Agentic enterprise and AI agents

From: aidotengineer

Presented by Hbert Mishtella of Novardis, this article explores the concept of the agentic enterprise and the profound impact of AI agents on workflows, organizational structures, and individual roles [00:00:15]. It addresses the challenges and opportunities for organizations preparing for a future run by AI [00:00:42].

The Rise of AI Agents

It’s projected that within three years, organizations might be run by AI agents [00:00:53].

What is an AI Agent?

An AI agent is an application based on a Large Language Model (LLM) that exhibits autonomy in its behavior [00:01:13].

Components of AI Agents and Capabilities

AI agents are composed of a language model that orchestrates specific actions [00:01:22]. They execute planning in a step-by-step fashion, utilizing various tools, and can adjust their actions based on previous steps, known as “dynamic action” [00:01:39].

Key capabilities provided by AI agents include:

• Reasoning for multi-step planning [00:02:26].
• Adaptability for dynamic work [00:02:28].
• Control and persistent memory [00:02:32].
• Contextual understanding, often leveraging Retrieval Augmented Generation (RAG) for specific knowledge not within the language model [00:02:35].
• Interactive workspaces (sandboxes, canvas) for collaborative refinement of tasks [00:02:49].
• Computer-using agents that can navigate graphical user interfaces to automate tasks previously performed by humans or Robotic Process Automation (RPA) [00:03:05].

Preparing for AI Agents

Every company should be working on digital assets and evolving them for the usage of AI agents [00:01:58]. Conceptually, every digital asset in an organization should become a tool for an AI agent [00:02:14].

The Workflow Perspective

A workflow is the execution of a process, typically defined by sequential steps [00:03:35].

Automation vs. Augmentation

Traditionally, technology has been used in two ways:

• Automation: Substituting a specific step with technology [00:04:20].
• Augmentation: Making a step quicker or better with technology [00:04:26].

While machine learning could achieve some of this before [00:04:33], AI agents offer a significant difference [00:04:45].

Agentic Workflows

The advent of AI agents enables “agentic workflows” [00:04:47] because:

1. AI agents can tackle a broader set of tasks, operating on cognitive steps often represented in natural language [00:05:07].
2. AI agents can act as “glue” between tasks, composing them together without human intervention [00:05:23]. Human feedback is only needed when the agent requires it [00:05:32].

This allows for the automation and delegation of not just one, but multiple steps in a workflow [00:05:45].

Takeaway 1: To start building agents, first identify your workflows and deeply understand the context around them [00:05:58]. This context (data used, systems, transformations) is often unrecorded and buried in employees’ heads [00:06:17].

The Network Perspective: Personas

Relying solely on traditional roles and job titles is insufficient for applying AI agents [00:07:51]. To understand how to apply an AI agent, one needs to know the exact tools to use, how to plan, and conditions for changing behavior [00:08:02].

Defining Personas

Thinking about each contributor to a workflow as a “persona” provides more useful information than job titles [00:08:42]. Examples of personas:

• Silent Achiever: Lower communication, high performance [00:08:57].
• Individual Contributor: Good communication and performance [00:09:06].
• Connector: Connects non-adjacent teams [00:09:09].
• Multiplier: Enhances the work of other team members [00:09:16].
• Knowledge Hub: Provides domain expertise or necessary information [00:09:21].

Defining personas helps project how AI agents will impact workflows and specific employees [00:09:39]. This global-level view helps identify similarities and differences across teams, which can guide agent optimization [00:10:15].

Takeaway 2: To help your organization and people adapt to AI, consider looking at workflows through the lens of personas [00:10:44].

Impact of AI Agents on Workflows and Personas

With AI agents, workflows can become more efficient:

• A “silent achiever” with a coding assistant can become so productive they can perform work previously requiring multiple people or steps [00:11:18].
• A “multiplier” persona’s impact can be magnified by an AI agent, extending their influence beyond adjacent team members to the entire team [00:12:04].
• A “knowledge hub” persona is most prone to substitution by an AI agent, as a well-prepared agent can dynamically support all team members [00:12:38].

The key is to understand the archetypes or personas of employees across different workflows to plan for the development of different agents [00:13:29].

Emerging Patterns and New Roles

When considering which tasks are better for humans vs. AI agents:

• Humans: Ambiguous workflows where outcomes depend heavily on specific situations, or when subjectivity is desired [00:14:42].
• AI Agents: Repetitive, tedious tasks, or when perfect, consistent execution is required [00:15:12].

Emerging patterns:

• Intelligence and Domain Knowledge as Commodity: These attributes become cheap and easily available, no longer providing a competitive edge for employees or companies [00:15:32].
• Shift in Skillset: Employees and companies need to pivot towards multidisciplinary know-how or much deeper specialization [00:16:09].
• Engineers and Domain Experts: Senior engineers may work directly with domain experts on product development, accelerating work with coding assistants and domain assistant agents [00:16:26].
• Agents as Super Connectors: AI agents can facilitate connections across teams [00:16:46].
• Value of Context Understanding: This will gain more value than raw intelligence or domain expertise [00:16:48].
• AI Supercharges Doers: People who perform tasks can do even more and quicker, potentially leading to a “100 times engineer” [00:17:04].
• Polarization of Positions: Fewer middle-ground positions, leading to new and highly specialized roles [00:17:27].

New roles in AI-mature organizations may include:

• Workflow Minor [00:17:48].
• Human-AI Orchestrator [00:17:55].

Enabling AI Adoption and Democratization

To democratize and accelerate the AI agent revolution, organizations should follow a step-by-step approach [00:18:21]:

1. General Assistants: Provide access to general AI assistants [00:18:32].
2. Individual Assistants: Offer individual assistants per employee, with memory, context understanding, and personal databases [00:18:39]. This allows employees to work more closely with their specific context [00:18:48].
3. Employee-Built Agents: Enable employees to quickly build agents themselves using coding assistants, no-code, and low-code tools [00:19:10]. This requires employees to develop “cognitive self-awareness” to translate their internal thought processes and steps into explicit instructions for agents [00:19:47].
4. Digital Twins: Utilize digital twins (like “time travelers” and “memory”) to retain organizational knowledge [00:20:24]. This prevents knowledge loss when experienced employees leave [00:20:31].
5. Multi-Employee Agents: Develop agents that serve entire teams or workflows, not just individual employees [00:21:07].
6. Swarm of Agents: Progress to managing and deploying swarms of agents [00:21:12].

This progression is now possible due to the maturing software stack around LLMs and AI agents [00:21:22].

Challenges and Opportunities

Challenges

Common challenges include groundedness, hallucinations, guardrails, and security [00:21:45]. Two significant considerations are:

• Adoption: A key question is whether organizations are ready for a paradigm shift where AI drives the interaction and delegates subtasks to humans, rather than the other way around [00:22:20]. This is a major cultural and adoption problem [00:22:41].
• Ethics of Autonomy:
  • Defining the values that AI agents should uphold [00:23:36].
  • Resolving conflicts between human and AI decisions [00:23:55].
  • As stated by John Lennox: “The greater the machine’s freedom, the more it will need a moral standards” [00:24:16]. This necessitates deeper engagement with philosophy and ethics in technology [00:24:25].

Opportunities

AI agents present numerous opportunities, including:

• Simulating synthetic reality before real-life activities in areas like marketing or lab work [00:24:57].
• Proactively identifying regulatory loopholes [00:25:06].
• Scouting for Mergers & Acquisitions (M&A) opportunities [00:25:13].

Executive Summary

Key takeaways for executives:

• Strategy Re-evaluation: If your strategy remains coherent after swapping “AI agent” with “machine learning” or “data,” it is likely outdated [00:26:29].
• New Capabilities: AI agents offer distinct new capabilities [00:26:56].
• Workflow Understanding: Detailed understanding of workflows is crucial for maximizing benefits from agents [00:26:58].
• Democratization: Democratizing agents is essential to accelerate the AI agent revolution [00:27:05].
• Human Aspect: Analyzing the network of personas is vital for successful AI implementation [00:27:13].
• Rapid Transformation: Organizations must be prepared for rapid transformation, as entire organizational charts may change drastically [00:27:19].
• Mindset Shift: A fundamental mindset shift is needed to redefine and adapt to new interactions with technology, where AI may sometimes lead [00:27:31].
• Ethics: Ethics remains a crucial part of developing and using AI [00:27:48].