Orchestrating Agentic Agreement Workflows with Docusign Iris

Key takeaways

• Docusign gives builders several ways to use Iris-powered agreement intelligence: in-platform agents and workflows, portfolio analysis with Worksheets, and external AI workflows through the Docusign MCP Server. 

• These capabilities help teams reduce manual agreement work while keeping people in control of consequential decisions. 

• Docusign Iris is the AI engine behind the agentic capabilities, turning agreement context into insights and action.

Agreement workflows rarely happen in one system. A renewal might need customer data from a CRM, pricing from a CPQ platform, legal approval, the right agreement template, and follow-up after the document is sent.

Each step may be straightforward on its own. The hard part is coordinating them reliably, with the right context, permissions, and human review. Builders can bring those disconnected steps together by designing agents and workflows that orchestrate actions across Docusign and other business systems.

Docusign Iris is the AI engine behind Docusign’s agreement intelligence and AI experiences. It uses agreement context, such as terms, obligations, amendments, negotiation history, and renewal provisions, to help teams move from understanding agreements to acting on them.

In this guide, “builders” includes admins, process builders, architects, developers, SIs, and technical teams across no-code to pro-code experiences. As agentic capabilities expand who can build, subject matter experts can also configure and shape workflows without a traditional development background.

Builders can leverage ready-to-deploy agents, create custom agents in Agent Studio, add agents to structured workflows in Workflow Builder, analyze agreement portfolios with Worksheets, or bring Docusign capabilities into third-party AI applications through the Docusign MCP Server.

Across these approaches, the most effective implementations focus on a specific outcome, provide the agreement context the workflow needs, define clear boundaries for automation, and keep people in control of consequential decisions.

Watch this overview to understand how Docusign’s agentic capabilities can support agreement workflows:

Choose the right approach for your workflow 

Docusign provides several ways to use Iris-powered agreement intelligence, depending on where users work and how much reasoning, orchestration, and customization a workflow requires.

Ready-to-deploy agents

Ready-to-deploy agents handle repeatable, multi-step agreement tasks without requiring a team to design an agent from scratch. 

These agents are intended for repeatable tasks such as retrieving agreement information, monitoring status and obligations, applying legal playbooks, and guiding users through standard agreement processes. Users can invoke agents through the AI assistant or use them as steps in Workflow Builder.

For developers and technical teams, these agents can become reusable capabilities within a broader solution. Developers can provide workflow inputs, connect the result to a user experience or downstream process, and use Workflow Builder, the Workflow Builder API, or the Docusign MCP Server to orchestrate the surrounding experience.

Where to start: Explore Docusign Agents and the Docusign Developer Center for APIs, workflows, and implementation resources.

Agent Studio

Agent Studio lets teams build custom agents for organization-specific processes. Subject matter experts in legal, sales, procurement, operations, and other business teams can describe the desired behavior in natural language and provide the policies, playbooks, and business context the agent should follow.

Custom agents are useful when a workflow depends on organization-specific knowledge, such as approval requirements, pricing policies, escalation rules, or standard operating procedures. 

After testing and refining the agent, teams can make it available to other users or invoke it from a broader workflow.

Example: Review past agreements before a renewal

Before a renewal, a legal operations team may need to review prior agreements, amendments, nonstandard terms, total contract value, and upcoming renewal dates. 

With Docusign Agent Studio, the team could create a custom agreement history agent with instructions such as:

Find the previous agreements for this customer. Summarize the key commercial and legal terms, identify deviations from our standard terms, calculate the total contract value, and list upcoming renewal dates.

The agent could then:

• Retrieve agreements associated with the customer

• Use relevant customer or account context from connected third-party systems like Salesforce and HubSpot, where configured through Extension Apps

• Extract the relevant terms and dates

• Compare the language against the organization’s legal playbook

• Flag exceptions that require review

• Return a structured renewal summary

The legal team still defines the policies and review requirements. The agent handles the repetitive work of collecting and organizing the information so reviewers can focus on judgment and negotiation.

Once tested, the same agent can be scaled across the organization to other legal or sales teams or invoked from a broader renewal workflow in Docusign Workflow Builder. 

Where to start: Explore Docusign Agents for custom-agent capabilities and the Docusign Developer Center for broader implementation resources.

Use agents in Workflow Builder

Custom agents become more powerful when they are part of a workflow. Builders can use Docusign Workflow Builder to standardize processes at scale by creating workflows that combine agent reasoning with rule-based routing, approvals, and downstream actions.

Developers and partners can then connect those workflows to external applications by using APIs to trigger or manage workflow activity. This lets teams design the workflow in Docusign while still connecting it to customer, partner, or system-of-record experiences.

For example, an agent can evaluate an agreement against a playbook and return a structured result. Workflow Builder can then use that result to route an agreement, request approval, escalate an exception, or trigger another downstream action.

When configuring a custom agent for a workflow, define a structured output schema so the agent returns machine-readable results. In Workflow Builder, add the custom AI agent step and use the output in conditional logic.

Where to start: Explore Workflow Builder and the Workflow Builder API.

How end users access agentic workflows with AI assistant

After builders publish a custom agent, the AI assistant serves as an entry point where end users can ask for information about their agreements and call the agent to take action.

For example, a user might ask the assistant to summarize a contract, identify an upcoming renewal, or explain how a clause differs from standard language. Because the assistant works with agreement context rather than standalone documents, it can provide more relevant answers than a general-purpose model working from isolated files.

In other words, builders define the workflow, context, controls, and human-review points. The AI assistant helps make those agentic capabilities accessible to end users in natural language.

Where to start: Review Docusign Iris and Docusign Iris AI agents. 

Worksheets

Powered by Iris, Worksheets in Agreement Manager help teams extract and compare specific data across thousands of agreements at once.

Teams can use Worksheets to identify vendor agreements that allow price increases above a certain threshold, find customer contracts with specific renewal or termination clauses, and surface obligations or caps that deviate from policy. 

Results are grounded in live signed agreements and can include citations, reasoning, and confidence indicators. This helps teams verify the source of each extracted value during a high-volume review.

For builders, Worksheets is the extraction layer that makes agreement data programmable. Rather than parsing PDFs or relying on manual tagging, developers can define what to extract (e.g., dates, commercial terms, risk signals) and Worksheets returns structured, queryable metadata across thousands of agreements at once. That structured output is what turns agreements from documents into data. It's the foundation for feeding agreement intelligence into agents, workflows, and internal tools in a way that replaces brittle parsing logic with a reliable, maintained extraction layer.

Where to start: Explore Agreement Manager and the Agreement Manager API.

Docusign MCP Server

Agreement workflows rarely start and end in one system. The Docusign MCP Server helps builders bring Docusign agreement actions into supported third-party AI applications, including Claude, ChatGPT, Gemini Enterprise, Copilot Studio, and Slack.

This lets developers and business users connect agreement workflows to the tools where teams already work. Depending on the supported environment and configuration, users can create, review, send, track, sign, search, analyze, and manage agreements with natural-language prompts.

Scenario: Send a renewal form from a third-party AI tool

Let’s take an example in sales. Suppose an account executive needs to prepare a customer renewal agreement. The process starts when the AE opens a renewal request, which notifies in-house legal counsel. 

A custom agent built by Legal summarizes agreement history, including total ACV, open agreements, and renewal dates, and attaches that context to the agreement desk ticket.

Working in a supported third-party AI interface, such as Claude, the AE can ask a custom Renewal Readiness Agent, purpose-built for this use case by a developer in the AE's organization, to prepare the renewal by:

• Retrieving company and customer note details from the CRM

• Pulling pricing and quote details from the CPQ system

• Confirming that the legal intake request has been closed

• Validating that the proposed discount is within the allowed range

• Presenting the renewal details for review

After the AE confirms the details, the agent can use the Docusign MCP Server to prepare and send the agreement for signature. The AE can later check status or send a reminder from the same AI interface.

This pattern connects third-party AI tools, business systems, and Docusign agreement actions while keeping consequential steps, such as sending an agreement, under user control.

Where to start: Start with the MCP Server developer documentation, then choose a setup path for Claude, ChatGPT, or GitHub Copilot.

Build with security and control in mind

As agentic workflows connect Docusign with external AI tools and business systems, security, permissions, and human control become essential to implementing these capabilities. Agents are limited to the agreement data the user can access. Permissions configured in Docusign Agreement Manager and other Docusign IAM services determine which agreements and agreement sets that user can retrieve.

When designing an agentic workflow, builders should also decide:

• What data the agent can read or write and which actions it can perform

• Which actions require human confirmation

• What data can be sent to external systems or models

• How instructions and context keep the agent grounded in its assigned task

• How the workflow handles missing data, exceptions, and errors, including when default values are appropriate

• What happens when an approval or policy check fails

• How the workflow verifies that the intended outcome was achieved

• Which actions and decisions need to be logged

These controls are particularly important for workflows involving confidential agreements, employment documents, regulated data, or financial commitments.

Availability

Availability varies by capability, plan, region, language, and launch phase. 

• AI assistant, ready-to-deploy agents, and custom agents in Agent Studio are generally available as of July 2026 and Worksheets in Agreement Manager as of August 2026 for customers hosted in North America data centers with IAM Professional, IAM Enterprise, IAM for Sales, or IAM Platform plans (English only). Using agents as steps in Workflow Builder is supported on IAM Professional plans and above, initially for accounts in North America data centers in English. 

• The Docusign MCP Server and connectors are available in global open beta in English, with support for AI environments currently including Anthropic (Claude), GitHub (Copilot), OpenAI (ChatGPT), Google Gemini Enterprise, opens in a new tabMicrosoft Copilot Studio, and Slack (Slackbot). General availability is targeted for a future release.

Because availability and supported capabilities can vary by region, plan, and launch phase, always check the latest product documentation and program requirements before beginning an implementation.

Additional Resources

• Build with Docusign and AI to explore the Docusign MCP Server, Agent Studio, Agreement Manager API, and other AI-focused developer capabilities

• Get started with the Docusign MCP Server to connect an AI assistant to Docusign agreement data and actions

• Explore the Agreement Manager API to retrieve AI-extracted agreement data for your applications and workflows

• Explore the Workflow Builder API to create, trigger, and manage agreement workflows programmatically