The Era of ‘Enterprise’ AI Agents Begins

The Era of ‘Enterprise’ AI Agents Begins

The enterprise software landscape is undergoing a dramatic shift. While SaaS providers continue to announce new AI-powered features , many forward-thinking companies are taking matters into their own hands. By developing in-house AI agents, they’re not just innovating—they’re slashing costs by millions, even billions, wiping out the need for expensive and outdated SaaS systems.

Case in point: Klarna.

The fintech giant recently made headlines for ending its partnership with Salesforce and Workday—two of the most dominant enterprise systems of record.?

Klarna’s approach? The company developed AI agents that could observe the inputs and outputs of these systems and eventually replicate their functions, creating a “digital twin”. Once these AI-driven replicas were functional, Klarna shut off the legacy systems, saving tens or even hundreds of millions of dollars in the process.

Industry expert Chamath Palihapitiya , in a recent episode of All-In podcast, lauded the move, and said it takes “enormous technical strength” and “tremendous executive leadership”, to do something like this. He said replacing deeply embedded enterprise software solutions like Salesforce and Workday isn’t for the faint-hearted. It demonstrates a clear shift in how companies are rethinking their tech stacks.

Agreed. Recently, Mark Zuckerberg questioned the authenticity of self-proclaimed tech companies, saying that a true tech company needs technically skilled leaders at all levels, from the CEO to the board and management team.?

“All these companies that called themselves technology companies were not really set up that way. It’s like the companies I was talking about—it’s like they, you know, the CEO wasn’t technical, the board of directors had no one technical on it, they had like one dude on the management team who was the head of engineering who was technical, and like everyone else wasn’t,” said Zuckerberg. “If that’s your team, then you’re not a technology company.”?

As more companies strive to become true technology companies, the latest advancements—particularly those developed by OpenAI, Google, Meta and other source initiatives—are making this transition easier than ever, thereby reducing reliance on traditional SaaS providers such as Salesforce, Oracle, SAP, and others.

OpenAI o1, for instance, ranks in the top 11% in competitive programming (Codeforces), stands among the top 500 students in the US in a maths competition (AIME), and surpasses human PhD-level accuracy in solving physics, biology, and chemistry problems (GPQA). There will be many more to follow, with the likes of? Google DeepMind , Anthropic and Meta , also launching advanced AI models in the coming months, taking agentic experiences to a whole new level.?

“We have these five levels of AI—the first was chatbots; the second, which we’ve just reached now, is reasoners; third is agents; the fourth is innovators, the ability to figure out new scientific information; and the fifth is full organisations,” said OpenAI chief Sam Altman, in an interview with T-Mobile president and chief executive officer Mike Sievert .?

He said the move from one (chatbot) to two (reasoners) took a while, but the latter enables level three (agents) relatively faster. “The agentic experience that we expect this technology to eventually enable—I thinkwill be quite impactful,” said Altman, adding that OpenAI o1 is like GPT-2 for reasoning . Developers and enterprise customers will eventually figure out ways to use it just like they did with the GPT-series of models.?

The End of Egregious Costs: The high costs of outdated enterprise systems are becoming increasingly troublesome. For example, NYC’s public college system recently paid Oracle $600 million to build a course management portal using Oracle’s PeopleSoft suite . “It’s built on top of Oracle’s PeopleSoft suite , which they refused to customise without an extra $400 million (to hit $1 billion ),” said the user on X, saying that New Yorkers are left with a subpar system and continue to pay over $5 million per year just for “hosting”.??

Palihapitiya believes that such projects would be impossible in the near future, as current AI systems offer far cheaper and more efficient solutions.?

The Future of Enterprise SaaS: The days of relying on expensive, monolithic platforms may be numbered, as AI-driven solutions offer highly customised, cost-effective alternatives. While enterprise providers are working hard to deploy AI agents on a massive scale , the real revolution may be happening within companies themselves, where internal teams are developing AI solutions tailored to their exact needs.

This raises a key question for the future: Will the SaaS giants or so-called technology enablers continue to dominate, or will we see a new era of in-house AI solutions that challenge their very existence?

One thing is certain—enterprises are no longer passive consumers of off-the-shelf software. They’re becoming active creators, empowered by AI to build, adapt, and optimise their systems in ways that were unimaginable just a few years ago.

Stay tuned as AIM continues to track these seismic shifts in enterprise technology.?


Top Stories of the Week >>?

Wayfair Goes All-In on GenAI; Partners with Google, OpenAI

Wayfair is undergoing a generative AI makeover, partnering with Google Cloud, OpenAI and Anthropic to redefine the future of e-commerce. With AI models like Gemini, GPT-4o, and Claude, Wayfair is optimising everything from product search to personalised customer experiences, allowing even non-technical teams to harness AI’s full potential.?

Check out the full story here .?

Why Google Will Make a Better Model Than OpenAI’s o1

While OpenAI celebrates the launch of o1 , Google is quietly preparing to surpass it with the upcoming release of Gemini 2, which promises better reasoning capabilities, longer context windows, and multimodal functionality. As competition heats up between these AI giants, Google’s advanced reinforcement learning techniques and chain of thought integration could give it the edge.

Click here to read the full article.


People & Tech >>?

The Final Battle: RAG vs Fine-Tuning

Parita Desai , lead architect at Fractal, in conversation with AIM, highlighted the difference between RAG and fine-tuning. She said that while RAG provides real-time, cost-efficient responses by incorporating live data, fine-tuning offers domain-specific accuracy through labelled datasets. The choice between the two depends on the specific problem an organisation aims to solve. Read more here.?

SambaNova’s 24TB Memory Powers 12T Parameters

In an exclusive interaction with AIM, SambaNova’s chief architects, Sumti Jairath and Raghu Prabhakar , highlighted how the company sets itself apart from competitors like Cerebras and Groq with its innovative SN40L chip and three-tier memory architecture.?

Unlike others, SambaNova can handle up to 12 trillion parameters with just one rack, thanks to its 24TB memory, offering unmatched scalability and efficiency. As enterprises increasingly seek to run multiple AI models, SambaNova’s full-stack solutions provide a cost-effective alternative, positioning it as a leader in the AI scalability race. Read more here.?

Why Bhavish Aggarwal is Bullish About Building ‘Made in India’ AI Chips by 2026

At the recent ‘Sankalp 2024’ event, Ola founder Bhavish Aggarwal made bold announcements about developing India’s first homegrown AI chips by 2026, highlighting the need for the country to own its tech infrastructure—from AI chips to cloud systems. Click here to read the full story.


AIM Videos >>?

AIM recently visited Ford’s Chennai office, where the Ford Business Solutions team is harnessing the power of AI and big data to revolutionise mobility. From connected, electric, and self-driving cars to global operations like manufacturing and marketing, the team is at the forefront of innovation.?

Ford is shaping the future of the automotive industry.

Check out the video below:?


AIM Events >>?

Cypher 2024 USA?

AIM is bringing Cypher 2024 to Silicon Valley, and we are already feeling the excitement peak as we get ready to host India’s biggest AI conference this week, in Namma Bengaluru.?

Next up: Santa Clara, California. Date: November 21-22, 2024

Register now!?

The DECODE Webinar is Back

Join industry experts Luis Carlos Cruz Huertas , executive director at DBS Bank, and Sriram Venkat , APAC head of data & analytics services at Google, as they share their deep expertise in cloud technology, data integration, and AI to break down data silos and drive seamless business operations in the second episode of the DECODE Webinar series.

Register now for the webinar.


AI Bytes >>?

  • Princeton Digital Group (PDG) is investing $1 billion to expand its AI-ready data centres in Mumbai and Chennai, increasing India’s data centre capacity to 230 MW as part of a broader $5 billion initiative to meet the growing demand for AI-driven cloud infrastructure across Asia.
  • OpenAI has made its advanced reasoning models, o1 preview and o1 mini, available to all ChatGPT Enterprise and ChatGPT Edu users , offering enhanced decision-making capabilities to businesses and educational institutions for complex tasks in fields like consulting, engineering, science, and tutoring.
  • Amazon has introduced Project Amelia, an AI-powered assistant for sellers, leveraging Amazon Bedrock’s AI models to provide instant insights, sales trends, and business metrics, making it easier to manage and grow their businesses. Learn more about Amelia here.?
  • NVIDIA and G42 recently partnered to launch the Earth-2 platform , using AI-enhanced, high-resolution simulations to improve climate and weather predictions. Supported by G42’s new Climate Tech Lab in Abu Dhabi, the collaboration will leverage over 100 petabytes of data to create tailored climate solutions and advance global sustainability efforts.
  • Intel is set to release its highly anticipated Gaudi 3 AI accelerator this week , designed to deliver double the power efficiency of its predecessor and run AI models 1.5 times faster than NVIDIA’s H100 GPU.?

要查看或添加评论,请登录

AIM的更多文章

社区洞察

其他会员也浏览了