The Rise of Agentic AI: Workato and Other Leaders Pave the Way for Enterprise-Wide Adoption

Workato doubles down on agents - but will its customers be able to keep pace?

At its recent user conference earlier this month in Las Vegas, enterprise workflow and integration vendor Workato doubled down on its vision of the agentic enterprise -- unveiling a range of new capabilities covering topics from platform scaling and ecosystem enablement to agent distribution and governance. These announcements built on the previous releases of Workato ONE and Workato GO earlier in the year -- and seem intended to round out these building blocks as part of a push to mature its platform for AI agent-based use cases. As we previously covered, Workato ONE combined the vendor's existing workflow infrastructure and enterprise search functionality (from its DeepConverse acquisition) with a new layer of tooling for building agents. This was quickly followed a couple of months later by the announcement of Workato GO, a unified interface which aimed to combat agentic sprawl by giving users a single place to interact with a broad range of data sources and agents. Together these releases set up the spine of a platform stretching from operations to user experience. Pushing further into this space during its annual customer event, Workato made a slew of announcements that represent a deep bet that its customers -- both old and new -- will increasingly want to build and run agentic systems. From extending the core platform for scale and pushing to create a deeper developer ecosystem, to announcing mechanisms for distributing and governing agents, Workato aimed to convince its customers that it was now ready to help them deliver agentic systems at scale -- moving beyond isolated experiments to enterprise-wide deployment. The foundation of Workato's vision remains Workato ONE, with the company aiming to use its platform to overcome the patchwork of current agent experiments it sees implemented by its customers. By providing a single integrated platform, the company believes it can satisfy demand for agentic systems while building on its existing guardrails and governance. In pursuit of this vision, the company announced a range of new agentic and orchestration capabilities as part of an extended Workato ONE platform -- from visual modelling of agent behaviours and contextual data retrieval via knowledge graphs, to data orchestration and the ability to reuse existing enterprise skills. Workato claims that these capabilities close the gap between traditional workflow and new agentic use cases, consolidating them into a single platform -- as Chief Product Officer, Amlan Debnath, explains: Workato ONE now gives businesses a single system to orchestrate AI, data, and action -- securely, at scale, and in production. For CIOs grappling with chaotic pilots this feels like it could be a strong attractor -- consolidating experiments into a single platform that already has all of the necessary actions and governance baked into the underlying infrastructure -- a point made by one of Workato's customers, Kim Huffman, CIO of Workiva: CIOs are facing a flood of AI use cases and point tools across every part of the business... Workato ONE gives us a single platform to surface, manage, and scale the agents. Workato's approach to agent development relies on their managed creation and deployment by skilled IT teams -- making developers a key constituency for the company. To scale its business, therefore, it needs to bring more developers into the ecosystem, growing the pool of people who choose to build with the platform. As part of this push Workato introduced a free Developer Sandbox -- a self-service environment that gives developers access to the full suite of Workato ONE capabilities (within limits) making it easy for developers to build and launch AI agents without waiting for traditional enterprise onboarding. In practice, enterprises might use the Sandbox as a proving ground for innovation -- while startups and side projects could use it to seed new agent use cases that ultimately flow back into the Workato ecosystem. As Adam Seligman, Chief Technology Officer, explained: Everything we've built over the past decade ... is now available to developers for free. The full platform. All the connectors. Agent Studio. Our remote MCP server. It's all here. By lowering the barrier to entry, Workato clearly hopes that the Developer Sandbox will broaden experimentation -- not only within its own customer base but with a broader ecosystem of curious developers. If successful it could encourage more usage of the platform -- whether by startups, side projects, or citizen developers inside large enterprises -- and lead to a stronger ecosystem of skills around the platform. The third major announcement from the company was the introduction of a fleet of production-grade 'Genies'. These prebuilt agents are designed for common business functions such as sales, support, HR, IT, marketing, and security. Each Genie is a packaged agent which comes embedded with KPIs, orchestration logic, and governance controls. Workato itself claims to have replaced more than 100 internal workflows with Genies, which it describes as operating not as copilots but as coworkers. Vijay Tella, co-founder and CEO, set out the company's ambition: The enterprise has been stuck in pilot mode, tinkering with copilots and waiting for AI to mature. Workato Genies act as true coworkers, operating at scale across the business and driving measurable impact today. While from my perspective the existing out-of-the-box assistants are a curiosity -- in much the same way as 'standard process templates' in a traditional workflow platform -- the more fundamental value of Genies, in my opinion, is their role as a packaging and distribution infrastructure for an organisation's custom agents -- representing Workato's attempt to make agents not only easy to build but also easy to distribute and consume. Agents -- like all complicated technologies -- end up with a long tail of data, configuration files, and definitions that often end up causing more trouble than the initial development. If Workato has cracked a way of making that packaging and distribution easy then it could be a real value for CIOs. Finally, Workato announced a new capability within Workato GO called Action Board -- a single interface for providing real-time oversight of agents in production. Workato claims that by using Action Board employees can monitor their Genies -- watching KPIs, viewing open action items, drilling into processes, and even conversing with agents to trigger follow-ups. Each Genie is presented as a widget with a defined role and job description, with role-based access controls ensuring appropriate visibility. By combining monitoring with conversational control, Workato says that Action Board aims to give enterprises confidence that agents are working safely and effectively at runtime -- while also giving them the tools to intervene when necessary. Bhaskar Roy, Chief of AI Products & Solutions, describes the intent: With Action Board, we're giving enterprise teams the visibility and control to treat AI agents like employees and not experiments. While we at diginomica aren't convinced by the concept of treating agents as employees, we certainly see the potential for unmanaged agents to cause chaos -- and so Workato's focus on providing clear observability and governance over their Genies is certainly a welcome move. For enterprises, Action Board could evolve into the equivalent of a management console for agents -- a place where managers track KPIs, CIOs monitor compliance, and employees gain confidence by seeing what their agents are doing in real time With these announcements, Workato is continuing its bold push into the agentic landscape -- despite some recent industry-wide challenges in terms of enterprise adoption and ROI. In this sense its investments represent a bet that current issues are just teething troubles and that by providing better tooling they can help their customers leverage agents for solid benefit. If it's correct then its continued rounding out of its platform with these four most recent pillars -- scale, ecosystem, distribution, and governance -- absolutely makes sense. Each maps to a practical need that enterprises will undoubtedly face as they begin moving agents from pilot to production. But it remains a bold push. Many AI deployments are still struggling to demonstrate value while research shows today's models still falter when faced with novelty or complexity. On the other hand, my own taxonomy of agents highlights that most real-world deployments remain in the safe zones of Instruction and Orchestration -- a position in which Workato excels by building on their existing infrastructure, to be fair. In that sense, Workato's emphasis on packaging, governance, and ecosystem is a pragmatic step in the right direction. Its platform might not answer the deeper questions of philosophical alignment that enterprises must face first -- but it does provide an increasingly capable infrastructure that makes such alignment safer to pursue. So if Workato can sustain momentum while others remain stuck in pilot mode, therefore, it may yet establish itself as an operating core of the agentic enterprise.

Early leaders in the agentic AI race show agents are here to stay - SiliconANGLE

Early leaders in the agentic AI race show agents are here to stay Generative artificial intelligence is oh-so-2024. Now that we're most of the way through 2025, all the AI buzz is around agentic AI. AI agents - large language model-powered autonomous software programs that can iteratively learn from multiple data sources to achieve the goals set out for them - are now attracting all the attention. The obvious question: How much of this noise is hype? Are there agentic platforms that actually work, leveraging the transformative nature of AI agents to solve business problems in fundamentally new ways? To answer this question, I interviewed several vendors that have risen to the top of the froth, developing early reputations as leaders in the agentic AI space. (Disclosures below.) Here's what I learned: Given generative AI's success powering chatbots with human language interfaces, agentic platforms that power customer support are an obvious place to start. Sendbird Inc. provides an agentic AI platform that focuses entirely on customer support. It works across multiple communications channels, including Instagram and Facebook Messenger. Sendbird can initiate autonomous support and sales conversations while keeping humans in the loop for more complex inquiries. It then proactively re-engages with end customers, enabling Sendbird's customers to offer automated 'personal concierge' services, even to mass market customers. It offers its own unit testing framework as well as security, privacy and compliance capabilities that provide the guardrails necessary for customer interactions with AI-based services. The agentic AI platform from MavenAGI Inc. also focuses on improving the customer experience. It can resolve support tickets, answering customer enquiries more than 90% of the time. MavenAGI has extensive multimodal support, working across copilots, chatbots, web search, voice, email, WhatsApp, SMS, Slack and Teams - as well as via application programming interfaces. MavenAGI's agents are smart enough to resolve most customer problems across different industries, for example, deliveries of the wrong item, fraud alerts or canceled flights. The company's speech intelligence is an important differentiator, as MavenAGI offers both speech-to-text and the more challenging voice-to-voice capability. These voice capabilities support its extensive telephony and call center integration. Customer support is but one of many areas where agentic AI has established a foothold. Perhaps the most interesting conclusion from my research is how AI agents can address challenges across multiple departments in the enterprise. Auditoria.ai Inc provides an agentic AI platform for finance operations, or FinOps. As a result, the company targets chief financial officers - notoriously skeptical purchasers of software. The intelligent nature of Auditoria agents, however, has converted many reluctant CFOs. The agents take over many of the mundane FinOps tasks such as managing communications, digitizing documents and handling status notifications. What convinces most CFOs, however, is Auditoria's AI-driven automation, which can handle invoice, collection, remittance and help desk processes, freeing personnel to focus on more strategic tasks. 7AI Inc. offers an agentic AI platform for cybersecurity analysts. The company's agents leverage alerts and other incoming data from multiple sources to perform security investigations, make conclusions and offload repetitive tasks. The company offers what it calls 'swarming agents.' Based on technology that offers a dynamic approach to generating agents on the fly, the platform may provide hundreds or thousands of agents for any particular customer. 7AI agents may behave either deterministically or nondeterministically. For example, they handle triage of alerts deterministically but takes a nondeterministic approach to creating mitigations based on researching various data sources, including tribal knowledge that human personnel have documented previously. Rox Data Corp. focuses its agentic AI platform on revenue operations - specifically sales productivity. Rox is looking to reinvent the customer relationship management market (as well as integrate with Salesforce and other CRM apps) with its 'agent swarm' approach. Underlying Rox's agent swarms are a data lake-based system of record that supports a knowledge graph that connects data from multiple sources. Rox's agents handle numerous tasks for go-to-market personnel, including research, tracking, and an understanding of org charts and opportunities. Finally, Resolve AI Inc. focuses its agentic AI platform on information technology operations, essentially providing an agentic AI-based system reliability engineer or SRE. The platform handles all alerts, performs root cause analysis of issues, and then troubleshoots incidents, acting as an on-call resource for human operators. Resolve leverages a knowledge graph that maintains information about all infrastructure dependencies in real-time. This knowledge graph understands the full enterprise production environment. Agents continually listen to alert streams and gather evidence from multiple sources to generate hypotheses about the root causes of issues. It then iteratively tests all such hypotheses to uncover the most likely causes. Though Resolve primarily focuses on the production environment, developers can also use its agents to troubleshoot incidents during development. While some agentic platforms focus on the needs of particular departments, there are also some offerings that seek to address the needs of employees across the organization. NinjaTech AI Inc. offers SuperNinja, a general-purpose AI agent that acts as a knowledge worker, vibe coder, analyst and researcher, among other roles. SuperNinja installs a Linux VM which hosts the agents, either in sandbox mode or for cloud-based deployment. Customers can then port the VM to their environment of choice. Its breath of capability is its most remarkable characteristic. It can build web sites, generate PowerPoint slides, analyze spreadsheets, automate processes, and more - all within a single agent. In addition to SuperNinja, NinjaTech also offers several simple agents that can handle individual tasks. In fact, NinjaTech's agents fall into three categories -- simple, complex and autonomous -- thus matching the level of complexity for each task. Glean Technologies Inc. provides an agentic AI platform that empowers any employee to build and run their own agents. Glean began its journey to agentic AI by offering AI-based search. It soon added LLM-based AI assistants, aka chatbots. Now the company offers a full-fledged agentic AI platform that leverages Glean's deep expertise with both search and AI assistants. Glean offers a no-code interface that anybody can use to build agents. In addition, the company provides low-code capabilities for developers to build agents rapidly, as well as "pro-code" capabilities via APIs and an SDK when low-code falls short. Under the covers, Glean's most important differentiation is how its knowledge graph keeps track of granular permissions for every data element it tracks. This thorough permissions infrastructure enables employees to build a wide variety of powerful agents that nevertheless conform to privacy, security and compliance policies. The sorts of agents that employees might build, however, are likely to be lightweight, point solutions that provide little value across the organization. In contrast, Moveworks Inc. provides an agentic AI platform that delivers agentic AI assistants for the entire enterprise workforce, including IT, HR, finance, engineering, sales, marketing and everyone else. Moveworks agents can automate even the most complex, enterprise-wide business processes. In addition to the aggregation and action-based agents that all platforms offer, Moveworks also offers "ambient" proactive agents that run continuously, monitoring for information relevant to their goals. Its most important differentiator, however, is its reasoning engine which manages all the agents, iteratively learning, experimenting and building agentic workflows. Moveworks' capabilities are so impressive, in fact, that ServiceNow Inc. announced earlier this year that it's acquiring the company. While developers can use many of the platforms above to create and deploy AI agents, a few standout vendors focus entirely on developer requirements. AgentOps.ai from Staf Inc. aka Agency offers a developer platform for building, debugging and deploying AI agents and other LLM-based applications. The platform also provides developers with observability into agent behavior. AgentOps enables developers to restart agents from the point of failure, thus accelerating the debug/fix process. Rounding out this category is Mastra from Kepler Software Inc., which offers a TypeScript agent framework that leverages its open-source Gatsby JavaScript stack. Mastra offers everything developers expect from an agent development framework, including a unified provider API, RAG support, flexible memory, prompt tuning and tool calling - all with TypeScript, a popular, strongly typed extension of JavaScript. I also uncovered some notable examples of vendors I have spoken with several times over the years, and which have now pivoted to agentic AI. Akka from Lightbend Inc. extends the mature Akka serverless, message-driven reactive runtime to the agentic AI world. Today, Akka is a platform for building, running and evaluating agentic systems, including orchestration, memory and streaming capabilities. Akka agents provide developers with both a design model and runtime for agentic systems, helping them define how agents gather context, reason and act. SnapLogic Inc. leverages its long history as a low-code enterprise integration platform to offer an agentic integration platform for developers to integrate AI, data, apps and services. It now offers its AgentCreator low-code tool for creating and deploying AI agents. AgentCreator supports multiple LLMs and provides agent evaluation, observability and data security capabilities. This category wouldn't be complete without mentioning UiPath Inc., which has a well-deserved reputation as a leader in the robotic process automation market. Its bots handle legacy integration and automation for numerous enterprises, despite bots' primary shortcomings of brittleness and adding to technical debt. (See my 2021 coverage of UiPath's RPA offering.) As a result, the company has pivoted to agentic AI, leveraging LLMs to support the autonomous behavior essential to flexible automations. UiPath's most important differentiator, however, remains its RPA bots, which it now orchestrates into agentic workflows when access to legacy applications and data sources is important for building agentic workflows. What the 14 vendors in this article have in common is that they all have fully functional agentic offerings - with no handwaving or "agent-washing." Furthermore, this list only scratches the surface. My research turned up several other vendors I might have just as easily included. The conclusion, therefore, is that agentic AI is real and here to stay. True, it's still an emerging technology and there's plenty of hype to go around. But as these vendors show, agentic platforms and the agents they support are solving business problems today - and furthermore, are transforming the way software addresses complex business needs.