Real-world deployments of agentic AI have so far been limited in scope, despite strong interest in using the technology to automate many of the business processes now handled by people.

One reason for the slow deployment of agents is the challenge of multi-agent orchestration: the ability of AI agents to communicate with each other and coordinate their activities across enterprise applications, workflows and even corporate firewalls.

There is growing recognition that developing a framework for multi-agent orchestration is essential for deploying agents on a large scale across the entire organization.

In this episode, we explore the main elements of a multi-agent framework, the problems it is meant to address, who is responsible for developing it and where to find ready-made frameworks and tools.

Featuring: Peter Hesse, Partner, 10Pearls

In today's episode, we'll also cover:

• Why the tendency of agents to work in harmony can make them less resilient when their scale expands.
• How a framework can support AI transparency and traceability.
• Using "policy as code" to enforce AI consistency and trust.

References:

• AI agent frameworks: A guide to evaluating agentic platforms
• Real-world agentic AI examples and use cases
• 10Pearls blog post on building enterprise AI agent frameworks

To learn more about enterprise applications, check out Search ERP.

Direct materials sourcing is a critical process in product design, engineering and manufacturing. That was never more apparent in 2017 when Tesla began to ship the Model 3, its first electric vehicle meant to be affordable for middle-income buyers.

The company had set two ambitious and unprecedented goals: building an EV for a base price of $35,000 and completing the development cycle in three years, half the typical turnaround in the automotive industry. The compressed timeline put tremendous pressure on employees to source quality parts economically and on time while keeping up with the tight design and production schedules.

But Tesla was working at a disadvantage. Much of its sourcing and procurement was still done manually, and an IT gap existed between the product lifecycle management system, where engineering and design were managed, and the ERP.

In this episode, we explore the impact inefficient procurement can have on profit margins, how a direct materials sourcing platform can close the technology gap for manufacturers, and the role played by AI.

Featuring: Spencer Penn, Co-founder and CEO, LightSource

In today's episode, we'll also cover:

• Why mastering the bill of materials is so important in sourcing's financial impact.
• How LightSource works, who uses it and where it fits in the product lifecycle.
• Lessons learned from Penn's experiences managing engineering finance at Tesla.

References:

• LightSource on procurement's overreliance on email and spreadsheets
• Tesla misses Q3 goals due to "production bottlenecks"
• Automotive supply chains can benefit from sourcing alliances: Here's why.

To learn more about enterprise applications, check out Search ERP.

Agentic AI is the hottest trend in ERP. It promises to infuse enterprise applications with AI that anticipates users' needs, communicates with them in natural language and handles more of the tedium of working in ERP systems.

SAP is one of the leaders of the push for agentic AI. It has aggressively added special-purpose AI agents and development tools for building custom agents and beefed up its data platforms to accommodate AI's needs.

But in practice, building agentic AI that works across SAP and non-SAP systems is challenging, and it often requires outside help from a consulting firm or system integrator.

In this episode, we explore the practical realities of implementing AI applications on SAP systems, including the design process, development tools and integration challenges.

Featuring: Gianluca Simeone, Vice President, CTIO and GenAI Leader, Capgemini

In today's episode, we'll also cover:

• Keys to successful agentic AI projects.
• Integration tools and standards for SAP AI.
• Whether non-programmers can use low-code development tools to play a meaningful role in agentic AI design.
• The status of multi-agent orchestration protocols.

References:

• SAP pitches role-based Joule assistants as ERP work partners
• Agentic AI explained: Key concepts and enterprise use cases
• Simeone explains procure-to-pay project in SAP video

To learn more about enterprise applications, check out Search ERP.