The rise of the platforms

The concept of the Agentic Organization is built on the promise that AI agents will work alongside humans to automate processes either partially or fully. Studies estimate that, on average, 25% of tasks across all industries will be automated by 2030. Achieving this level of automation, however, requires overcoming several key challenges.

·       Skills Shortage: There is a global shortage of skilled machine learning and AI engineers, and attracting and retaining top talent is an expensive exercise. Developing internal capabilities is a costly endeavor that only large enterprises can typically afford.

·       Speed of Implementation: To fully realize the benefits of Agentic AI, organizations must redesign their processes. This is a time-consuming task, especially when there is a need to develop custom AI agents.

·       Data Readiness: Data readiness is consistently cited as a major barrier to AI projects. Organizations must find ways to quickly prepare and provide data to AI agents, ensuring that access rights are correctly managed.

·       Governance: Even deploying a small number of AI agents requires a robust governance framework. Most IT departments, as well as other business units, are not yet equipped to manage this responsibility.

·       Business Context: Integrating business context into AI solutions is a critical task, often handled by consulting firms. Bridging the gap between a generic technology and specific business applications demands specialized skills that are in short supply (this is on area where we excel).

So, the question remains: Is it possible to rapidly deploy large numbers of AI agents for "straightforward" scenarios, fast and secure?

Agentic AI Platforms: Addressing the Challenges

Agentic AI platforms are emerging to tackle today's challenges and set new standards for AI-driven digital transformation. These platforms provide an environment to develop and manage AI agents with their integrations, data, governance rules, and security roles. Their goal is to deliver speed, ease of use, and business context, enabling organizations to deploy AI agents on a scale.

The first mover was, Salesforce, announcing Agentforce in the summer of 2024 (prior to this, "Agentic AI" was a term used mainly among developers). Agentforce leverages customer data from its data cloud and understands the Salesforce ecosystem, allowing rapid, secure deployment of AI agents for various scenarios.

ServiceNow followed, launching its Agentic platform with an emphasis on business workflows, providing cross-application coordination and integration for AI agents.

Workday then introduced its own platform, by adding the angle of HR and the importance of “human in the loop” in the agentic processes.

However, the most promising platforms come from industry leaders such as Microsoft (Co-Pilot Studio), IBM (Watson), and Google (Gemini Enterprise). Reason being, these vendors control key component data, large language models (LLMs), security, applications, development environments, and integrations—making their platforms the easiest and fastest way for organizations to leverage the agentic AI technology.

Advantages of Agentic AI Platforms

·       Time to Market: For organizations using a vendor's technology stack, deploying an agentic platform is a routine IT task. Developers accustomed to these environments can work naturally, while access to multiple LLMs streamlines testing and deployment.

·       Ecosystem: These platforms offer AI agent marketplaces and pre-built templates, making deployment for simple scenarios possible within weeks. Over time, this will foster an ecosystem where all AI agents comply with security and governance standards.

·       Ease of Use: Designed for configuration by AI platform specialists, not necessarily data scientists, these platforms can address business needs as long as the context is understood. Development will still be needed but this will also be done using natural language and some Python😊.

·       Governance: Centralized management through a single admin panel ensures that all AI agents adhere to unified governance rules, simplifying oversight compared to managing agents individually across platforms.

·       Integration: Support for protocols like A2A, API, and MCP enables communication between AI agents and third-party applications, allowing companies to consolidate agents into multi-agent systems within one platform.

Limitations of Agentic AI Platforms

·       Limited LLM Options: While platforms offer a selection of LLMs, none provide every option. For example, developing on Google's Gemini model while using Microsoft's Co-Pilot platform its not possible and requires work outside the platform and integration A2A.

·       Development Flexibility: Although these platforms are user-friendly, complex scenarios—especially those involving legacy applications or data sources—may require custom coding rather than relying solely on low-code environments.

·       Business Context: Platforms may not fully understand third-party applications or external data sources, necessitating outside AI agents to be called from within the platform.

·       Cost: These platforms typically charge monthly subscription fees based on user and consumption metrics. While they are suitable for quickly deploying a few agents, long-term costs may outweigh the benefits as usage grows.

·       Control and Stability: Rapid technological evolution results in frequent platform updates, which can disrupt existing developments and increase support costs due to extended testing requirements.

Final Notes

As with all emerging technologies there will be winners and losers. In our view the platforms that will prevail will have the following characteristics:

·       Proximity to developers and a large user base, driving the proliferation of AI agents.

·       Volume and variety of AI agents, with simple development, deployment, and management processes fostering a rich ecosystem.

·       Strong data context, including robust connections to mainstream systems and support for zero-copy data access.

·       Competitive total cost of ownership, factoring in licensing and operational costs.

Ultimately, the platforms that will prevail will RISE to become the backbone of Agentic Organizations.