For the past few months, we’ve been working with a client who’s aiming to revolutionise the carbon certification industry with a tech-driven, AI-enabled solution geared towards efficiency, collaboration and user-friendliness.
Our client has a lot of knowledge about the market, connections with key stakeholders and several ideas about how the carbon credits certification process can improve. Currently, they are looking to raise a seed round to launch the MVP.
In our first interaction, we combined their expertise in the carbon credits market and our technical and product knowledge to build a proof of concept to harness the power of artificial intelligence in this domain.
Why do a Proof of Concept instead of building an MVP?
In this case, it would make sense to build a PoC instead of an MVP for four main reasons:
- Validation of technical feasibility: a PoC serves as a preliminary investigation into whether the AI technology can perform as expected in real-world conditions. It’s a low-risk way to validate the technical feasibility without the commitment of resources required for an MVP.
- Stakeholder buy-in: a successful PoC can help build confidence among stakeholders, including investors and potential users, by providing proof that the idea is possible and relevant. This can be crucial for securing the necessary support and resources to develop an MVP.
- Cost and time efficiency: PoCs have fewer requirements and need less polishing which will require less time and financial investment compared to developing an MVP. They help prevent the larger expenditures that come with developing features or products that may not meet market needs or technical expectations.
- Risk mitigation: By conducting a PoC, it’s possible to identify potential risks early on. This may include technical limitations, user acceptance issues, or unforeseen complexities with data integration. Addressing these issues at the PoC stage is less costly and less complex than during the MVP phase.
In summary, a PoC is a strategic step that allows for testing assumptions, understanding technical and market challenges, and refining the product concept in a controlled, risk-managed environment. It sets the stage for a more informed, efficient, and effective development of an MVP, ultimately increasing the chances of success for the final product.
The proof of concept in action
After having more clarity about the main issues of the carbon credit certification process, the main stakeholders and the opportunities around this market, we were able to define the goal for this PoC.
We decided to focus on the first stage of the process: the development of the Project design document. A stage where project developers spent a large amount of time gathering all the information needed. Our goal for this Proof of Concept was clear: to demonstrate that AI could significantly reduce the time project developers spend crafting Project Design Documents.
Our PoC was centred around two main features: AI-enhanced preliminary drafts and an AI-powered suggestion engine. With the AI-enhanced preliminary drafts, we wanted to validate how AI could generate a preliminary draft of the PDD by merging data from an initial survey with insights gathered from other PDDs.
This functionality had a high impact in the initial drafting phase of the PDD because it’s typically the most time-consuming one involving gathering and organising a large volume of information. On the other hand, there was also an automation potential because much of the information required is standard across PDDs, making it ripe for automation using AI.
The AI-powered suggestion engine had the goal of analysing content from other successful PDDs approved by a specific standard and using AI to suggest specific components of the PDD aligned with the standards of the industry. This way we could use the learning capabilities of AI to provide valuable insights that might be missed.
By targeting these two areas, we aimed to address both the efficiency and quality of the PDD creation process, which are critical for project developers.
Conclusions and feasibility analysis
Our proof of concept has demonstrated that AI can revolutionise the creation of project design documents. By feeding example documents to OpenAI’s GPT-4 model, we were able to automatically generate draft PDDs that follow a standardised structure and contain most of the standard data and specific data of the project.
The results of our feasibility analysis were very promising. We found that high-quality draft PDDs could be produced in just 30 seconds, drastically reducing the time spent. The drafts were consistent in their structure and scaled easily by parallelising requests to the AI model. At a cost of only €0.04 per draft, this approach has immense business potential.
In conclusion, leveraging AI for automated PDD generation is incredibly efficient, cost-effective, and reliable. Given the outstanding proof of concept results, we believe this technology could streamline and enhance the product by rapidly generating draft PDDs for each new carbon credit project. The data and insights of this PoC have laid a robust groundwork for the next phase by providing a compelling case that we believe will be instrumental in attracting the necessary investment for the MVP.
By leveraging innovations like AI generation, we can provide fast, tailored solutions to meet the specific needs of our clients. The success of this initial project shows one way we can deliver value, and we look forward to applying what we’ve learned to new use cases.
Right Talent, Right Time
Turnkey technical teams and solo specialists
when you need them for greatest impact.
