The context & the Client’s case

RegenX operates in decarbonization financing, a sector where credibility, data integrity, and speed of execution all matter. The need was clear: deliver a photo and text based AI assistant for agronomists and farmers, designed for remote connectivity, low-tech end users and in a matter of weeks, not months.

What we were hired to do

We designed and delivered the solution from the ground-up, from the data architecture to the AI orchestration engine. By coordinating with a team of agronomists, we made the concept of “an agronomist in every farmer’s pocket” a reality.

Agronomists and farmers simply take photos of their farm, crop and fertilizer bag. In the backend, our solution identified the farmer, the crop and the most appropriate fertilizer practice in the upcoming season to provide personalized agronomic advice to each farmer.

The key constraints

The economics were the primary constraint. Standard quotes for functional MVP mobile development in Vietnam start around $30,000, with most vendors requiring at least 3 months to deliver. That’s before factoring in the back-and-forth, the delays, and the gap between “delivered” and “production-ready.”

The client had weeks, not months and a tight startup budget.

The secondary constraint was quality and AI alignment. In the context of decarbonization financing, we had to make sure the personalized recommendations provided to each farmer nudged them towards regenerative practice, which tend to lower carbon emission and unlock carbon financing. Nudging the other way would have been catastrophic.

What we did

We combined our business acunment, deep experience in AI and SDLC (software development lifecycle) expertise to structure the entire development approach around AI but also build a production-grade AI orchestration pipeline for the end client.

Technically, our AI solution had to:

• classify and detect intent from simple inputs (photos or just a few words from the farmer)
• augment simple prompts from low-tech users to derive context
• align recommendations on regenerative and low-carbon practices, without discarding more traditional farming practices
• combine with external data sources about the farm to personalize the recommendations (weather, crop being planted, farm size, etc)
• nudge farmers to track their own fertilizer usage to derive carbon emissions for the buyer, but also optimize their yields

Outcome

Cost came in under $2,500. Delivery took 3 weeks. The client launched immediately. In comparison with traditional dev shops, we built a production-grade product at 10x less the cost and 4x faster.

Within the first 60 days, 70% of their target users had adopted the app. This reflected a combination of product fit, a clean UX, and the fact that launching early created a better adoption window than a delayed release would have.

Could this apply to your organization?

If your organization is already outsourcing some work but is frustrated with cost or delivery time, you can use AI to compress both. However, you will certainly need support in delivery production-ready and reliable products, not just prototypes.

What matters is not just using AI, but using it correctly:

• keeping humans in the loop for architecture, quality, and alignment
• structuring development so AI accelerates, rather than introduces risk
• designing for real users from day one, not retrofitting later

If you are evaluating whether this applies to your case, we can assess it quickly based on your constraints, workflow, and target outcomes.

This model works particularly well when:

• Speed to market directly impacts business outcomes (revenue, adoption, capital deployment)
• The workflow is well understood, even if the technical solution is not yet built
• You need production-grade quality, not a demo or experiment
• Internal teams lack the bandwidth or expertise to execute quickly