Weekly analysis of AI & Emerging Tech news from an analyst's point of view.
1️⃣ServiceNow agentic AI acquisitions - Movewors and Cuein - set pace for enterprises
Details:
On Jan. 17, ServiceNow bought Cuein, makers of AI conversation data analysis software, for an undisclosed amount. On Monday, ServiceNow publicized its $2.85 billion deal for agentic AI software vendor Moveworks, the largest acquisition in ServiceNow's history. The purchase will give ServiceNow a front-end AI assistant and enterprise search tools to combine with its Now Platform. Moveworks and ServiceNow already have multiple joint customers. Moveworks is in FedRAMP Marketplace--a first for agentic AI platforms--and counts HP, Databricks, Unilever, Marriott and Toyota as customers. Here's a look at Moveworks' architecture.
Key items about the ServiceNow acquisition of Moveworks include:
- Moveworks agentic AI platform has nearly 5 million employee users in 18 months.
- 90% of Moveworks customers have deployed it to all employees.
- The companies have about 250 mutual customers.
Analysis:
With these moves and the Jan. 29 release of its AI Agent Orchestrator, ServiceNow has a chance to set the tone for the next major evolution of AI for the IT operations market. ServiceNow has already become the de facto enterprise standard in many services and support workflow management. While ServiceNow's agentic AI and automation strengths in their platform have been around for a while, they needed a way to build front-end AI assistants quickly and easily, and Moveworks helps them in that angle.
Moveworks also brings AI-based enterprise search to agentic AI workflows. Along with some 500 employees with expertise in AI. Particularly in regulated industries, where information needs to follow heavy workflow governance components, ServiceNow is a leader.
My full analysis and quotes can be seen in this TechTarget article: https://www.techtarget.com/searchitoperations/news/366620527/ServiceNow-agentic-AI-acquisitions-set-pace-for-enterprises
2️⃣New Databricks tools tackle lingering GenAI accuracy issues
Details:
Databricks on Monday unveiled new features, including governance and monitoring capabilities, designed to enable customers to scale generative AI beyond pilot projects and low-risk applications. While studies show widespread interest among enterprises in developing and deploying generative AI tools, they also show that concerns regarding the accuracy of generative AI outputs and data security prevent many organizations from putting such tools into production. To alleviate such concerns, Databricks introduced centralized governance for all AI models through the AI governance framework Mosaic AI Gateway and real-time performance observability with Lakehouse Monitoring for Agents, among other features.
Highlights:
- Custom large language model (LLM) provider support in Mosaic AI Gateway, now in public preview, so that customers can govern all their AI models in a central location, including open source and proprietary SaaS models.
- Lakehouse Monitoring for Agents, a feature in beta testing that deploys MLflow Tracing and LLM judges so that users can track the performance of AI agents.
- An API in public preview that enables developers to integrate Genie, a conversational interface that lets users interact with data using natural language, into custom-built applications and productivity platforms.
- Batch inferencing capabilities with Mosaic AI Model Serving to simplify infrastructures needed to integrate unstructured data and train models.
Analysis:
The tools -- now in various stages of testing and preview -- do address concerns related to generative AI accuracy. However, whether enterprises want to use Databricks for governance, observability and other features that address accuracy remains to be seen. The features address the enterprise adoption of GenAI, for sure. But ... almost every platform provider, every hyperscaler, every model provider and many startups are working to provide similar solutions. While Databricks has a leg up with its data platform, I'm not sure they will fully convert those with enterprise AI needs to their platform.
Each feature is purposeful, with centralized AI governance potentially the most significant. Centralized governance for all AI models is an interesting solution, Integrating and managing both open source and proprietary SaaS models in one place and the ability to set governance policies centrally can be compelling for large enterprises which have distributed units that work independently on AI models and consumption.
Things like usability and user experience have been a major customer complaint. And integration with more third-party tools such as vector databases, better optimization of AI workloads, fast and efficient analytics for hybrid environments, and cost efficiency are areas they could improve.
In addition, real-time monitoring could spur more enterprise adoption of generative AI.
My full analysis and quotes can be seen in this TechTarget article: https://www.techtarget.com/searchdatamanagement/news/366620484/New-Databricks-tools-tackle-lingering-GenAI-accuracy-issues
In other news,
-
US plans to use AI to identify foreign students supporting Hamas and revoke their US visas.
Per Reuters, the US State Department is exploring using AI to identify the foreign students who support Hamas and other outlawed organizations, revoke their visas, and deport them. This has raised significant concerns among human rights advocates and free speech supporters. The program, called "Catch and Revoke," will use AI to continuously monitor the social media accounts of tens of thousands of foreign student visa holders to watch for potentially pro-Hamas statements.
Numerous students and groups across the US have organized protests expressing concern over the plight of Palestinians in Gaza. Secretary of State Marco Rubio stated on social media that the US has zero tolerance for foreign visitors who support terrorism, and violators will face visa denial, revocation, and deportation.
-
Microsoft develops a new AI strategy. The relationship with OpenAI seems to have soured.
Microsoft is concentrating on its own models, MAI, rather than fully relying on OpenAI. While OpenAI models are the model of choice for Microsoft Copilot and other Azure initiatives, Microsoft is planning to release MAI API this year. Microsoft will also make MAI the default choice for all its applications, including Microsoft Copilot. This will avoid heavy reliance on OpenAI models. Microsoft is also experimenting with models from Anthropic, xAI, DeepSeek, and Meta's open-source offerings to diversify its AI mode offerings. Apparently, training the MAI model took more than a year, and in that process, Microsoft also lost a lot of senior executives. It will be interesting as Microsoft has invested over $13 billion in OpenAI.
-
Many AI companies remove DEI claims from their AI model disclosure and web pages.
Google updated its "Responsible AI and Human-Centered Technology (RAI-HCT)" webpage, removing references to "diversity" and "equity." Previously, the webpage used terms like "marginalized communities," "diverse," "underrepresented groups," and "equity" to describe the team's work on AI safety, fairness, and explainability. Phrases like "all," "varied," and "numerous" are now substituted for "diverse.” Google announced in February that it would be abandoning its diversity hiring goals and reviewing its diversity, equity, and inclusion (DEI) programs.
AI company Anthropic quietly removed its commitments to AI safety from its website and from Anthropic's transparency center, which lists the company's "voluntary commitments" to responsible AI development. While not legally binding, these commitments pledged to share information and research on AI risks (including bias) with the government.
Many non-AI companies changed their DEI programs recently as well. The list includes Walmart, PepsiCo, Tractor Supply, Meta/Facebook, Ford, Toyota, Lowe’s, and Target, among many others. However, some companies are standing by their support for DEI programs - Apple, Ben & Jerry’s, Costco, Delta, and JPMC, among the few.
-
McDonald's improves restaurant operations with AI technology.
McDonald's is leveraging artificial intelligence (AI) to improve operations across its 43,000 global restaurants, aiming to ease employee workloads. According to the Wall Street Journal, McDonald's began rolling out an edge computing platform in select US restaurants last year, with plans for broader deployment by 2025. This technology enables innovations such as computer vision to verify order accuracy via kitchen cameras before customer pickup. It is also testing AI-powered automated ordering, streamlining the drive-thru process, and improving order accuracy and efficiency. McDonald's has installed sensors to collect real-time data, predicting potential failures in fryers, ice cream machines, and other equipment. This proactive maintenance minimizes service disruptions caused by equipment malfunctions. Generative AI "virtual managers" are also experimented with to significantly improve scheduling efficiency.
-
AI coding assistant cursor may raise funding at $10 billion valuation.
Anysphere is considering a new funding round at a valuation nearing $10 billion per Bloomberg. They just raised $100 million at a $2.5 billion valuation. Per the New York Times, Anysphere's previous valuation was based on its $1 million annual recurring revenue (ARR). This is insane. Their competitor Codeium, a company developing the AI coding editor Windsurf, is currently seeking funding at a valuation near $3 billion. Their valuation is approximately 70 times its $40 ARR.
-
AI21 labs releases Jamba 1.6: Long-form text processing that supports multiple languages.
AI21Labs recently released its latest large language model, the Jamba 1.6 series. Jamba has 2.5 times the inference speed of comparable transformer models. The Jamba 1.6 series includes Jamba Mini (1.2 billion parameters) and Jamba Large (9.4 billion parameters). This model is released under the Jamba Open Model License, allowing for both research and commercial use under the terms of the license. Jamba 1.6 series' knowledge cutoff date is March 5, 2024. It supports multiple languages, including English, Spanish, French, Portuguese, Italian, Dutch, German, Arabic, and Hebrew, ensuring global accessibility.
-
AI agent Manus takes off; Invitation codes resell in the black market for a high price.
The world's first general-purpose AI agent Manus has garnered significant attention in the tech world. Manus possesses the ability to think independently and execute complex tasks, delivering complete results and demonstrating powerful versatility. It can handle daily tasks, as well as conduct in-depth market research, personalize travel plans, and much more. Because of the popularity, the demand has hit through the roof. In the secondary market, Manus invitation codes are priced anywhere from $500 to as high as $8000. While this demand might subside soon, it is interesting to note that the invite codes are sold in the black market.
-
OpenAI launches "PhD-Level" AI agent with a monthly fee of $20,000.
OpenAI recently announced a "PhD-level" AI agent designed to meet the high-end needs of industries such as finance, healthcare, and manufacturing. The AI agent boasts a monthly fee of up to $20,000, offering various service types with pricing based on the economic value created for clients. While the high cost has prompted some jokes, OpenAI clearly targets large enterprises, not individual users.
-
Mistral unveils Mistral OCR.
Mistral AI officially launched its latest document recognition model, Mistral OCR. Mistral OCR supports the accurate extraction of text from complex PDFs, images, tables, mathematical formulas, and multilingual documents, surpassing Google Document AI and Azure OCR in both speed and accuracy. It is capable of processing up to 2000 pages per minute. Mistral OCR has dramatically improved recognition rates, achieving nearly 99% accuracy in multiple languages. This performance is not only evident in multilingual text processing but also in the recognition and formatted output of complex mathematical formulas, meeting the urgent needs of academic and professional fields. It is priced at $1 per 1000 pages for the API and the bulk processing is available at $1 per 2000 pages!
#GenerativeAI #GenAI #AI #LLM #Grok #OpenAI #Meta #Amazon
