journalfront

Software

Amazon AWS SageMaker Unifies Data Control in 2024

2024-12-03

Amazon Web Services (AWS) has been at the forefront of cloud computing for nearly a decade. Their platform, SageMaker, has been a key player in creating, training, and deploying AI models. This year, however, the focus shifted towards streamlining. At the re:Invent 2024 conference, AWS unveiled SageMaker Unified Studio, a revolutionary tool that brings together data from across the organization in one place.

Unveiling SageMaker Unified Studio

SageMaker Unified Studio is a single destination for data discovery and work. It combines tools from various AWS services, including the existing SageMaker Studio, to assist customers in preparing and processing data for model building. As Swami Sivasubramanian, VP of data and AI at AWS, stated, "We are witnessing a convergence of analytics and AI, and SageMaker Unified Studio provides the necessary tools in one place."Customers can publish and share data, models, and other artifacts within their teams or the entire organization using SageMaker Unified Studio. The service offers data security controls and adjustable permissions, along with integrations with AWS' Bedrock model development platform.AI is seamlessly integrated into SageMaker Unified Studio through Q Developer, Amazon's coding chatbot. It can answer questions such as "What data should I use to gain a better understanding of product sales?" or "Generate SQL to calculate total revenue by product category." AWS explained in a blog post that Q Developer supports development tasks like data discovery, coding, SQL generation, and data integration within SageMaker Unified Studio.

Expanding the SageMaker Product Family

In addition to SageMaker Unified Studio, AWS launched two small yet significant additions to its SageMaker product family. SageMaker Catalog allows admins to define and implement access policies for AI apps, models, tools, and data using a single permission model with granular controls. SageMaker Lakehouse, on the other hand, provides connections from SageMaker and other tools to data stored in AWS data lakes, data warehouses, and enterprise apps.AWS emphasizes that SageMaker Lakehouse works with any tools compatible with Apache Iceberg standards, an open-source format for large analytic tables. Admins have the option to apply access controls across data in all the analytics and AI tools that SageMaker Lakehouse interacts with.

Improved Integration with SaaS Applications

In a related development, SageMaker now integrates better with software-as-a-service applications. Thanks to these new integrations, SageMaker customers can access data from apps like Zendesk and SAP without the need for extraction, transformation, and loading. AWS wrote, "Customers often have data spread across multiple data lakes and a data warehouse. A simple way to unify this data would benefit them. Now, they can use their preferred analytics and machine learning tools on their data, regardless of where it is physically stored, to support various use cases including SQL analytics, ad-hoc querying, data science, machine learning, and generative AI."

Product teams frequently accumulate a plethora of screen recordings and screenshots that go unutilized. Traditionally, video production from these sources has been a time-consuming and costly endeavor. However, two former Microsoft employees took the initiative and founded Lica, an innovative AI tool that simplifies the process of creating tutorials and product videos from screenshots and screen recordings.

Founders and Their Backgrounds

Lica was established in 2023 by Priyaa Kalyanaraman and Purvanshi Mehta. Kalyanaraman, with her extensive experience working as a product manager at Microsoft, Snap, and Waymo, and her work on products like PowerPoint and Microsoft Designer, including adding AI features, recognized the significant demand from business users for AI-assisted video creation. Mehta, who worked at Microsoft as a data scientist and project manager, was also engaged in building projects for content personalization. A mutual friend introduced them, and they saw an opportunity to build a more effective tool for video storytelling.

How Lica Works

Lica offers a powerful set of features. It has a tool that enables users to build product videos or explainers directly from screen recordings. The tool automatically adds transitions, background music, and effects, enhancing the visual appeal of the videos. Additionally, users have the freedom to add manual elements like text for narration. Through prompts, they can guide the AI assistant to give the video a specific voice, such as a Gen-Z language tutorial-style video. Once the video is generated, further edits can be made through prompts, including adjusting the voiceover tone, adding captions, changing the language and style, and selecting music.The founders emphasize that Lica understands design aesthetics well. Even when presented with unusual design or color choices, the tool ensures that the final product looks pleasant and not off-putting. The AI assistant consists of two models - an orchestrator that combines different parts of the presentation and selects the best narration voice, and a layout generator that manages the appearance of different screen recording or text elements on the screen. For other parts like audio generation, a mix of open-source and close-source models is used.

Free and Paid Tiers

Lica currently provides a free tier, allowing users to create 10 videos with a limit of 3 minutes per video and 3 downloads per month. For those who need more extensive capabilities, a monthly subscription of $49 offers unlimited 10-minute videos and 10 downloads, along with access to branded templates.

Future Expansion

While the current focus is on product and tutorial videos, in the coming months, Lica aims to expand its AI assistant to handle more formats of videos, such as marketing, presentation, social media, and investor pitches. Although it may not have direct competitors in the exact form, companies and startups often use a combination of tools like Zoom calls, screen recordings, and polishing tools like Loom. AI-centric startups like D-ID and Syenthesia have used avatars for tutorial videos or internal messaging.Accel's Sameer Gandhi believes that Lica stands out with its combination of a high-quality team and a product approach that combines advanced AI capabilities with intuitive design. It allows users to maintain creative control while benefiting from AI-assisted features, positioning it uniquely to address a key market need that other solutions have yet to fully capture.

Clarifai, a prominent name in the AI industry, has undergone significant transformations. Once associated with controversial projects, it has now shifted its focus to building AI tools for enterprise and government customers. This shift has led to the unveiling of a new chapter in its trajectory, enabling users to orchestrate and coordinate compute and resources from a single control panel.

Unlock the Power of Unified AI Computing with Clarifai

From Controversy to Enterprise Focus

Clarifai, founded in 2013 with an initial focus on computer vision, has come a long way. In its early years, it was involved with the Pentagon's Project Maven, which raised questions about AI ethics. However, over the past few years, the company has redirected its efforts towards serving enterprise and government customers. This shift has allowed it to build a reputation as a vendor-agnostic full-stack AI tools provider, helping with tasks like data labeling, model training, and workflow management.Today, Clarifai's vendor-agnostic AI life cycle platform offers automated management, allocation, and optimization of computing resources. As Matthew Zeiler, the founder and CEO, explained, this helps customers scale their AI more efficiently while making better use of their existing compute resources. The company's control center dashboard, released in October, allows companies to view all their compute sources on one platform, regardless of whether they are in the cloud or on-premise machines. Today's announcement adds another layer of functionality to this system, enabling customers to manage multiple clusters from a central place in a secure way.

Streamlining Costs and Resources

One of the key advantages of Clarifai's platform is its ability to bundle smaller AI models from different sources. This gives customers more control over their compute resources and helps them manage costs more effectively. As Zeiler mentioned, the cost of compute in AI is extremely high, and this solution provides a way to handle it in a more streamlined manner. By connecting different compute sources and using a set of tools, customers can optimize their AI operations and achieve better results.For example, OpenTable, Siemens, and Canva are among the companies that have benefited from Clarifai's services. The startup has built more than 1.5 million AI models, demonstrating its expertise and capabilities in the field. With its VC funding of over $100 million, including a $60 million round in 2021 led by NEA and valued at $775 million, Clarifai is well-positioned to continue its growth and innovation in the AI space.In conclusion, Clarifai's journey from controversy to enterprise focus has been remarkable. Its unified control platform and focus on streamlining costs and resources are set to make a significant impact on the AI industry. As the company continues to evolve, it is likely to play a crucial role in shaping the future of AI computing.