A new AI-powered development environment called Copilot Workspace has been launched by [GitHub](https://github.com/) ahead of its annual GitHub Universe conference in San Francisco. By utilizing "Copilot-powered agents," Copilot Workspace is intended to support developers from the very beginning of ideation and planning through coding, testing, and implementation. The leader of GitHub's research and development team, Jonathan Carter, characterizes Workspace as an expansion of the AI coding helper Copilot, improved with new features including Copilot Chat for interactive coding help.

[According to Carter](https://techcrunch.com/2024/04/29/copilot-workspace-is-githubs-take-on-ai-powered-software-engineering/), research revealed that developers frequently found the first stages in solving a code problem to be the most difficult. By providing AI-assisted starting points and collaborative tools throughout the full codebase, Copilot Workspace seeks to streamline this process. Copilot struggles to turn a profit even with over 1.8 million paying users and 50,000 commercial clients; monthly [losses are estimated to reach $20 per user](https://www.wsj.com/tech/ai/ais-costly-buildup-could-make-early-products-a-hard-sell-bdd29b9f). With competitors like Amazon's CodeWhisperer and several companies joining the market, competition is also growing.

Workspace offers a complete toolkit for code validation and revision by using OpenAI's GPT-4 Turbo to recommend coding activities within the development environment. This solution streamlines the software development process and is available via a new "Open in Workspace" option on GitHub. Workspace is now in technical preview, thus it is not protected by GitHub's IP indemnity clause. However, GitHub hopes to improve Workspace's functionality and strategy for the market during this time.

This breakthrough coincides with broader conversations about how AI is affecting coding techniques, raising questions about higher error rates and security vulnerabilities. GitHub, on the other hand, is still upbeat about the use of AI in software development, stating that the combination of human and artificial intelligence skills could improve coding productivity and creativity.


thumbnail_pexels-pixabay-270348 (1).webp

pexels-pixabay-270348 (1).webp

GitHub Launches Copilot Workspace: An AI-Driven Evolution in Software Development


Boston Dynamics and Agility Robotics are working on methods to improve how well their bipedal robots manage falls. Boston Dynamics wowed the public last week with two big announcements. The first was the [launch of the electric Atlas](https://techcrunch.com/2024/04/17/boston-dynamics-atlas-humanoid-robot-goes-electric/), which garnered almost five million views in a short period of time. The day prior, they announced the [retirement of the first hydraulic Atlas](https://techcrunch.com/2024/04/16/atlas-shrugged-boston-dynamics-retires-its-humanoid-robot/), commemorating ten years of use with a celebratory film and blooper reel that emphasized the process of learning from mistakes made repeatedly.

Similar to humans, bipedal robots are prone to falls, but what distinguishes their advancement is how well they bounce back. "Everyone falls sometimes, it's how we get back up that defines us," [Agility recently said](https://www.linkedin.com/feed/update/urn:li:activity:7187912431577239552/). Teaching robots to fall safely is another example of this idea in action. Fall safety is a vital skill for both the longevity and functionality of robots.

Robots falling during jobs are a good indicator of real-world testing, according to Pras Velagapudi, the recently hired CTO of Boston Dynamics, and are necessary for long-term operation in these contexts. Human-understood safe fall techniques, like head protection and utilizing limbs to guide the fall, are being adapted for use with robots. This method has been essential in creating robust robots that can bounce back and carry on with their tasks following a fall.

Boston Dynamics has also gained a lot of knowledge from Spot, its quadriplegic robot, which exhibits improved stability but nevertheless stumbles in challenging operational settings. Reducing the frequency of these falls and improving robot resilience have been made possible by the company's acceptance and analysis of them.

Another factor that has helped in this area is agility, especially since their Digit robot now has arms to help with balance and recovery after falls. They even went so far as to show off this ability in a live presentation, where even though the demo had a high success rate, the rare fall offered insightful teaching moments and an opportunity to highlight the robot's rapid recovery capabilities.

Robots must be able to fall safely and rebound on their own. This is essential for their incorporation into "brownfields," or existing industrial environments, where they must function dependably with little assistance from humans. Every fall brings new lessons, advancing the field of bipedal robotics and guaranteeing its longevity in practical uses. Bipedal robots are not only technological wonders but also robust, adaptive beings that are ready for further integration across multiple sectors because of their constant learning from falls.

thumbnail_taiki-ishikawa-PecQftb1ubg-unsplash (1).webp

taiki-ishikawa-PecQftb1ubg-unsplash (1).webp

Boston Dynamics and Agility Teach Bipedal Robots to Fall Safely

In a successful Series B fundraising round, [Parloa](https://www.parloa.com/), a German conversational AI platform that specializes in customer service, raised $66 million. This substantial amount is just one year after it secured $21 million to support its international expansion from a number of European investors. The business is now concentrating more on the US market after opening an office in New York. Thanks in part to its new American headquarters, this calculated move has already drawn the interest of "several Fortune 200 companies" as clients. [Altimeter Capital](https://www.altimeter.com/home), a U.S. venture capital firm well-known for its investments in significant businesses including Uber, Airbnb, Snowflake, Twilio, and HubSpot, is leading the funding round.

While the customer service industry has long been involved in AI and automation, the introduction of sophisticated [large language models](https://en.wikipedia.org/wiki/Large_language_model) (LLMs) and generative AI infrastructure has reignited interest in advanced, intelligent conversational AI technology. Not to be outdone, other major companies in the market have also received a substantial amount of money lately. Kore.ai, for example, secured $150 million from investors, including Nvidia. In addition, Bret Taylor, the former CEO of Salesforce, raised nearly $100 million in February to establish Sierra, an AI-based customer experience platform.

Since its founding in 2018, Parloa has grown significantly, counting Swiss Life and Decathlon among its notable clientele. These companies utilize the platform to automate consumer communications via email and instant messaging. Malte Kosub, co-founder and CEO of the firm, highlights Parloa's distinctive speech-first approach [in an interview with TechCrunch](https://techcrunch.com/2024/04/24/parloa-a-conversational-ai-platform-for-customer-service-raises-66m/), which he thinks distinguishes them by providing the most significant customer experience through AI-driven voice interactions that closely resemble human communication.

Co-founder and CTO Stefan Ostwald says AI has always been a key component of Parloa's strategy. The business creates models especially for speech-to-text applications that are appropriate for customer service scenarios and phone audio quality by combining proprietary and open-source LLMs. Furthermore, Parloa has created a custom telephony infrastructure to lower latency, which is essential for voice automation.

After raising $21 million in its Series A in 2023, Parloa has raised a total of over $98 million to date. With this amount, the company is well-positioned to continue growing in the European and American markets. According to Kosub, the company has tripled its yearly sales over the last three years and, since 2023, has made substantial inroads into the U.S. market, with outcomes that have above projections. The Series B funding round also saw contributions from EQT Ventures, Newion, Senovo, Mosaic Ventures, and La Familia Growth.


thumbnail_pexels-mart-production-7709166 (1).webp

pexels-mart-production-7709166 (1).webp

Parloa Raises $66M in Series B Funding to Expand Conversational AI Platform in the U.S.


[Adobe](https://www.adobe.com/) has revealed Firefly Image 3, which has been showcased at the [Max London conference](https://www.adobe.com/uk/max-london.html), as their most recent breakthrough in AI-powered image production. Adobe claims that Firefly Image 3 is a major advancement in creating realistic graphics, even though previous iterations were criticized for frequently warping photos and failing to catch the nuances of user prompts. This new model, which is now available in the beta versions of Photoshop and the Firefly online app, promises to be able to interpret lengthier, more complicated prompts and sceneries with better lighting and text production accuracy.

Firefly Image 3 is designed to produce more accurate representations of line art, raster graphics, typography, and iconography. According to Adobe, this model performs noticeably better when it comes to depicting crowded areas and individual people with finely drawn features that convey a range of emotions. [Adobe Stock](https://stock.adobe.com/ie/promo/firstmonthfree?gclid=Cj0KCQjw_qexBhCoARIsAFgBletxmyosKdH-y6j-jk5inDRQEPDsWp6Y5qBk5e_ir8Kk3hULMw_IeVYaAiXNEALw_wcB\&ef_id=Cj0KCQjw_qexBhCoARIsAFgBletxmyosKdH-y6j-jk5inDRQEPDsWp6Y5qBk5e_ir8Kk3hULMw_IeVYaAiXNEALw_wcB:G:s\&s_kwcid=AL!3085!3!697186390480!e!!g!!adobe%20stock!1747617084!68493315775\&as_channel=sem\&as_campclass=brand\&as_campaign=ES%7CCPRO%7CStock%7CPURCH%7CAs_Brand_Exact%7CGG%7C%7CEN\&as_source=google\&mv=search\&mv2=paidsearch\&as_camptype=acquisition\&sdid=BMVV34ZM\&as_audience=core\&gad_source=1), a vast royalty-free media repository that is constantly growing and contains licensed and public domain content whose copyrights have expired, is where the training data for Image 3 comes from.

In response to ethical concerns, Adobe launched a scheme to pay contributors to Adobe Stock whose photos are included in training datasets, setting themselves apart from other AI vendors. With this action, competitors hope to steer clear of the contentious practice of training on photographs from any source. Concerns over the usage of possibly copyrighted material in AI-generated photos seen in Adobe Stock for model training have been brought to light by recent studies. In response, Adobe asserts that AI-generated photos make up a very small portion of its training data and go through a stringent moderation procedure to make sure they don't violate copyrights or feature names or characters that are easily recognized.

Apart from its improved ability to create images, Firefly Image 3 adds a number of new features to Photoshop. These include an auto-stylization option and a new style engine that enable a wider variety of colors, backdrops, and subject positions. With the use of an image, users can influence the model and align the tone or color of future generated material by using the Reference Image feature. Additional tools that make use of Image 3's power include Enhance Detail, Generate Similar, and Generate Background. These programs replace backdrops, provide variations for specific photo elements, and sharpen images.

Adobe is demonstrating its commitment to incorporating artificial intelligence (AI) into its core products with the release of these new features in Photoshop, which have been under beta testing in the Firefly online app. In addition to Image 3, the Firefly web app is getting tools called Structure Reference and Style Reference. These tools allow users to generate images that either match the structure of a reference image or take on the style of another image, essentially enabling style transfer under a different mask and improving creative control.

Adobe has said that it will not be changing its pricing structure for the time being, in spite of these technological advancements. Firefly's most affordable premium subscription is still $4.99 a month, which is reasonable when you consider the prices of competitors like Midjourney and OpenAI. Adobe also plans to keep up its generative credit system, watermarking AI-produced content, and its indemnity policy covering copyright disputes connected to generated works. Furthermore, regardless of whether photos are created from scratch or altered using generative features, Content Credentials—which automatically attach metadata to identify AI-generated media—will remain consistent across all images generated in Photoshop and on the web.

thumbnail_szabo-viktor-qmvl6m4Tots-unsplash (1) (1) (1) (1).webp

szabo-viktor-qmvl6m4Tots-unsplash (1) (1) (1) (1).webp

Adobe Launches Firefly Image 3 Offering Unprecedented Image Accuracy

A serious security vulnerability in the [Chirp Systems](https://www.chirpsystems.com/) app has been made [public by the US authorities](https://www.cisa.gov/news-events/ics-advisories/icsa-24-067-01). This flaw permits unwanted remote access to smart home locks. It allows anyone to remotely override any lock in thousands of rental homes across the United States thanks to a smart access control system that is extensively utilized in these residences. Chirp Systems has been made aware of the problem, however it has not been resolved.

The inhabitants use the app to replace their keys, but the [Cybersecurity and Infrastructure Security Agency (CISA)](https://www.cisa.gov/) released a security advice pointing out that the program improperly maintains hardcoded credentials, granting external control of any smart lock that is compatible with Chirp. Because they are easily retrieved and exploited to imitate app functionality, such as remotely locking or opening doors over the internet, hardcoded credentials in apps provide a security risk.

The vulnerability was classified as having a low attack complexity and the potential for remote exploitation, with CISA rating the vulnerability's severity at 9.1 out of 10. Chirp Systems has not been in touch with CISA or the researcher who found the vulnerability in spite of numerous alerts. Although security researcher Matt Brown notified Chirp of the bug in March 2021, it still exists.

After the alert was published, Chirp stated that it had not discovered any data to substantiate its assertions, although it did state that it was working on a patch to fix the problems.

In 2020, RealPage bought Chirp Systems, a company in the expanding property tech sector that provided keyless access. Later that year, [Thoma Bravo acquired RealPage](https://www.thomabravo.com/press-releases/realpage-to-be-acquired-by-thoma-bravo) for a whopping $10.2 billion. RealPage's rent-setting software is currently at the center of legal disputes, and the firm has neither recognized the software's flaws nor disclosed its intentions to alert impacted residents to the security dangers.


thumbnail_sebastian-scholz-nuki-IJkSskfEqrM-unsplash (1).webp

sebastian-scholz-nuki-IJkSskfEqrM-unsplash (1).webp

Security Flaw Found in Chirp Systems' Smart Home Lock App

[Vector databases](https://en.wikipedia.org/wiki/Vector_database) are becoming more and more important as the AI industry approaches a peak of interest and development, drawing the attention of both established database giants and up-and-coming companies. [Large language models](https://en.wikipedia.org/wiki/Large_language_model) (LLMs) and generative artificial intelligence (GenAI) have made it possible for vector databases to grow by effectively managing the complexity of unstructured data, including texts, photos, and videos. This non-predefined style of material is difficult for traditional relational databases like Postgres or MySQL, which arrange structured data into rows and columns. In order to close this gap, vector databases transform unstructured data into vector embeddings, which are numerical representations of the data that maintain its relational contexts and semantic properties. This improves machine learning applications and artificial intelligence capabilities.

In addition to completely changing the way data is handled and stored, vector databases are essential for improving the comprehension and contextualization of talks, which is a key feature of AI models such as [OpenAI's GPT-4](https://openai.com/gpt-4). This technique can quickly match user searches with suitable results based on semantic similarity, which is especially useful for real-time applications like tailored content recommendations on social networks or e-commerce platforms. Furthermore, by supplying extra contextual information not present in the initial training datasets, vector search technology can reduce the likelihood of "hallucinations" in LLMs.

The new wave of investment in the field emphasizes how crucial vector databases are to contemporary AI applications. Vector database startups, like [Qdrant](https://qdrant.tech/lp/high-performance-vector-search/?utm_source=google\&utm_medium=cpc\&utm_campaign=21184756527\&utm_content=160520941865\&utm_term=qdrant\&hsa_acc=6907203950\&hsa_cam=21184756527\&hsa_grp=160520941865\&hsa_ad=696691988577\&hsa_src=g\&hsa_tgt=kwd-1329481093586\&hsa_kw=qdrant\&hsa_mt=e\&hsa_net=adwords\&hsa_ver=3\&gad_source=1\&gclid=CjwKCAjw26KxBhBDEiwAu6KXt9bWEg1DNbKSA__dO22JmWg1-69WPBVAKVpsHlR03kSCsPCieaqe4RoCMZMQAvD_BwE), have received a lot of money, indicating the market's belief in the increasing importance of these technologies. In an interview with [TechCrunch](https://techcrunch.com/2024/04/20/why-vector-databases-are-having-a-moment-as-the-ai-hype-cycle-peaks/), Andre Zayarni, CEO of Qdrant, outlined the critical function vector databases play in handling big datasets by offering a productive and intuitive method of managing vector embeddings. With its recent $28 million fundraising round, the company has solidified its standing as one of the fastest-growing commercial open-source startups.

Furthermore, the industry is not just for startups. Vector search capabilities are becoming more widely available from well-known database companies and cloud service providers, which points to a larger industry trend toward specialist database solutions. Elastic, Redis, OpenSearch, and major cloud platforms such as Microsoft Azure and Amazon AWS are among the companies that are extending their services to incorporate vector search, realizing that vector search can improve data handling and AI application performance considerably.


thumbnail_claudio-schwarz-fyeOxvYvIyY-unsplash (1).webp

claudio-schwarz-fyeOxvYvIyY-unsplash (1).webp

Vector Databases Gain Importance as AI Technologies Advance and Demand Increases


In order to help businesses that wish to stay independent of a single large language model (LLM) supplier, the German startup [Langdock](https://www.langdock.com/de) has successfully raised $3 million in funding, with [General Catalyst](https://www.generalcatalyst.com/) serving as the lead investor. While many big businesses are keen to use AI, many are hesitant to settle on a single core model too quickly. In order to solve this, Langdock provides a chat interface that acts as a go-between for businesses and LLMs, giving freedom in selecting different models without committing to anything long-term. The Berlin based startup raised seed money recently from Y Combinator, La Famiglia, and a number of well-known German businessmen, including Rolf Schrömgens (Trivago) and Hanno Renner (Personio). Co-founder and CEO of Langdock, Lennard Schmidt, stressed that the company's interface offers choices between using alternative basic, open-source, or proprietary models, preventing [vendor lock-in](https://en.wikipedia.org/wiki/Vendor_lock-in). In order to enable the safe and controlled integration of LLMs into business operations, the organization also prioritizes maintaining compliance with European regulations. Security solutions and on-premises and cloud-based choices are among Langdock's other products. Clients like Merck recognize the potential of the technology to improve efficacy and efficiency, and the company has implemented the interface for its 63,000 workers. Although there are other companies in this sector that are creating LLM interfaces, such as Paris-based [Dust](https://dust.tt/), Langdock's approach is more versatile and accessible to all employees.

thumbnail_priscilla-du-preez-OEdkPaxYMXU-unsplash (1).webp

priscilla-du-preez-OEdkPaxYMXU-unsplash (1).webp

Langdock Secures $3M Funding to Improve AI Usage in the Corporate World 


Tesla has initiated a recall of all 3,878 Cybertrucks it has distributed so far due to an issue where the accelerator pedal may become stuck, increasing the risk of accidents. The National [Highway Traffic Safety Administration](https://www.nhtsa.gov/) made this [announcement](https://www.nhtsa.gov/vehicle/2024/TESLA/CYBERTRUCK#recalls), which signifies a difficult time for Tesla. The company lost two key executives and fired nearly 10% of its workforce earlier last week. Not long after, Tesla forced a re-vote on CEO [Elon Musk](https://en.wikipedia.org/wiki/Elon_Musk)'s huge compensation package, which had earlier in the year been declared unconstitutional by a court ruling.

Reports of accelerator pedal issues with the Cybertruck have surfaced in recent weeks. Tesla responded by temporarily stopping the car's delivery. Musk noted Tesla's cautious approach in a statement on the social media platform X. The company informed the NHTSA that there have been no reported crashes or injuries linked to the defect. However, Tesla has acknowledged to the NHTSA that the pedal can become dislodged and trap itself in the surrounding footwell trim.

A consumer reported the issue for the first time on March 31 and then again on April 3. Following testing, Tesla announced the recall on April 12 after learning that the pedal's attachment was compromised by an unapproved modification including the use of lubricant during assembly. Tesla has now promised to replace or fix the accelerator pedals on all Cybertrucks that are impacted, and it has started to install a revised pedal in freshly built cars. They are also taking care of the problem in cars that are in route or at delivery hubs.

This recall is not the first for the Cybertruck, despite the fact that it started shipping late last year; the prior one was quite small and included changing the software to change the warning lights' font size. The truck was first unveiled by Tesla [in 2019](https://techcrunch.com/2019/11/21/behold-the-tesla-cybertruck-is-here/).
