How BMC's CCO Built AI Agents That Triple Productivity ft. Sofia Barbosa (BMC Software)

In this episode, Sofia Barbosa, CCO at BMC Software, walks through five AI agents her team built or is building across a 415-person global CS org supporting 400-plus products. She tripled success story output in under three weeks. She cut SOW creation from three weeks to a day. And she got to zero support escalation red alerts for the year. None of it started with a business case. It started with a prototype.

---

Listen on YouTube, Spotify and Apple Podcasts

---

Three Things We’re Taking Away:

1. Start with the output bottleneck, not the strategy. Sofia’s first agent targeted success stories because she had a specific target she couldn’t hit manually without one to two months of effort per batch. The agent did it in under three weeks with 3x the output. That result created internal momentum for everything that followed.

2. Build dedicated agent teams, not just agent tools. Sofia created two formal internal functions: Succeedo Forge, which builds and designs the agents, and Succeedo Care, which maintains and supports them post-launch. Treating agents as infrastructure that needs ongoing ownership is what separates deployments that stick from ones that quietly die.

3. Sentiment-based ticket prioritization is a faster path to zero escalations than SLA management alone. SupportLogic reads the emotional state of customer communications inside existing priority tiers and surfaces which P3s are about to become P1s before they do. BMC had zero support escalation red alerts for the year.

🎯 The [Un]Churned Take: Five Agents, One Platform, a 415-Person CS Org

Sofia Barbosa, Chief Customer Officer at BMC Software, oversees a global CS organization of 415 people supporting 400-plus products across professional services, customer success management, support, and customer experience. When the company split last September and her team had to do more with less, she didn’t reorganize. She built Succeedo.

The Five Agents and What They Actually Do

Succeedo is BMC’s internal AI platform built on Microsoft Copilot Studio, connected to Salesforce, Gainsight, and external data sources. It’s now home to five discrete agents at various stages of deployment.

The first, and the one that started everything, handles customer success stories. CSMs enter a company name and use case. Succeedo pulls data from connected systems and the web, drafts the story, and a CSM validates before it’s published internally or sent to the advocacy team. Output tripled in under three weeks. What used to take one to two months per batch now takes a day.

The second is value summaries: an annual document per customer that pulls from pre-sales business cases, Gainsight adoption data, and customer records to build a narrative of what value the customer has actually realized. The output is a PDF the CSM reviews and presents at a QBR. The goal is to arrive at every renewal conversation with documented proof of value already signed off by the customer.

The third is an education agent that maps BMC’s full catalog of 400-plus training courses and certifications into a personalized learning path for individual users or teams. Currently internal, Sofia wants to push it to customers through a newly launched customer portal. Metrics will follow training attendance, course completion, and product adoption.

The fourth is SupportLogic, which does sentiment-based prioritization of the support backlog. It reads the emotional state of customer communications across all ticket tiers and surfaces which ones are critical regardless of their assigned priority level. A P3 that a customer is genuinely distressed about gets flagged before it escalates. The result: zero red alerts for the year.

The fifth is still in development: an agent for statements of work. Complex professional services engagements involve dozens of dependencies across BMC, the customer, and any third-party partners. Succeedo pulls from similar past projects and pre-populates a SOW template. Sofia estimates it will save three weeks of manual work per engagement. The next step is automating the delivery of the finished document directly to the customer.

The Org Structure That Makes It Sustainable

Sofia formalized two internal teams to own the platform. Succeedo Forge designs and builds new agents. Succeedo Care maintains and supports them once live. These aren’t new hires. They’re reclassified roles. But making the distinction explicit is what keeps the agents from drifting or degrading after launch.

The Advice She’d Give Every CCO

“Stop sometimes speaking so much. We need more action. With action will come knowledge.” CFOs are already asking for AI ROI, and most CS leaders don’t have clean answers yet. Sofia’s position: the way to get clean answers is to experiment, measure what you can, and fail forward. The sensitivity matrices she’s built on Succeedo’s early results are more useful than any pre-implementation business case would have been.

Wrapping Up

Five agents. Two dedicated internal teams. One platform. Built in months, not years. The CCOs who’ll have the ROI answers CFOs are asking for are the ones who started experimenting before the question was asked.

See you next week 🧠

---

Where to find Sofia Barbosa:

• LinkedIn: https://www.linkedin.com/in/sofiabarbosa/

---

Referenced in This Episode

• Microsoft Copilot Studio — The platform BMC built Succeedo on. Allows non-engineers to build and deploy AI agents connected to existing enterprise data sources. copilotstudio.microsoft.com

• SupportLogic — The AI-powered support intelligence platform BMC uses to analyze customer sentiment across their ticket backlog and flag which cases are at risk of escalating regardless of priority tier. supportlogic.com

• Gainsight — One of the core data sources feeding Succeedo’s value summaries and customer success story agents. gainsight.com

• Salesforce — BMC’s CRM, also connected to Succeedo as a data source for customer records across agent workflows. salesforce.com