The Challenge
Onyx AI’s original Zapier-based architecture created insurmountable barriers to growth, including scalability constraints that prevented handling enterprise loads, limited AI customization which reduced accuracy, integration restrictions that blocked advanced features, and inconsistent performance plagued peak periods. These technical limitations resulted in a credibility gap with high-value buyers, preventing the company from targeting enterprise customers.
Solution Implemented
Valere delivered a complete platform transformation by rebuilding the solution from the ground up on AWS. The solution featured a custom serverless backend, an iterative AI refinement methodology including prompt engineering and LLM fine-tuning, and a complete enterprise-grade UX/UI redesign with accessibility enhancements. This robust architecture prioritized production-quality AI outputs and the scalability required for mission-critical enterprise licensing decisions.
About the Company
Mission
Onyx AI provides AI-powered, expert-backed licensing advisory solutions designed to help Microsoft partners and Cloud Solution Providers navigate complex licensing decisions. The platform addresses a $25 billion market opportunity, helping CSPs reduce support overhead and maximize revenue through its “CSP Growth Engine.” By delivering instant, accurate guidance based on real-time data, Onyx AI empowers partners to capture commercial opportunities that were previously bogged down by licensing complexity.
Key Leadership
Chris Brown (CEO) leads the company with clear vision for modernizing partner success in the Microsoft ecosystem. His commitment to understanding the specific needs of CSPs and their customers was evident throughout the engagement, driving the platform’s evolution from prototype to enterprise-grade solution capable of serving customers with $100M+ in annual revenue.
The Challenge: Developing a Scalable AI Infrastructure to solve Enterprise Growth Barriers
When Onyx AI approached Valere, their platform had hit a technical ceiling. The Zapier-based architecture that enabled rapid prototyping had become their biggest limitation, creating insurmountable barriers to enterprise growth.
Technical Constraints
The Scalability Crisis
Zapier workflows were unable to handle enterprise-level loads, resulting in performance degradation during peak usage. The infrastructure lacked auto-scaling capabilities for variable AI workloads and was fundamentally not designed for production-grade AI inference at scale.
AI Quality and Customization Barriers
The platform was constrained by Zapier’s generic AI integrations, preventing the use of advanced prompt engineering or infrastructure for iterative model refinement. It produced inconsistent response quality across use cases, variable confidence levels in licensing recommendations, and unpredictable conversational tone that undermined user trust.
Market and Business Impact
These technical failures prevented Onyx AI from closing deals with target enterprise customers ($100M+ revenue) who required robust, scalable infrastructure.
In addition to dealing with limited revenue potential due to addressable market constraints, the fixed cost structure of Zapier subscriptions at scale created an unsustainable economic model where the company paid for capabilities it couldn’t fully utilize.
The GenAI Development Challenge
Beyond infrastructure, Onyx AI had to bridge the gap between “demo” and “enterprise” quality. This required managing the inherent volatility of GenAI, specifically subjective tone requirements, unpredictable model behavior, and the need for comprehensive, iterative refinement that generic platforms simply couldn’t facilitate.
Why Onyx AI Partnered with Valere for a Production-Grade AI Transformation
Chris Brown selected Valere after evaluating multiple AI/ML development firms, citing several key differentiators:
- Proven AI Competency: Valere demonstrated a non-theoretical and comprehensive understanding of LLM fine-tuning, prompt engineering, and the specific requirements of enterprise deployment.
- Honest Assessment of AI Development Complexity: Rather than overselling GenAI capabilities, Valere provided realistic expectations about AI development complexity, timelines, and the iterative refinement required to achieve production-quality outputs. The team was transparent about trade-offs and honest about what could be achieved within specific timeframes.
- Commitment to Quality Over Scope: Valere demonstrated willingness to iterate beyond initial scope to meet quality standards. The team treated Onyx AI’s success metrics as their own, with leadership remaining engaged throughout the project lifecycle rather than delegating to junior team members after contracts were signed.
- Strategic Partnership: Valere operated as an extension of the Onyx AI team, challenging assumptions and advising on technical architecture decisions that would impact long-term scalability rather than just acting as an outside vendor.
The Solution: Rebuilding an Enterprise-Grade AI Platform with Scalable AWS Infrastructure to unlock high-value corporate markets
Valere engineered a complete platform transformation, rebuilding Onyx AI from the ground up to enable enterprise-grade AI capabilities and a foundation for future growth.
Infrastructure & Architecture
Valere architected a scalable and secure AWS infrastructure purpose-built for AI-native applications:
- Compute: AWS Lambda for serverless, auto-scaling AI inference loads.
- API & Integration: Custom API development through AWS API Gateway to support specialized enterprise features.
- Data Layer: Optimized data storage using AWS RDS and S3 for transactional and document data.
- Security: Production-grade controls including IAM policies, encryption, and compliance readiness.
Iterative AI Model Refinement (Most Complex Challenge)
The AI optimization process became the most technically demanding aspect of the engagement. Valere implemented a continuous improvement methodology featuring iterative prompt engineering cycles, LLM fine-tuning for Microsoft licensing domain expertise, comprehensive output validation and quality assurance testing, and response accuracy improvement through real-world scenario testing. The team deployed prompt engineering changes, tested with actual use cases, measured accuracy and user satisfaction, identified edge cases and quality gaps, refined prompts and validation logic, and repeated until production quality was achieved. This iterative approach addressed the inherent unpredictability of GenAI behavior and subjective quality requirements that evolved through testing.
Enterprise UX/UI Transformation
The interface was completely redesigned to meet enterprise-grade design standards with user experience optimization for professional CSPs, responsive design across devices, accessibility enhancements, and performance optimization. The new design reflected the credibility and sophistication that enterprise customers with $100M+ revenue expected from mission-critical business software.
Phased Implementation Approach
Valere executed a comprehensive transformation spanning discovery and technical assessment, AWS architecture design and infrastructure provisioning, backend API development with Microsoft licensing data integration, frontend implementation with enterprise interface, iterative AI refinement cycles extending beyond initial scope, comprehensive load testing and security validation, and phased production deployment with go-live support. The engagement model emphasized transparency about timeline adjustments required for AI quality refinement, with weekly meetings focused on discussing features, testing results, and success criteria.
Post-Launch Optimization Partnership
Even after successful launch, Valere continued partnering with Onyx AI to refine AI outputs based on real-world customer feedback. Activities included real-world edge case identification, prompt engineering micro-optimizations, user experience refinements, and performance tuning based on usage patterns.
"What impressed me most was their willingness to iterate beyond the initial scope when our subjective quality requirements kept evolving. They treated our success metrics as their own and didn't declare victory until we were confident in the product."
The Results
The transformation fundamentally repositioned Onyx AI, evolving the platform from a constrained prototype into a mission-critical enterprise solution.
Key Outcomes
AI Performance & User Experience Transformation
- 50% improvement in chatbot response accuracy through enhanced prompt engineering framework and iterative model refinement methodology
- 30% increase in user satisfaction scores driven by enterprise-grade user experience, improved conversational tone, and consistent response quality
- Production-quality outputs replacing demo-quality prototypes, with edge cases handled effectively and ongoing refinement capability established
- Faster response times enabled by optimized AWS infrastructure and serverless architecture
Enterprise Market Positioning Success
- Unlocked $100M+ revenue customer segment previously inaccessible due to infrastructure limitations
- Infrastructure credibility established with enterprise buyers requiring robust, scalable platforms
- Security and compliance standards met, enabling conversations with sophisticated procurement teams
- Competitive advantage achieved through production-grade platform in $25 billion Microsoft CSP market opportunity
Operational Efficiency & Scalability
- Eliminated recurring Zapier subscription costs through pay-as-you-go AWS model optimized for actual usage patterns
- 40% efficiency gain in feature implementation enabled by flexible AWS architecture and custom API infrastructure
- Auto-scaling capabilities for variable AI workloads ensuring consistent performance during peak usage
- 99.9%+ uptime reliability providing enterprise-grade availability standards
Strategic Business Impact
Market Position Transformation
Before the AWS migration, Onyx AI was limited to small-to-mid market customers with an infrastructure credibility gap preventing enterprise conversations. After the transformation, the company is actively targeting enterprise customers, confidently discussing infrastructure capabilities in sales conversations, and positioned as an innovator in the Microsoft partner ecosystem with CSP Growth Engine featured in IT Channel Oxygen press coverage.
Revenue Growth Foundation
The platform transformation positioned Onyx AI to capture share of the $25 billion Microsoft CSP market with enterprise segments now addressable, higher customer lifetime value potential, and scalable infrastructure supporting aggressive growth plans without technical constraints.
The Client’s Perspective
"Valere's AI capabilities are the real deal. Many firms claim generative AI expertise, but Valere's team has demonstrated actual competency in prompt engineering, output validation, and iterative model refinement. The team doesn't oversell what AI can do and is honest about the work required to achieve production quality. …If you're serious about building production AI solutions—not just demos—they're worth a conversation."
Check the full client reviews on:
