ENACT AI

Businesses want to use AI. Building it is another matter.

For many startups, SMEs and specialist organisations, adopting advanced AI means dealing with significant barriers.

The opportunity behind NAPSC was to reduce this complexity and make the path from business data to a usable AI system more accessible.

From proprietary data to a customised AI solution.

NAPSC is designed to guide organisations through the AI development lifecycle in a no-code/low-code environment.

Items marked as future concepts were part of the platform's longer-term direction rather than functionality presented as shipped.

Fine-tuning isn't always the answer.

One important product consideration was avoiding the assumption that every proprietary AI use case requires model fine-tuning. The platform concept considered both fine-tuning, for cases where adapting model behaviour is appropriate, and retrieval augmented generation, for cases where organisations primarily need an AI system to work with their proprietary knowledge.

The technology should follow the use case, rather than forcing every customer through the same AI development path.

AI capability without an entire ML department.

Particularly organisations that have valuable proprietary data, a defined AI use case and limited internal ML infrastructure. NAPSC was not positioned as simply AI for everyone.

Start focused, then expand.

Rather than attempting to serve every industry immediately, the initial market strategy identified renewable energy as a potential entry vertical. This created an opportunity to explore industry-specific workflows and templates rather than expecting every customer to configure an AI system from scratch.

NAPSC is intended to make sophisticated AI development more accessible while still giving organisations visibility and control over how their data, models and evaluation processes are handled.

Another product challenge was how to expose sophisticated machine learning concepts to users who may not be ML specialists.

Model evaluation could include technical measures such as perplexity, but the product experience needed to translate model quality into information users could understand and act on. I treated this as a product and user experience problem rather than a case of presenting technical metrics without context.

NAPSC sits between an organisation's proprietary data and the underlying AI models, rather than being another interface on top of a single model.

ENACT AI is not simply another application built on top of a single AI model. The product opportunity is the workflow and infrastructure layer that helps organisations move from proprietary data and a business problem to a specialised AI system without having to build the entire machine learning stack internally.