform for Africa’s B2B importers and exporters, takes in $13M equity, debt by Tage Kene-Okafor originally published on TechCrunch

Qarnot creates green data centers by putting servers in central heating boilers

Running a data center means that you have to find innovative ways to manage heat from the servers. And French startup Qarnot is standardizing its process to reuse fatal heat and turn it into an asset.

While the startup has been around for a while, Qarnot now wants to reach the next level. It has raised a €12.5 million funding round ($13.3 million at today’s exchange rate). It has also negotiated a €22.5 million credit line ($24 million) to finance its future projects.

Société Générale Ventures, ADEME Investissement, Demeter, la Banque des Territoires and Colam Impact are participating in today’s funding round.

Qarnot started with electric heaters for construction companies looking for heaters for their new buildings. The company put servers in those heaters so that CPUs and other components would generate heat. The rest of the heater would act as a passive cooling system and warm residential and office buildings.

At the other end, companies such as BNP Paribas, Société Générale and 3D animation studio Illumination are renting those servers for their own needs. This is an innovative way of building decentralized data centers.

“Ever since we launched Qarnot 12 years ago, our goal has been to valorize fatal heat from computers. Today, it’s a hot topic and there are a lot of people talking about it,” co-founder and CEO Paul Benoit told me.

But the main issue with this system is that you don’t need a heater all year round. Qarnot has found a way to counter this seasonality effect by building a new product —scalable boiler systems.

These modules pack up to 12 servers with a standard form factor based on Open Compute Project server designs with AMD Epyc and Intel Xeon CPUs. Each module is then connected to the water system so that cold water enters the module and is turned into hot water. Up to 95% of computer waste heat is turned into hot water.

I asked about the seasonality of central heating boilers. And it turns out that heating networks work all year round. For instance, they produce hot water, which is always required.

“We want to tackle the baseload. We’re going to run all year round and the heating network is going to rely on a complimentary source,” Benoit said. Buildings or neighborhoods are still going to use electricity or natural gas directly for those cold winter nights.

Qarnot can provide multiple modules and make them work in parallel. It generates more hot water and it increases the total compute power of the Qarnot platform.

“The idea is that we should be able to deploy data centers in locations where people consume heat. It’s a completely different approach compared to traditional data centers,” Benoit said.

Image Credits: Qarnot

Clients include social landlords, property developers, local authorities, swimming pools and heating network operators. Qarnot already rolled out a pilot data center in Finland with 100kW of compute. It’s a small data center, but the company is already looking at other locations with 500kW up to a few MW of compute capacity.

“With our system, you don’t need a cooling system and we can sell the heat that is generated from our servers. In the data center industry, when you have servers that require 1kW, they also require up to 500W in air conditioning,” Benoit said.

And this is key to understanding Qarnot’s appeal. From a business perspective, Qarnot uses electricity in two different ways. It runs computers and it sells heat, which is great for the company’s bottom line. From an environmental perspective, Qarnot greatly improves the carbon footprint of data centers.

“We think the future looks like a plethora of mid-sized data centers. Edge computing is slowly showing up. Today, we don’t really know how it’s going to look like but we know that there are operators looking for hundreds of data centers with a few kilowatts of capacity,” Benoit said.

Qarnot could profoundly change the data center industry. If the company becomes massively successful, there would be no reason to construct dedicated buildings to host hundreds of thousands of computers anymore.

Image Credits: Qarnot

Qarnot creates green data centers by putting servers in central heating boilers by Romain Dillet originally published on TechCrunch

Singapore-based Supermom helps parenting brands navigate a post-cookie world

Supermom, a parenting platform with 20 million users in six Southeast Asian countries, offers parents price comparisons, communities and the chance to earn money by completing surveys. For brands, it gives them a way to conduct market research and collect first-party data, which is important as marketers prepare for a post-cookie world.

The Singapore-based startup announced today it has raised an oversubscribed Series A of $8 million SGD (about $6 million USD) led by Qualgro, with participation from AC Ventures. Supermom currently has a presence in Indonesia, Malaysia, Singapore, Vietnam and Thailand, and plans to expand into more markets. Over 200 consumer brands use Supermom for marketing research, including Kimberly Clark, Procter & Gamble and Philips.

Supermom was founded by Joan Ong, Luke Lim, Lynn Yeoh and Rebecca Koh, and originally started as a platform in 2013 for community groups and in-person events. “Motherhood is quite a lonely journey and I wanted to support other parents,” said Ong. The next five years were focused on Singapore. Then in 2019, Supermom’s team decided they wanted to scale across Southeast Asia, so they pivoted their focus from community events to digital marketing.

Supermom’s market research platform for businesses lets brands perform market testing with very targeted groups of people. Its backend tracks the demographics of parents in its database, including aggregated data about the age of their kids, their occupations and interests. In addition to writing reviews or participating in surveys, parents can apply to be brand ambassadors through Supermom, and when they do, brands can see their number of followers on social media and what groups they are in.

Qualgro associates Jeremy Soh and Neo WeiSheng, with Supermom founders Rebecca Koh, Joan Ong, Lynn Yeoh and Luke Lim

Lim says that quantitative surveys of about 1,000 people usually takes months to put together, but can be done within an hour through Supermom. Surveys can be launched simultaneously in six countries, with specific users invited to take them. To set up a survey, brands select a country, then filter for specific demographics like age of kids. Supermom’s platform also guides them through the process of writing questions.

In addition to their own surveys, brands also get information like brand indexing, or a ranking of which brands are most trusted by consumers. That data can be broken down by demographic—for example, researchers can see if a brand is used by 50% of mothers aged over 30 years, and then narrow that down by disposable income level.

Supermom will use its new funding on its data and product capabilities, and expansion in Southeast Asia.

“As companies all move into a cookieless world, having direct access to consumers is very, very important,” said Ong. “Because they can no longer do retargeting and use cookies to target consumers. So with the millions of users that we have, we are helping a lot of brands to get this critical first-party data so they can actually access their target audience.”

In a statement about the financing, Qualgro general partner Weisheng Neo said, “We recognize the current gap in brands’ access to reliable, secure data on parents, a large and growing customer segment. Brands will find the Supermom platform to be a treasure trove of insights and first-party data activation.”

Singapore-based Supermom helps parenting brands navigate a post-cookie world by Catherine Shu originally published on TechCrunch

Pin It on Pinterest