Overview
Indra Energia is developing a platform that, using generative AI with Amazon Bedrock, will allow millions of users to migrate to the Free Energy Market. In operation since 2019, Indra Energia is a company specializing in energy trading and has been supporting customers who want to migrate to the Free Energy Market. Indra Energia’s partner, Thiago Veiga, explains that by 2028, this possibility is expected to be open to all consumers – residential and corporate – which should greatly expand migration processes.
He explains that this process begins with an analysis of the customer’s electricity bill, which enables the company to indicate the best product according to that consumption profile. “We are asked by our customers to carry out this process, but we are talking about a market with the potential for millions of migrations,” he explains.
Opportunity | Standardizing energy bills
That’s why Indra thought about developing a platform that would allow interested customers to upload their bills and, based on an analysis, obtain automated cost estimates. “We initially tried to develop an in-house solution, but the analysis of non-standardized information became a challenge,” he explains, pointing out that, in addition to each utility having its own layout for the bills, they were received in different formats, such as color and B&W images, text, .pdf, etc.
In view of this, Indra’s data team came up with a solution that applies a generic extraction of data from the different layouts worked on and the use of Generative Artificial Intelligence to make sense of this data, making it possible to analyze it and automate responses to customers. “Since Indra was created, our processes have run on AWS, so we sought information from them to understand how they could help us with this project,” he says.
“It takes us a lot of time to do this process of data input and account analysis, which is also very prone to errors. When we’re up and running in the way we expect, with the support of AWS, this time will be much shorter.”
Thiago Veiga
Partner at Indra Energia
Why AWS | Identifying data and semantics
After initial contacts, AWS appointed Flexa Cloud to develop the project, which aims to extract information from different energy bills, standardize it and provide answers to users. Indra Energia’s data scientist, Ruan Medina, explains that the information extraction process was divided into two stages.
“In the first, we transform everything into text that can be read by the platform and, in the second, the data is identified and given semantics,” he explains, noting that the process allows us to identify more than 180 different document models, all with the same data but presented differently. “AI is fundamental for identifying this data and the semantics of this context,” he says.
The idea is to avoid Indra having to collect this data manually for each new client. “When we saw the possibility of having contact with Flexa, which was already working with Generative AI, and of inserting a tool into the process with the support of AWS, we found a way of technical development,” he says.
The CEO of Flexa Cloud, Deivid Bitti, explains that two layers of artificial intelligence are being used to develop Indra’s platform. The first, based on Amazon Textract, extracts data from documents in different formats. “Here, we’ve managed to create a process that extracts and converts the image into text and, in the end, returns a series of data with high assertiveness,” he explains, citing the examples of the customer’s personal data indicated on the bill, energy concessionaire, consumption history and contracted demand, bill details such as fines and charges, tax rates, group marking and tariff framework, consumer unit identifiers, among others.
The second layer relies on Amazon Bedrock to use generative Artificial Intelligence, a critical stage in the process. Generative AI makes it possible to identify and facilitate the semantic processing of information extracted from various processes on the platform. “Bedrock was chosen because it allows us to test a variety of foundational models quickly and at low cost. With Bedrock, we can test different foundational models until we identify the best one,” explains Bitti.
Anthropic’s Claude-2 foundational model was chosen after performance tests showed excellent results for Brazilian Portuguese. The context window of 100,000 tokens was critical as it allows us to work with all the data present in the electricity bill, which can be several pages long.
The architecture also relies on Amazon S3 to provide secure, temporary storage for the energy bills. Amazon Aurora in PostgreSQL stores all the platform’s data in a structured way. “We’re going to turn this system into an API where it will be possible to present the file and receive the data. It will be a 100% serverless solution, fully pay-as-you-go, following demand upwards or downwards,” he explains.
Results | Preparing the structure for the future
For Veiga, Indra’s use of Generative AI will represent a disruption in the company’s processes. “It takes us a lot of time to do this process of data input and account analysis, which is also very prone to errors. When we are up and running in the way we expect, with the support of AWS, this time will be much shorter,” he says.
The executive emphasizes that Indra is preparing for 2028, when the Free Energy Market will also be accessible to residential customers. “We’re talking about tens of millions of consumers. Only 2% of this market has more than 2 million customers who, to be won over, would require us to analyze around 10 million invoices. We have to be tremendously robust to meet this volume,” he says, noting that today it takes three to four days for the company to send a quote – which is already faster than many competitors in the market – and that by 2028 it will be done in seconds and via cell phone.
Indra’s CEO, Ingrid Catherine Santos, agrees. “The market is opening up and if we don’t prepare ourselves, we won’t be able to keep up. This project is not disruptive from the word go,” she says, stressing that the solution should bring benefits such as improving the customer experience, who will be able to count on high-quality interaction through a self-service platform.
In addition, the platform on AWS speeds up and makes scalable the acquisition of customers in volume, with few manual actions, and drastically reduces the time spent manually reviewing documents to collect data.
Next steps
Medina goes further, and says that Indra will have modules to meet every step in the journey of its customers and partners. “What we have here is the customer research, identification and acquisition module, but we’re still going to expand the frontend for these accesses, manage the benefits received by partners and implement various connections to internal and external data to analyze the data extracted from energy bills. There is a whole demand for development that will use this solution and we will have a lot to use from the AWS infrastructure in this process,” he predicts.
About Indra Energia
Indra Energia is a Brazilian energy company that since 2019 has been helping customers reduce costs, make utility bills more predictable and reduce their CO2 footprint.
If you’d like to discuss the Energy Market, please click the link below to send us an email.