What are watsonx agents?

watsonx agents can perform many business functions without human input, such as automating workflows, integrating with tools and communicating with customers in real time.

Businesses stand to potentially reap several key benefits from using IBM watsonx Orchestrate for their agentic AI operations. These advantages include:

• Quick deployment: Integrate prebuilt agents complete with domain expertise and enterprise-grade logic ready for immediate use.

• Centralized management: Streamline agentic AI operations management with a single interface.

• Workflow integration: Turn preexisting workflows into agentic workflows with support for numerous apps, tools and systems.

• Enhanced efficiency and scalability: Implement agentic AI solutions to automate tasks without the need for step-by-step prompting.

• Enterprise-grade security: Set guidelines and guardrails to control how watsonx agents behave.

IBM watsonx offers prebuilt, customizable enterprise-grade agents for use cases spanning numerous business operations. The no-code Agent Builder and included software development kit (SDK) suite both allow users to create custom agents and AI applications.

watsonx Orchestrate’s prebuilt agents cover four domains:

• Procurement agents

• Sales agents

• HR agents

• Customer care agents

Each agent comes with the domain expertise, enterprise logic and app integrations it needs for its respective role.

The AI models available as procurement agents automate the procurement lifecycle so employees can focus on strategic impact that optimizes business gains. Procurement agents can help with:

• Supplier management: By integrating insights from Dun & Bradstreet, supplier management agents can proactively flag potential compliance and risk management issues. Automated supplier onboarding enhances data accuracy and profile maintenance.

• Purchasing: Agents shorten procurement cycles and streamline helpdesk operations by empowering employees to check purchase requests, orders and goods receipts. Employee self-service benefits also cover purchase request creation and modification as well as operational issue resolution.

• Sourcing and contracts: Generative AI (gen AI) agents can help with event creation, request for proposal (RFP) drafting, supplier recommendation and contract workflow management. watsonx agents generate content that aligns with organizational templates and policies.

• Invoice management: Agents help improve invoice accuracy by checking for errors and discrepancies, clearing blocked invoices and streamlining collaboration between stakeholders for invoice processing.

The sales agents in the watsonx Orchestrate Agent Catalog are designed to help sales teams automate routine tasks and generate insights they can use to maximize growth. Sales agents help with:

• Client outreach: Gen AI agents draft and send personalized emails to verified clients for scalable, faster outreach.[CS1] 

• Client search: The Dun & Bradstreet integration provides enriched company and contact data for targeted prospecting.

• Client relationship management (CRM): Agents connect with Salesforce to view and maintain accounts, opportunities, leads and contact records, helping ensure sales teams always have current data.

• Product enablement: Curated content from the Seismic sales enablement platform helps agents answer seller questions with the right messaging, USPs (unique selling points) and sales collateral.

• Sales prospecting and conversion: Agentic AI moves early-stage leads through the funnel by answering questions and helping sales teams engage.

• Sales research: Agents compile data from multiple sources to equip sales personnel with the data they need to pitch effectively and close leads.

watsonx Orchestrate’s prebuilt HR agents automate complex tasks to deliver employees and HR personnel the services and functionality they need. HR agents help with:

• Employee support: HR agents handle mundane tasks including time-off requests, payslip retrieval, benefits education and onboarding, freeing HR personnel for more demanding jobs.

• Talent acquisition: Recruitment is an AI arms race on both sides of the equation. Agents help automate job posting, candidate sourcing, resume screening, interview scheduling and offer drafting.

• Learning and development: Agents further upskilling and advancement by identifying skill gaps, issuing personalized training recommendations and automating learning workflows.

• Performance management: Managers can use agents to help set personalized goals for individual employees, provide continuous feedback and streamline performance reviews.

The customer care agents in watsonx Orchestrate help customers resolve queries without having to involve support personnel—leveraging first-party business data and escalating to support staff only when needed. Customer care agents can provide:

• Omnichannel customer self-service: Agents can help assist customers across the range of support touchpoints, including web chat, phone, SMS and social apps. The foundation models powering these agents use natural language processing (NLP) for contextual awareness and reasoning to resolve multi-step tasks on their own.

• Call center modernization: Voice AI agents use natural language understanding (NLU) to power speech recognition and chat with customers on the phone.

• Agent assist: AI agents support human staff with relevant answers and suggested next actions to simplify decision-making and increase efficiency.

In addition to its suite of prebuilt agents, watsonx Orchestrate offers two solutions to organizations needing custom agents. The no-code Agent Builder provides an intuitive agent framework for AI development. Users with development experience can use the Agent Development Kit (ADK) to design and build agents with Python, YAML and JSON.

The Agent Builder is a no-code interface for creating, testing and deploying AI agents. Custom-built agents can integrate with the systems and tools organizations already use through API keys. Users can combine third-party agents, such as those created with the open source frameworks LangChain and LangGraph, with Orchestrate agents for unified workflows.

With support for direct coding, the ADK gives developers the space and tools they need for agent creation and deployment. Its command line interface (CLI) utilities and Python modules can be used to build agents and tools for watsonX Orchestrate AWS, IBM Cloud and other implementations.

watsonx Orchestrate simplifies the process of designing multi-agent workflows. Agent orchestration connects agents so they can work together, sharing context and managing dependencies. Each agent works modularly to reason, integrate with other systems or delegate tasks to other agents.

Orchestration in watsonx is based around the following features:

• Nested agent calls: High-level agents break down tasks into subtasks and delegate them to other agents.

• Sequencing and control flow: Multi-step agents can execute tasks in order, using one agent’s output as input for the next or progressing along adaptive paths.

• Context propagation: Orchestrated agents share workspace context for smooth collaboration.

• Reusable agent catalog: watsonx’s prebuilt agents can fit into workflows or serve as templates for further customization.