Overcoming Governmental and Enterprise Barriers – Without Breaking the Bank

The ability to extract meaningful insights from customer interactions is an increasingly hot corner of the business intelligence market. Whether it’s sales calls, customer service engagements, or even internal meetings, these conversations are rich with data that can fuel AI engines – if captured correctly. But for AI software developers building applications that need to extract transcripts – and therefore data – from Microsoft Teams interactions, what looks straight-forward on paper is incredibly complex in practice.

In a recent conversation with Empowering Cloud’s Tom Arbuthnot, Numonix’s CEO Michael Levy and CMO Stephen Denny explore this issue and shine a bright light on how interaction capture in Teams environments is coming under increased scrutiny from governmental as well as enterprise-level priorities.

The Constraints AI Developers Face When Building Around Teams

Microsoft Teams presents unique challenges for developers aiming to capture and process interaction data securely and compliantly. As Tom Arbuthnot explains, there are two primary ways to access Teams call data:

Certified Recording APIs: Originally designed for compliance, these APIs are now being repurposed for AI-driven insights. However, they require certification and deep integration with Microsoft’s evolving Graph API.
Bot-Based Capture: Some third-party tools use bots that join meetings as guests to record interactions. While technically feasible, this method raises serious concerns around data privacy, enterprise security, and regulatory compliance – especially in regions governed by privacy laws like GDPR.

The problem is compounded by Microsoft’s increasing restrictions on guest bots, including lobby controls and capture limitations. Make no mistake, enterprises are asking for these restrictions for real policy-driven reasons. Most modern enterprises are pushing back against unauthorized recording – and by extension, data extraction – and Microsoft is responding by tightening access. Developers who rely on bot-based capture are very much at risk of being left in the lobby.

Moreover, the Graph API itself is complex and constantly evolving. As Arbuthnot notes, “Graph is a beast,” and staying current to changes requires significant resources. For AI developers focused on building value through insights, these technical hurdles can be a major distraction.

How Numonix’s TRAAS Overcomes These Challenges

Enter Numonix’s TRAAS—Teams Recording as a Service—a solution purpose-built to address these constraints. Michael Levy outlines how TRAAS provides developers and enterprises with a secure, compliant, and flexible way to capture high-quality audio from Teams interactions.

Key advantages of TRAAS include:

Certified and Secure Architecture: Numonix is ISO 27001 compliant and our TRAAS product is a component of our SOC 2 type II accredited IXCloud platform, ensuring enterprise-grade security. This certification is often a prerequisite for working with large organizations.
High-Quality Audio Capture: TRAAS delivers stereo audio at PCM 16,000 Hz, enabling clean transcripts and accurate AI processing. This is critical for feeding AI engines with reliable data.
Rules-Based Recording Engine: Built on Numonix’s iXcloud platform, TRAAS allows granular control over what gets recorded. Developers can exclude sensitive calls (e.g., HR meetings) and target specific interaction types like PSTN calls or internal meetings.
Rapid Adaptation to Microsoft Updates: Numonix has made a deep commitment to keep pace with Microsoft’s changes. For example, when Teams introduced support for conference mode and call queue recording, Numonix was among the first to implement it—ensuring uninterrupted service for customers.

TRAAS also supports legacy integration, allowing companies transitioning to Teams to maintain familiar interfaces while gaining access to new media sources. This flexibility makes it easier for developers to plug into existing workflows without reinventing the wheel.

What Customers Are Saying

One of Numonix’s marquee customers, Gong, recently shared their experience with TRAAS. Gong, a leader in revenue intelligence, needed a secure and certified way to extract media from Teams for their AI-driven platform. As Arbuthnot recounts, Gong was impressed by the ease of integration and the ongoing support provided by Numonix.

“They wanted a proper way to get the media out of Teams,” Arbuthnot said. “And they’re working with enterprise customers, so they need to hit those requirements or they’re not getting in the door.”

Levy adds that Gong’s focus on revenue intelligence aligned perfectly with Numonix’s specialization in secure capture. “We power their revenue intelligence under the hood,” he said. “When we first engaged, we didn’t have SOC 2. They held it up as a requirement, and we delivered.”

This partnership highlights the value of TRAAS not just as a technical solution, but as a strategic enabler for AI companies. By offloading the complexity of media capture, developers can focus on building insights and delivering value to their customers.

Final Thoughts

For AI developers building around Microsoft Teams, the path to reliable, compliant, and high-quality interaction capture is fraught with challenges, from evolving APIs to enterprise security concerns. Numonix’s TRAAS offers a compelling solution that removes these obstacles, enabling developers to focus on what they do best: creating transformative AI experiences.

As Levy puts it, “You don’t want to be in the weeds developing your own mechanism to feed it. You need reliable, clear capture that is secure. That’s exactly where we play.”

If your AI platform depends on Teams data, TRAAS might just be the missing link between your vision and execution.