Kore.ai on Wednesday launched what amounts to a ground-up reinvention of its core technology: the Artemis edition of its Agent Platform, a system designed to let enterprises build, govern, and optimize AI agents using AI itself — compressing what has traditionally been months of engineering work into days.
The platform arrives at a moment when every major technology vendor — from Microsoft and Salesforce to Google and ServiceNow — is racing to become the default infrastructure for enterprise AI agents. Kore.ai's answer to that crowded field is a bet on neutrality, a proprietary intermediary language for defining agents, and a philosophy that AI, not human developers, should do most of the heavy lifting.
"We're trying to change the paradigm about how people design, build, deploy and optimize agentic AI applications," Raj Koneru, the company's founder and CEO, told VentureBeat in an exclusive interview ahead of the launch. "The whole theme that we are now coming out with is you do AI with AI — you design with AI, you build with AI, you test with AI, you deploy with AI, manage with AI, and optimize with AI."
At the technical core of the Artemis platform sits Agent Blueprint Language (ABL), a compiled, declarative language built on YAML that standardizes how AI agents, workflows, and multi-agent systems are defined, validated, and governed. Kore.ai describes it as an intermediary layer that sits between the natural-language instructions a business user might provide and the production infrastructure where agents actually run.
ABL comes with its own parser, compiler, and runtime. It supports six built-in orchestration patterns — supervisor, delegation, handoff, fan-out, escalation, and agent-to-agent federation — that govern how multiple agents coordinate on complex tasks.
Koneru framed ABL as addressing a fundamental gap in the current AI landscape. "There's a lot of value in generating code, and that code is used by developers to build applications," he said. "What we saw is a gap between generating code and actually running it on infrastructure — with the deployment, version management, governance, and observability that production requires."
Because ABL artifacts are YAML-based, they can be stored in GitHub, version-controlled through CI/CD pipelines, and reviewed by both developers and business stakeholders — a design choice intended to bridge the divide between no-code platforms and traditional software engineering. "The final artifact is ABL, a YAML-based construct — you can put it in GitHub, you can version-control it," Koneru said. "It gives business people, developers, and IT a single standard to build on."
The second major innovation is Arch, an AI system that translates business requirements into production-ready ABL. Users provide specifications, data sources, and business rules in natural language. Arch then designs the multi-agent topology — selecting from the platform's six orchestration patterns — generates the ABL code, produces test data, deploys the application, and monitors it in production.
Critically, Arch also handles optimization. It observes whether deployed agents are meeting their goals, identifies where and why they fall short, and automatically regenerates and redeploys refined ABL to improve performance.
"Think of it this way," Koneru explained. "In the beginning, I wanted 50% automation for a particular use case. I'm getting 30%. Because of that cycle of optimization, it moves the needle to 50% by adjusting the application based on actual usage data."
This closed-loop approach — design, build, test, deploy, manage, optimize — is Kore.ai's bid to differentiate from both the no-code configuration platforms that dominated the previous era of chatbot development and the pro-code frameworks emerging from companies like Anthropic and OpenAI, which Koneru argues place too much burden on individual developers. "So that's a paradigm shift in the way AI agents have been built up until now," he said, "either with no code, configuration-based platforms — and we were one of them — or pro code capabilities that you get with Cloud code or a Codex or something else, which then puts the onus on the developer to build a platform for themselves."
Perhaps the most architecturally significant element of the Artemis platform is what Kore.ai calls its Dual-Brain Architecture: two cognitive engines — one for agentic reasoning powered by large language models, the other for deterministic execution of business rules — operating in parallel through shared memory within a single runtime.
This design reflects a hard lesson Kore.ai has learned from more than a decade of deploying AI in banking, healthcare, insurance, and telecommunications. In those environments, leaving all decision-making to a language model is a non-starter.
"Enterprises are not going to completely relegate decision-making to a model," Koneru said. He drew a sharp contrast with newer AI-native startups: "A number of the AI-native companies that have emerged recently, especially in Silicon Valley, are essentially frameworks built as a wrapper around an LLM. That means much of the decision-making is left to the model — you're heavily reliant on it, and the model itself is the one implementing the guardrails."
Kore.ai's approach flips that. Guardrails — both input and output — are enforced at the platform layer, not by the model. Evaluations run inside the platform's governance engine. Business rules can execute deterministically when precision matters, while the LLM handles conversational responses and reasoning where appropriate. In a healthcare scenario where an AI agent is processing prescription refills for millions of consumers, or in a banking environment where an agent is advising clients on portfolio management, the consequences of a hallucinated response or an improperly executed workflow are severe. Kore.ai is positioning the Dual-Brain Architecture as the engineering answer to a trust problem that has slowed enterprise AI adoption across regulated sectors.
Artemis launches initially on Microsoft Azure, integrating natively with Microsoft Foundry, Microsoft Agent 365, Entra ID, and the Microsoft Graph API. Kore.ai is a launch partner for Agent 365 and is working toward becoming a native Azure service within Azure Foundry.
The Microsoft partnership runs deep. Koneru described multiple co-build initiatives spanning the past year: agents built on Kore.ai's platform can run on Azure Foundry using its models and infrastructure; Kore.ai's AI for Work product integrates with Microsoft Copilot so that enterprise data and agentic workflows surface directly in the Copilot interface; and AI for Service integrates with Dynamics 365 as a joint go-to-market offering.
"There is a deep relationship," Koneru said. "In fact, I'm at their CEO Summit, and then for the next three days."
Stephen Boyle, CVP of Enterprise Partner Solutions at Microsoft, offered support for the partnership in the Artemis press release, noting that the platform "integrates with Microsoft Foundry and Microsoft Agent 365, giving customers a governed environment to build, deploy, and operate AI agents."
Yet Kore.ai simultaneously pitches itself as the vendor-neutral alternative to Microsoft and its peers — a tension the company addresses head-on. "All of the vendors or tech companies that you mentioned have a legacy that they're trying to protect," Koneru said when asked why a CIO should choose Kore.ai over an incumbent. "There's an inbuilt lock-in to their legacy, whether that's a Salesforce application, ServiceNow application, Microsoft Azure cloud, or whatever." The platform supports 175 different AI models — including those from OpenAI, Anthropic, and open-source providers — deploys across Azure, AWS, Google Cloud, and on-premises environments, connects to any data source via tool calling or MCP, and delivers across more than 40 voice and digital channels.
Kore.ai's claims about enterprise readiness are backed by deployments that rank among the largest AI implementations in the world.
One of the largest pharmacy chains in the United States — which Koneru declined to name but described in enough detail to make identification straightforward — receives approximately 750 million calls from consumers annually. The chain signed with Kore.ai at the end of March 2025, deployed on its own infrastructure, had half of its 9,000 stores live within three months, and reached full deployment across all stores within six months.
"The speed at which they were able to build out very complex functionality — which requires understanding what the prescription is all about, being able to answer questions about them, then tying it to their backend systems to fill the prescription, refill it — all of those processes was done essentially," Koneru said.
A second example involves the world's second-largest investment bank, which deployed Kore.ai's AI for Work product to 135,000 employees and contractors. The bank uses the platform to give more than 30,000 financial advisors access to proprietary research and client portfolio data through a conversational interface, with agentic workflows handling routine tasks. The deployment went from initial users to global rollout within a year. A third customer — a major semiconductor manufacturer with 35,000 employees across multiple countries and languages — deployed AI for Work starting with HR use cases like onboarding, benefits management, and performance reviews, with backend integration to Workday, and has since expanded into IT, legal, and facilities management workflows.
The Artemis launch lands in one of the most fiercely contested markets in enterprise technology. Microsoft's Copilot Studio and Agent 365, Salesforce's Agentforce, Google's Vertex AI Agent Builder, and ServiceNow's AI Agents all target the same CIO budget. Meanwhile, a wave of well-funded startups — from established players like UiPath to AI-native entrants — is flooding the market with agent-building frameworks and platforms.
Kore.ai's competitive position rests on several pillars. The company has earned consistent recognition from major analyst firms: it has been named a Leader in the Gartner Magic Quadrant for Enterprise Conversational AI Platforms (positioned highest for Ability to Execute, according to the company), a Leader in the Forrester Wave for Cognitive Search Platforms with the highest ranking in the Strategy category, and an Emerging Leader in Gartner's Emerging Market Quadrants for both Generative AI Engineering and GenAI Applications. Everest Group has also positioned Kore.ai as a Leader in its Agentic AI Products PEAK Matrix Assessment for 2026.
The company's financial trajectory adds further credibility. In January 2024, Kore.ai raised $150 million in a round led by FTV Capital with participation from Nvidia, bringing total funding to approximately $223 million. TechCrunch reported at the time that the company's annual recurring revenue exceeded $100 million, with the platform automating 450 million interactions daily. In January 2026, the company secured an additional strategic growth investment led by AllianceBernstein Private Credit Investors, with continued backing from Vistara Growth, Beedie Capital, and Sweetwater Private Equity. The company now claims more than 500 Global 2000 customers and partners, with 75% of its customer base in regulated industries and support for over 300 enterprise integrations.
The Artemis platform is available today at kore.ai, launching initially on Microsoft Azure with broader cloud availability to follow. Koneru said existing customers — many of whom built their current deployments on Kore.ai's previous no-code platform — are planning migrations to the new architecture, while all new customers are starting on Artemis.
The portability question remains partially unresolved. While ABL itself is a YAML-based artifact that customers can store and manage in their own systems, the runtime required to execute it is not yet available as a standalone component. Koneru said a lighter version of the runtime will be made available in the future for customers who want to run ABL outside the full Kore.ai platform, but acknowledged that the initial release prioritizes the integrated enterprise experience.
For CIOs navigating an increasingly crowded and fast-moving market for enterprise AI agents, the Artemis launch poses a clear choice: bet on a hyperscaler's native platform and accept the lock-in that comes with it, or adopt a neutral layer that promises to orchestrate and govern agents across any model, any cloud, and any vendor — but requires trust in a company that, for all its scale and analyst recognition, remains far smaller than the giants it competes against.
"If I'm going to go down the path of one hyperscaler or one SaaS company that provides an agentic platform, I'm getting locked in in some fashion or the other," Koneru said. "We need standardization. We need a central way to build and deploy. We need a central way to govern."
It is a bold claim from a company that has spent 12 years building the plumbing for enterprise AI while flashier names grabbed headlines. But if the next chapter of the AI revolution is defined not by which model is smartest but by which platform can be trusted to run agents safely at scale, then Kore.ai's long apprenticeship in the unglamorous trenches of compliance, governance, and regulated industry deployment may turn out to be exactly the right résumé for the job.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。