Birdie.ai: GenAI to transform feedback into actionable insights

Google Cloud Results

96% accuracy of the models developed with Vertex AI and Gemini 1.5 Flash
80% time savings in the analysis and classification of data collected for one of the company’s clients
3.5x lower costs per processed text
Acceleration of daily processing times for millions of feedback entries: from 4 hours to 15 minutes
Possibility of scaling from zero to hundreds of machines in seconds to run ML models, enabled by Cloud Run

Thanks to Google Cloud’s tools, Birdie.ai managed to drive its computing power, improve end customer experience, and boost customer development.

Birdie.ai, based in Palo Alto, California (USA), is an organization specialized in customer feedback analytics that provides actionable insights to companies. Its mission is to make the voice of customers heard through a globally distributed team, driving strategic decisions that boost customer experience.

As the organization operates in the data and insights industry, it leverages inputs from several sources: from databases and customer service/satisfaction platforms to social media and review sites, among others.

“We use data to help companies identify trends, issues, and opportunities, enabling them to make decisions more quickly and accurately,” explains Everton Alvares Cherman, Co-founder and CTO of Birdie.ai.

In recent years, Birdie.ai has undergone some technical transformations to enhance the development of its solutions and customer services. In 2022, it chose Google Cloud as it is a pioneer in the AI and data industry.

Staged migration and tool integration

After deciding to migrate its operations to the public cloud, Birdie.ai took its first step towards transformation. At first, Google Cloud’s engineering and technical teams helped the company throughout the transition process, validating the ideal methods to make migration feasible and paving the way for the execution of PoCs.

The company’s journey kicked off in September 2022 and was completed in three months, mainly leveraging BigQuery features.

“Throughout this cycle, we migrated our Kubernetes clusters and related databases and data pipelines to BigQuery. Later, we began a pipeline based on microservices, and Pub/Sub became a key solution,” explains Birdie.ai’s Co-founder and CTO.

One of BigQuery’s main features is how easy it is to use, especially its computing power for analytics services. And in 2023, the company added another product: Cloud Run.

The most interesting feature of Cloud Run is that it helps us keep ML processes simpler, at scale, and with lower costs. We have also achieved flexibility gains.

Everton Alvares Cherman

Co-founder and CTO, Birdie.ai

Cloud Run enables the organization to scale from zero to hundreds of machines in only seconds—reducing costs by 3.5x per processed text—as well as to train and use SLMs with CPU efficiency. The company also gained time savings in the daily processing of millions of feedback entries: from 4 hours to 15 minutes.

In this scenario, another highlight relates to app responses. This statistic was improved by 83% based on RAG technology, a type of data retrieval process.

Innovating with Gemini to improve results

More recently, in February 2024, Birdie.ai started to use Gemini 1.5 Flash and Vertex AI. Right after migrating another API to Vertex AI, the company also began hosting open-source models, which are easily managed in the cloud.

Today, Birdie.ai’s solution is twofold: a data platform and an app. Although the platform deals with various complex aspects, such as third-party system integration, data ingestion, enrichment, indexing and storage into databases, and the availability of APIs that enable the app to leverage these simplified services, the application serves as the front-end solution for users.

GenAI aids in the data structuring and enrichment process of the platform, involving tasks such as chat summarization, text classification, and data extraction.

In the app, GenAI is applied to various tasks, such as summarizing collections of filtered feedback, uncovering opportunities among existing feedback entries, and “chatting” with data.

Thanks to Gemini 1.5 Flash, our models are now 9% more accurate, attaining a 96% accuracy rate. We’ve also reduced the processed unit costs by 15%, compared to the previous resources we used.

Everton Alvares Cherman

Co-founder and CTO, Birdie.ai

Technology in line with customer experience

Following these improvements leveraged by Birdie.ai’s solution, the customers it serves clearly get most of the benefits. For example, Mercado Bitcoin, a leading crypto platform, has revolutionized its Voice of Customer program. With 4 million users and transactions over USD 15 billion, the company can now convert tons of feedback into true growth.

Before adopting Birdie.ai, Mercado Bitcoin’s team worked on collecting and orchestrating insights and reports, instead of planning and working on initiatives related to product enhancements.

After implementing this CX intelligence platform and integrating it into the Voice of Customer workflow, the company can consolidate all the information and centralize it into a single source.

This means that teams can comprehensively access customer feedback and its impact, applying several defined metrics. Among other benefits, the company also leveraged aggregation features, automated analytics, and 80% acceleration in the assessment and classification of collected data.

Apart from enabling the implementation of projects to better serve Birdie.ai’s customers, migrating to the cloud allowed the company to simplify MLOps processes and enhance user experience, thanks to lower processing times.

After centralizing its services in Google Cloud, Birdie.ai intends to leverage all these positive impacts to train, assess, and strengthen MLOps processes using further automations, and ultimately continue with the flow of new trends.

More recently, we started working with audio transcriptions at scale. That’s why we expanded the use of GPUs, with low costs, large scale, and high availability. This is another major step that we will soon offer to all our customers.

Everton Alvares Cherman

Co-founder and CTO, Birdie.ai

Birdie.ai provides companies with a comprehensive view of top CX improvement opportunities, leveraging the processing of feedback received across multiple sources throughout customer journeys. As a result, organizations can make continuous improvements on several indicators, including satisfaction, engagement, customer service, and churn rates.

Industry: Technology

Location: United States

Products: Google Cloud, BigQuery, Cloud Run, Gemini 1.5 Flash, Google Kubernetes Engine, Pub/Sub, VertexAI