Iliana AI
Iliana AI is a B2B sales technology company that has built one of the first AI-powered digital sales agents for enterprise lead qualification — a real-time streaming avatar that engages website visitors in natural conversation, asks discovery questions, and routes qualified leads directly to the sales team.
Case Details
Clients: Iliana AI (www.ilianaai.com)
Period: February 2026 – July 2026
Tags: Artificial Intelligence, B2B SaaS, Streaming Infrastructure, Software Testing
Project Duration: 1 year
Let’s Work Together
Call us directly, submit a sample or email us!
The Story
Iliana AI set out to solve one of the most persistent challenges in B2B sales: the gap between inbound interest and qualified pipeline. Their product — an AI-powered avatar embedded directly into a company’s website — was designed to bridge that gap, qualifying visitors through real conversation at any hour of the day, without the overhead of a live SDR team.
The technology behind it is sophisticated by any standard: a live streaming video avatar with synchronized voice and lip movement, real-time speech recognition, a proprietary conversation intelligence layer trained on enterprise sales methodologies, and a WebRTC transport infrastructure capable of sustaining high-quality audio and video over a standard browser connection.
As the product matured and conversations grew more complex, Iliana AI sought a quality engineering partner capable of working across the full depth of the stack — from the AI infrastructure layer down to the end-to-end visitor experience. That partnership is what brought them to SQA Service.
The Challenges
Building a reliable AI sales agent is not a single engineering problem — it is a coordination challenge across multiple specialized systems that must perform together, in real time, under real visitor conditions.
The product combines a third-party streaming avatar platform, a WebRTC transport layer, a custom backend proxy, a conversation intelligence service, and a browser-based widget. Each layer has its own operational characteristics and failure modes. Ensuring that the full system behaves correctly — and recovers gracefully when any component is under stress — requires a testing approach that treats the product as a system, not as a collection of individual parts.
At the same time, the product’s user-facing layer — the website, the widget, and the conversation interface — needed to deliver a consistent and professional experience across browsers, devices, and deployment contexts. For a product that represents the first impression a prospect has of a client’s sales process, that consistency is not optional.
The key challenges included:
System-level reliability at scale: Ensuring the AI infrastructure performs consistently across the full duration of real sales conversations, not just in controlled demo conditions.
Cross-system observability: Building the visibility needed to understand how the product behaves end-to-end, across layers that were not designed with shared monitoring in mind.
Interface consistency across environments: The widget is deployed on clients’ existing websites, each with its own design context, browser base, and device mix. Validating visual and functional consistency across that range required a structured and repeatable approach.
Quality coverage for a fast-moving product: Establishing a testing practice that could keep pace with ongoing product development, catching issues before release rather than after deployment.
The solution
SQA Service engaged with Iliana AI as a full-scope quality engineering partner, working across the AI infrastructure, the application layer, and the user-facing product.
AI infrastructure quality assurance. We built instrumentation and telemetry to observe the full streaming session lifecycle under real conditions — capturing timing, transport-layer metrics, and system behaviour across the end-to-end request flow. This gave the team objective, data-driven visibility into how the product performs during extended conversations, and the foundation for ongoing infrastructure validation as the platform evolves. We also delivered a strategic technical roadmap for long-term reliability and architectural alignment with the underlying platforms.
Functional testing. We developed and executed a comprehensive functional test suite covering the complete visitor journey: from initial page load through avatar session initiation, active conversation, lead qualification, and session completion. Coverage was extended to include edge cases and boundary conditions that only emerge under sustained, real-world usage patterns. Test documentation was structured to serve both the current release cycle and future regression testing.
User interface testing. The product’s visual layer was tested across browsers, devices, and screen sizes. The avatar widget — which must render correctly within the design context of each client’s website — was validated for layout consistency, interaction states, loading behaviour, and the full range of session states a visitor might encounter. Findings were documented and prioritised to support the development team’s release planning.
UX review and validation. Beyond functional correctness, we evaluated the end-to-end visitor experience against the product’s core purpose: qualifying leads efficiently and without friction. This included the flow of the conversation interface, the clarity of in-session status indicators, the coherence of the experience across the full session lifecycle, and the handling of exceptional conditions from the visitor’s perspective.
Ongoing quality partnership. SQA Service continues to work with Iliana AI as a standing quality partner — reviewing new features ahead of release, maintaining and extending the test suite as the product grows, and providing structured input into the development process to embed quality earlier in the cycle.
The result is a product that performs reliably under real production conditions, presents a consistent and polished experience to visitors across environments, and has a quality practice in place to sustain that standard as the platform continues to scale.
Impact
The engagement delivered measurable improvements across every dimension of product quality that matters to Iliana AI’s business.
Session reliability. The AI avatar now sustains the full duration of real sales conversations without degradation. Extended sessions — the ones where qualification actually happens — complete as intended, giving the sales intelligence layer the complete conversation context it needs to score and route leads accurately.
Visitor experience. The product presents a consistent, professional experience across browsers and devices. Session states are communicated clearly to the visitor at every stage, and exceptional conditions are handled gracefully rather than silently. The interface holds up in the deployment context of client websites, regardless of design environment or browser base.
Development confidence. With a structured test suite in place and a standing QA partner reviewing releases, the Iliana AI team ships new features with significantly greater confidence. Issues are identified during development rather than after deployment, reducing the cost of fixes and the risk of regressions reaching production.
Strategic clarity. Beyond immediate quality improvements, the engagement gave Iliana AI a clear technical roadmap for the evolution of their platform infrastructure — a prioritised, phased plan that the team can execute against as the product scales.
For a product that sits at the intersection of AI, real-time video, and enterprise sales, quality is not a support function — it is a core enabler of the value proposition. That is the foundation this engagement was designed to build.
