DocuSign Iris AI Agents: B2B RevOps + Sales-Ops Guide

DocuSign used its Momentum ’26 conference on May 21 to make the most consequential platform expansion the company has shipped since launching Intelligent Agreement Management in 2024. Three new capabilities went into early access in the US the same day: an AI assistant built on Iris, the company’s agreement-tuned AI engine; a set of agents that review contracts, monitor obligations, and trigger next steps without human follow-up; and Agent Studio, a no-code builder for teams to deploy custom agents against their own renewal, approval, and deal-management workflows. US general availability lands in July.

The release also formalizes DocuSign’s bet on the Model Context Protocol. Iris connects natively to Anthropic Claude, Google Gemini, and OpenAI ChatGPT via MCP, alongside deep integrations with Coupa, Microsoft Copilot, Salesforce, SAP, and Slack. A separate legal-AI partnership brings Harvey, Legora, and CoCounsel by Thomson Reuters into the agreement workflow. Pricing for the assistant and agents has not yet been disclosed; the MCP server is in beta today globally, in English.

For B2B RevOps, sales-ops, legal-ops, and procurement teams, this lands at a specific moment. HubSpot and Salesforce have spent April building out the front-office and back-office sides of the agentic-CRM market; DocuSign is now planting the third flag on the agreement workflow that sits between them. The pieces are in early access; the planning question for the next 60 days is which workflow to pilot first when GA opens in July.

DocuSign Iris vs. HubSpot Breeze vs. Salesforce Agentforce

What DocuSign Actually Shipped

The release sits inside the Intelligent Agreement Management platform (IAM) DocuSign rebuilt over the last 18 months. Iris is the AI engine underneath; it analyzes agreement data, surfaces key terms and obligations, and lets teams ask questions in natural language. The new assistant and agents are the user-facing layer on top of Iris. The agents do three things the assistant cannot: they review contracts against the company’s standards, they keep work moving automatically in the background between human touchpoints, and they let RevOps build custom workflows in Agent Studio without engineering involvement.

Review acceleration is the marquee capability. Agents can check a draft against company standards, suggest specific edits, and prepare the document for approval in minutes rather than days. The autonomous-monitoring layer is the structural one. Agents watch contracts in the background, flag risks (renewal-window obligations, indemnity clauses outside policy, missing signatures on amendments), track obligations across an entire portfolio, and trigger next steps without anyone having to remember to follow up. That is the operational gap most legal and procurement teams have been filling with spreadsheets and calendar reminders for years.

Agent Studio is the load-bearing piece for RevOps and sales-ops teams. It lets a non-engineering owner build a custom agent against a specific workflow: monitor renewals 90 days out, escalate to the CSM when a renewal-stage discount request hits a threshold, route procurement spend over a certain dollar amount to legal review automatically. The Studio canvas is no-code, the agents deploy without engineering bandwidth, and they connect to the existing IAM workflow library. Whether the abstraction holds for genuinely novel workflows or breaks the moment a team tries to build something outside the template set is the question 90-day pilots will answer.

Why This Positions Iris as the Third Leg of the Agentic-CRM Stool

The B2B revenue-tech market entered 2026 with three platform vendors all pointing at the same agentic-workflow surface from different starting positions. HubSpot’s Breeze suite occupies the marketing and customer-service side, with the Customer Agent and Prospecting Agent already on outcome-based pricing. Salesforce’s Agentforce occupies the sales and operations side, with the Operations SKU shipping back-office workflows on consumption pricing. DocuSign’s Iris now plants the flag on the agreement workflow that lives in the middle, where sales hands off to legal, legal hands back to sales, procurement intersects both, and renewals get owned by whoever has the calendar reminder set.

That third-leg positioning is stronger when measured against the Agentforce paid adoption gap, because Salesforce has awareness before it has universal deployment.

Our read: the three platforms are no longer competing on AI features. They are competing on which surface the agent gets to live in. HubSpot agents live in the CRM record. Salesforce agents live in the operational workflow. DocuSign agents live in the agreement itself. For a B2B team running all three platforms, the agents will collaborate or compete depending on which workflow surface owns the source of truth for any given action. A renewal triggered from a CRM record (HubSpot or Salesforce) but executed against a contract obligation (DocuSign) now has three agents that could plausibly fire the next step. The question for RevOps is which one wins, and on what authority.

The competitive pattern mirrors the same shift documented across the agent stack. Salesforce Agentforce Operations launched April 29 with the same end-to-end-workflow framing DocuSign just adopted: agent completes the work, surfaces only exceptions. HubSpot’s outcome-based pricing for Breeze agents moved the SaaS pricing model in April from per-seat to per-resolved-outcome. DocuSign has not yet disclosed pricing for Iris assistant or agents, but the architectural choice (agent does the work) implies the same consumption-or-outcome pricing logic will follow. Per-seat pricing on agents that operate without seat-holders is the model nobody is shipping in 2026.

The MCP-First Architecture Is the Real Differentiator

DocuSign’s choice to ship MCP-native integration with Claude, Gemini, and ChatGPT is the technical decision that matters for B2B teams thinking past 12 months. The Model Context Protocol is the open standard that lets external AI clients query and act on Iris-managed agreement data without bespoke integration work. A sales-ops team running Claude as their primary internal AI assistant can ask Claude to surface unsigned amendments across the portfolio, and Iris responds through the MCP layer without anyone writing custom API glue.

That architectural bet aligns DocuSign with the broader infrastructure shift across the B2B martech and revenue-ops stack. Twilio’s conversation layer made GA at SIGNAL 2026 on the same principle: persistent state, managed handoffs, and an infrastructure tier sitting between the underlying communication channels and the AI agents working on top of them. The agreement layer is the same idea, applied to documents instead of conversations. The teams that win the back half of 2026 will be the ones whose agreement, conversation, and CRM data all expose themselves to the same agents through open protocols, rather than the ones stitching point-to-point integrations between proprietary stacks.

Agent Studio is also a direct architectural mirror of a competitor pattern shipping in the same quarter. Outreach launched Omni and Agent Studio in April with a near-identical pitch: no-code agent builder for RevOps, custom agents deployed without engineering, visual canvas plus ready-to-run workflow templates. The fact that two enterprise vendors converged on the same form factor inside six weeks is the strongest evidence that Agent Studio is the right shape for the no-code-agent category, not just a one-vendor experiment. The teams that pilot both will end up with portable mental models for what custom-agent ownership looks like inside RevOps, regardless of which vendor wins the budget line.

The legal AI partnership tier is the part of the announcement most B2B teams will under-rate. Harvey, Legora, and CoCounsel by Thomson Reuters are the three names dominating in-house legal AI procurement conversations right now. Wiring all three into the Iris workflow at the platform layer means the legal review step inside a sales contract, a procurement agreement, or an HR offer letter can fire the legal-AI tool directly without anyone having to copy-paste a clause into a separate window. The legal-ops gain is structural: the agreement workflow becomes the integration point for the legal AI tools, rather than a separate decision the legal team has to make about which AI to standardize on.

What B2B RevOps and Sales-Ops Should Do Before July GA

Four moves for the next 60 days.

Decide which workflow goes first, before the early-access pilot consumes the runway. Renewals, vendor onboarding, sales-contract review, HR offer-letter routing, and procurement spend approval are all candidates. The wrong pilot is a complex, judgment-heavy, low-frequency process. The right pilot is high-frequency, low-judgment, and currently lives in spreadsheets and email. Renewal monitoring is the highest-yield first pilot for most B2B teams because the workflow is well-defined, the cost of a missed renewal is large, and the agent’s signal vs. noise ratio is easy to measure against a known baseline. Cresta’s real-conversation persona model adds a useful testing pattern: build repeatable simulations before the customer-facing workflow reaches production.

Map the Iris-Salesforce-HubSpot agent overlap before the GA cycle. If your stack runs Salesforce or HubSpot as the CRM and DocuSign for agreements, the agentic layer is about to double-fire on renewal and contract-amendment workflows. Define explicitly which platform owns the source of truth for any given trigger. Without that decision, the agents will compete to fire the same next step, and the rep will get conflicting nudges from two systems. The same coordination question already applies to HubSpot’s Prospecting Agent and Outreach Omni on the outbound side; the renewal motion is where the DocuSign-CRM overlap surfaces.

Audit your agreement template library for agent-readiness. The agent is grounded in your company standards and your existing agreements. If your templates live in mixed formats (PDFs, Word docs, scattered Google Drive folders, an unmaintained CLM), the agent’s review logic will inherit the inconsistency. Use the early-access window to consolidate the template set, document the redline standards explicitly, and tag the clauses the agent should treat as policy boundaries. A clean template library is the prerequisite for trustworthy agent suggestions.

Plan the legal-AI integration tier now, not at GA. If your legal team is already piloting Harvey, Legora, or CoCounsel, the Iris MCP layer means the integration can route automatically. If they have not yet picked a legal AI tool, the DocuSign partnership shortlist is the place to start the evaluation. Legal-ops vendor selection takes 60-90 days; running it in parallel with the Iris early-access pilot means the integrated workflow is ready when GA opens in July rather than three months after.

The release is narrow in scope at launch (US early access, English MCP beta) and broad in implication. Every B2B platform that touches agreements is now under pressure to decide whether to build agentic capabilities themselves, integrate with Iris through MCP, or accept that DocuSign is becoming the agent-addressable surface for the agreement layer of the stack. Teams that pilot in Q2 will have read the architecture before procurement decisions hit in Q4. Teams that wait will inherit other people’s pilots.

Frequently Asked Questions