Docusign Developers

Taking an MCP server to production means solving a different class of engineering problems. Connecting an LLM to tools is one thing. Making its behavior reliable and safe across model updates, tool changes, and releases is another. In this engineering deep dive, Arvind Kurmi, Sr. Engineering Manager at Docusign, shares how our engineering team approached production readiness: - Guided MCP Apps for predictable, critical workflows - Automated evals to validate tool calls and task completion - Scheduled production checks to catch behavior drift - Continuous security testing built into the release pipeline - Quality gates that stop changes when evals fail One measurable outcome: regression testing went from ~80 hours to under two. 🚀 See how we engineered the Docusign MCP Server for production. https://lnkd.in/eNJYdhvW

Find the right agreement records and return only the data your workflow needs. Video 4 in our Agreement Manager API series walks through using OData with Postman to filter agreements by status, type, and date, select specific response fields, and search records by keyword. 🎥 Watch the full walkthrough and build along in the series. https://lnkd.in/eevKww-S

Since getting 100% of our engineers on AI tools, our teams have built a coding agent that turns repetitive tasks into PRs, an internal platform with over 300 custom agents, and an observability agent that watches telemetry across 400+ services. Getting from those individual wins to something the whole org could build on came down to systems. I shared the five levers on stage at ELC Annual last week from individual wins to something the whole org benefits from came down to systems. 1. Define "good" before you scale. Treat adoption as an input and value and workflow outcomes as the scorecard. 2. Scale your goodness. Turn each win into a reusable skill, agent or workflow, so the next team starts from it. 3. Put accountability and guardrails in the pipeline. Access is regularly governed, Code is tested before review, AI-generated code is labeled, and a human owns every merge. 4. Make it sustainable. Give every capability an owner and a feedback loop, so it outlives the person who built it. Measure and optimize your cost/tokens and tokens/task. 5. Teach the AI *your* systems, *your* language, *your* organization and *your* rules, so it starts in the right place (aka Context is key). I also walked through our AI adoption journey with Jerry Li on the Engineering Leadership Podcast: https://lnkd.in/gnGrgEhU

I recently had the opportunity to be featured in Docusign’s “Meet the Engineer” series. In the article, I share some reflections on leading our post-quantum cryptography work for Trust Products, solving complex customer problems, and what engineering leadership means to me. One principle that guides me is staying close enough to the technical details to ask the right questions, while giving experienced teams the space to make decisions and grow. The strongest results come when people feel trusted, supported, and accountable for the impact of their work. If you’re interested in working on challenging problems in digital trust, cybersecurity, and next-generation technology with talented people who challenge assumptions and stay close to the customer, take a look at the open roles at Docusign. Read the full article: https://lnkd.in/dSTj5wq4 #Docusign #LifeAtDocusign #Engineering #EngineeringLeadership #DigitalTrust #TrustProducts #PostQuantumCryptography #Cybersecurity #Innovation #MakeYourMark #TechCareers #Hiring

Our Director of Data and AI, Jan Štihec, sat down with Sagnik Nandy, CTO of Docusign, to explore what it takes to build AI that actually understands your business. Smarter models, longer context windows, and out-of-the-box RAG aren’t enough. AI agents need context that captures your business rules, workflows, relationships, and user permissions to work reliably. Watch for practical advice on solving the context problem - from structuring enterprise data and capturing business knowledge to managing permissions, measuring accuracy, and scaling AI reliably. Watch the full interview on AugmentedU: https://lnkd.in/gnqq2e-5

Build more flexible agreement workflows by programmatically updating agreement classifications. Video 3 in our Agreement Manager API series walks through how to retrieve agreement records, fetch available types with GET /agreement-types, update an agreement with a PATCH request, and verify the change using Postman. 🎥 Watch the full walk through and continue building along in the series. https://lnkd.in/eHD92ZM9

Build AI agents around the workflows your organization already trusts, instead of rebuilding automation from scratch. In this Microsoft Copilot Studio Agent Academy mission, see how to connect the Docusign MCP Server to Copilot Studio and invoke existing Docusign workflows through conversational AI. You’ll build an HR onboarding agent, incorporate human approval, and combine Docusign and Microsoft MCP tools to automate a broader business process. Great to see our partner ecosystem putting these integrations into practice and making it easier for builders to connect AI agents with existing enterprise processes. 🚀 Try the hands-on mission and earn a digital badge. https://lnkd.in/e-NXsTXm

Docusign API or MCP Server? The right choice depends on how you need to work with agreement data. Contracts can describe the same concept in different ways, making keyword-based searches unreliable across large repositories. Paige Rossi walks you through how the Docusign Agreement Manager API and MCP Server can help you retrieve, filter, and query AI-extracted agreement data, with examples of when each approach makes sense for your specific task. https://lnkd.in/eXKhY6RT