Overview

ProRata, an emerging player in the AI and content attribution space, envisioned a method to attribute outputs from large language models (LLMs) - such as ChatGPT - back to their original source materials, and then calculate payments accordingly. From the outset, the endeavour presented both a rare opportunity to pioneer in the AI sector and significant uncertainties about how best to translate a novel concept into tangible product features. ProRata enlisted Spicy Mango’s expertise to collaborate on a proof of concept (POC) of viability of their AI-based attribution model, meet tight investor timelines, and spark genuine innovation.

About ProRata

Founded in 2024, ProRata’s mission is to ensure that generative AI platforms compensate and credit content owners for the use of their material. ProRata builds technology that enables generative AI platforms to attribute contributing content sources and share revenues on a per-use basis, protecting and rewarding creators while helping to prevent unreliable content from compromising AI results.

The Challenge

Devise a platform that could identify the origins of AI-generated content and quantify how much of that content was derived from each source. Traditional AI engines process language through complex tokenisation methods, often mixing data from countless unseen sources - raising profound questions about ownership and fair compensation. Adding further pressure was the need to deliver concrete, demonstrable progress rapidly, to convince prospective investors that ProRata’s vision was not just viable, but on the cusp of revolutionising the market.

Several core uncertainties shaped the project from the outset:

Engaging Multiple LLMs

Each large language model has its own architecture, biases, and limitations, complicating attempts to refine outputs and apply consistent attribution methods.

Creating Pioneering IP

ProRata required innovative algorithms and processes that reliably mapped AI-generated outputs to their sources, carving out a new IP landscape in an emerging field.

Innovating at Pace

Tight investment timelines demanded continuous development of functional prototypes and real-world demonstrations - nothing resembling mere “vapourware.”

Navigating Collaborative Unknowns

Spicy Mango and ProRata had to co-create methods and procedures on the fly, requiring transparent communication and agile thinking to overcome unforeseen technical obstacles.

Our Collaborative Approach 

1. POC Under High Uncertainty

Spicy Mango co-located resources with ProRata in Los Angeles, forming a dedicated, cross-functional team that tackled challenges head-on. This arrangement enabled spontaneous problem-solving, rapid iteration, and immediate feedback loops. We tested a variety of LLMs - ranging from broad-market generative AI engines to more specialised language models - to determine which processes could most effectively attribute content back to its origins.

2. Crafting Pioneering Technologies & IPR

Drawing on our solid foothold in the AI and Gen AI space, Spicy Mango leveraged experience gained from working with digital publishers, media powerhouses, and innovative start-ups. Together with ProRata’s in-house experts, we worked on token-based formulas capable of dissecting AI outputs and mapping them to precise segments of underlying source material.

3. Conditioning & Refining AI Outputs

We engaged in extensive experimentation to refine text-generation outcomes and analysed token distributions and model outputs. Prompt engineering and post-processing techniques were crucial in turning opaque AI processes into traceable, verifiable data. We worked on accuracy and transparency to assess the viability of ProRata’s attribution distribution model.

4. Rapid Development for Tangible Results

The entire project was underpinned by an urgency to deliver demonstrable features that showcased tangible progress to prospective investors and early adopters. This was not about hypotheticals but genuine functionality. We relied on frequent sprints, rigorous testing, and direct feedback from key ProRata stakeholders to ensure every milestone felt like a meaningful step forward.

5. Building on AWS with Microservices

After validating the core attribution concepts, Spicy Mango architected a microservices platform on AWS that ensured each component - such as LLM orchestration, token analysis, and payment calculation - could be scaled independently. We adhered to robust security and automation best practices, allowing ProRata to expand quickly without having to overhaul its underlying infrastructure.

The Benefits of Prototype Engineering

One of the foundational pillars of the ProRata project was the deliberate choice to pursue prototype engineering rather than a prolonged “big bang” development cycle. This approach offered several advantages:

De-Risking Innovation

By releasing early and often, the team could identify technical stumbling blocks, test new ideas, and refine the approach without having invested an excessive amount of time and resources.

Investor Confidence

High-velocity prototype builds demonstrated to potential backers that ProRata’s technology was both feasible and commercially relevant, increasing the likelihood of securing additional capital.

Faster Time to Market

Prototype engineering meant each development sprint delivered something demonstrable, evaluable, and occasionally even usable in real scenarios - significantly reducing the journey from concept to tangible product.

Collaborative Problem-Solving

Ongoing prototyping fostered a culture of open dialogue and iterative learning, enabling Spicy Mango and ProRata to jointly develop cutting-edge AI solutions.

Key Outcomes

By uniting ProRata’s visionary concept with Spicy Mango’s deep AI expertise and prototype-driven methodology, the teams achieved the central achievements that underscored the project’s success and paved the way for ProRata’s continued evolution in the AI-attribution landscape below:

Proven AI Attribution Model

ProRata successfully demonstrated an operational method for mapping AI-generated text to its original sources, effectively addressing the “black box” challenge of contemporary LLMs.

Accelerated Investor Readiness

The live, AWS-hosted MVP gave stakeholders hands-on experience with the platform, fuelling ProRata’s momentum and attracting crucial capital investment.

 Flexible, Scalable Infrastructure

A microservices architecture on AWS enabled ProRata to add new features, integrate additional AI engines, and expand internationally - without disruption to existing services.

Enduring Collaboration & Innovation

Co-locating in Los Angeles created a strong partnership between Spicy Mango and ProRata, driving a shared ethos of discovery, adaptation, and iteration that has led to ongoing breakthroughs in AI-based attribution.

Conclusion

ProRata’s quest to pioneer a fair attribution model for AI-generated content stands as one of the most compelling frontiers in contemporary AI technology. By harnessing Spicy Mango’s robust expertise in AI and Gen AI, and by embracing an agile, prototype-led methodology, ProRata advanced its nascent concept into an early stage uniquely scalable platform - that helps positions ProRata at the forefront of AI-based attribution and revenue distribution.

About Spicy Mango

Spicy Mango is a specialist technology consultancy with extensive experience in AI, Gen AI, and media innovation. We partner with established digital publishers, broadcasters, and forward-thinking start-ups to accelerate platform transformations, establish new business models, and develop pioneering products. Whether you’re modernising legacy systems or creating first-of-its-kind AI IP, our hands-on, iterative approach transforms bold ideas into production-ready solutions. Contact us to discover how we can help you navigate the intricacies of AI-based innovation - and turn your vision into reality.

Learn more about ProRata at https://www.prorata.ai