An Engineering Lead at Rilla discusses the journey from identifying market gaps to developing a revenue-generating, sales coaching platform that is reshaping sales performance.
The landscape of sales enablement is undergoing a notable shift, driven by the integration of artificial intelligence. Historically, sales coaching has been a reactive process, with managers providing feedback after calls are completed.
This model is now being challenged by emerging technologies that offer real-time guidance, aiming to influence outcomes as they happen rather than just analyzing them in retrospect. At the forefront of this trend is Alexander Jabbour, an Engineering Lead at Rilla, where he directs the development of the company's real-time product initiative.
His work, which has already generated multiple 7 figures of revenue for Rilla and is projected to reach 8 figures by the end of the year, provides insight into the complex process of creating disruptive technologies. With a background in building innovative products from the ground up, Jabbour has focused his career on turning nascent concepts into market-ready solutions.
From Frustration to Innovation
Traditional sales management suffers from a critical flaw: the gap between live customer conversations and feedback sessions. This delay transforms what could be immediate learning moments into missed opportunities, creating the urgent need for real-time coaching platforms that bridge this divide.
This gap was a recurring theme in user research conducted by Jabbour. He notes, “I kept hearing the same frustration in user interviews: coaching is almost always reactive. Managers only see what happened after a call ends, which late to influence the outcome or build a rep's confidence in the moment.”
A structured sales enablement framework often helps identify such points of friction. Jabbour recalls the realization: “That was the spark: if we could listen, understand, and surface the right nudge while the call is still live—and give managers a way to actively support reps in the moment-we could change outcomes, not just reports.”
This focus on immediate intervention reflects a broader industry trend, with reports indicating that 90% of companies have either implemented AI in their sales processes or plan to do so.
Overcoming Initial Product Hurdles
Transforming an idea into a functional product involves navigating significant technical and strategic obstacles. For real-time audio analysis, the technical demands are stringent.
Jabbour identifies early hurdles, including “The challenge of capturing and streaming high-quality audio from devices and networks with minimal interruptions, while preserving user consent and privacy,” alongside achieving low-latency speech-to-text in noisy environments.
Different AI sales coaching platforms address intervention in various ways, but a clear strategic roadmap is essential. This aligns with the 'Exploration' phase of organizational AI enablement, which focuses on analyzing workflows to identify opportunities.
“Before writing code, I produced a long and detailed engineering design document that laid out the architecture, failure modes, and measurement plan,” Jabbour explains. This detailed planning was critical for establishing internal alignment and moving forward without compromising system integrity.
Identifying Core User Pain Points
To ensure a product addresses genuine need, a deep understanding of user pain points is necessary. Through interviews and shadowing live sales calls, several key themes emerged as pain points: timing, consistency, and scale.
These issues prevent sales teams from reinforcing best practices effectively. “The biggest lost opportunity was the lack of on-the-spot coaching. Managers knew what they would have said—only after the call,” Jabbour states.
This insight underscores a key challenge that real-time coaching aims to solve, as sales reps can forget up to 70% of training within a week. Jabbour mapped these pain points directly to product features, such as live call visibility and alerts.
This user-centric approach ensures technology is applied with purpose. “That ensured we weren't building ‘AI for AI's sake,’ but targeting the exact friction managers were living with,” he adds. This focus aligns with reports that teams using AI-powered coaching are 36% more likely to achieve higher win rates.
Designing for Speed and Scale
The architecture of a real-time system must be optimized for speed, as insights delivered minutes too late lose their value. The design process involved breaking down the problem into manageable components with strict performance budgets.
Jabbour shares, “I decomposed the problem into a low-latency streaming pipeline with strict budgets at each hop,” which included audio streaming, transcription, and an AI model for insights.
This modular structure is common in multi-agent systems that use decentralized communication protocols to coordinate tasks. The guiding philosophy was centered on immediate utility.
“The design principle was simple: if it can't show up in seconds and materially help a human make a better decision, it doesn't belong in the real-time path,” Jabbour says. This principle is a core tenet of effective agentic workflows, where components interact with external resources to achieve a goal.
Validating Impact with Data
Before scaling a new product, its real-world impact must be validated through measurable results. The initial rollout was approached methodically with a beta test involving an in-person company with ~40 sales reps.
“We iterated with them weekly, and after six weeks, determined it was the right time to conduct an ROI analysis,” Jabbour recalls. This process of evaluating a pilot program requires clear benchmarks to measure progress.
The outcomes of this initial test were significant, showing a close rate increase from 29% to 34%. “This impact validated our approach,” he states. This data-driven validation provided the confidence to invest further in features with proven value. Such results are consistent with broader industry findings, where deals using AI-recommended actions have a 50% higher average win rate.
From Prototype to Platform
Transitioning a product from an early prototype to a stable platform requires a strategic shift from discovery to reliability and scalability. This evolution involves moving from disruptive concepts to disciplined execution.
“As we scaled our user base, it was clear that we needed to shift the focus from building 0->1 and attempting to figure out what works, to making the product stable, scalable & performant under high load,” Jabbour explains.
This journey reflects the difference between disruptive and incremental innovation, where an initial breakthrough must be followed by continuous improvement. This process also involved team growth.
“I built the first versions solo, then brought in more engineers, and now lead a team of four across engineering and design. That lets us scale features without sacrificing our speed and quality bar,” Jabbour notes. This expansion enabled a focus on productization, security, and ROI analysis, crucial steps for any company moving from a low-end disruptive innovation to a full-market solution.
Balancing Speed and Stability
When building mission-critical business applications, developers face a fundamental challenge: moving fast while building something reliable. Once your system goes live and real users depend on it, trust becomes everything.
A core principle is to prioritize depth over breadth in the early stages. “Get one golden path unbelievably solid before adding breadth. Quality over quantity - Build guardrails on day one,” he advises.
This approach is supported by technical safeguards such as feature flags, which prevent a delicate feature release from compromising reliability. Early investment in system monitoring is also crucial. As Jabbour states, “Invest early in observability. It makes the difference between a five-minute blip and a five-hour incident.”
These practices are crucial when companies want to add AI-powered tools to their existing workflows. The goal is to make processes smarter and more efficient without creating new risks that could hurt the business.
The Future of Agentic AI
Looking ahead, the evolution of AI is pointing toward more autonomous and capable systems. The next generation of products will likely move beyond providing insights to executing complex tasks.
“One thing that excites me is building agentic copilots that help humans make better decisions across every customer-facing moment—sales, success, and support,” Jabbour says.
This vision extends to creating systems that can manage end-to-end business functions, a concept central to the reshaping of business execution through agentic AI. Such advancements in agentic AI for sales could automate everything from lead qualification to proposal generation.
Jabbour envisions: “Agents that are voice-first and multimodal, safe by default. They'll pair frontier models with on-device compute and new hardware to deliver ‘personal assistants’ for everyone.”
The development of real-time AI coaching represents a critical step in how organizations support their sales teams. The journey from identifying a market frustration to building a scalable, data-validated platform highlights a disciplined approach to 0-to-1 product creation.
By focusing on tangible outcomes and user-centric design, innovators are showing that technology can transform not just processes, but also performance.
