Saves 15% in overall costs by running the cloud more efficiently
Reduces scaling time by 87% to improve service delivery
Accelerates incident investigation from days to real-time, enabling rapid identification and resolution of performance bottlenecks
Achieves 99.95% API success rate to improve real-time personalization of content delivery
Bobble AI, a conversational keyboard platform, migrated to Google Cloud to enhance scalability, improve performance, and strengthen data privacy, delivering a seamless user experience and fostering business engagement.
When Bobble AI launched its flagship Indic keyboard in 2020, it became a game changer for everyday conversations. Users now use the Bobble AI keyboard on their favorite messaging app or Android device to jazz up their conversations with customizable stickers, GIFs, and more. Businesses can also be part of those conversations through branded campaigns and contextual ads.
Timely delivery is a key factor to ensure user experience and campaign success. Like an awkward pause in real-life interactions, any drop in app performance such as keyboard lags or slow responses can frustrate users, leading to a loss of engagement. The frontend user interface of the Bobble AI keyboard needs to work in tandem with the backend for a seamless conversation experience.
In 2024, Bobble AI faced significant challenges in maintaining smooth and reliable performance as its infrastructure became increasingly complex. Running microservices within a VM led to competition for shared resources such as CPU, resulting in performance degradation. Additionally, manual scaling diverted developer time from core development to infrastructure tasks. As the team grew, effectively tracking interdependencies between microservices became more difficult, hindering efficient development and troubleshooting.
"By adopting an infrastructure-as-code approach on Google Cloud, we can optimize our cloud resources, streamline our operations, and accelerate time to market," said Kunal Dawn, Senior Vice President for Engineering at Bobble AI. "This enables us to introduce innovative AI features while maintaining a unified view of our API and insights workloads."
By adopting an infrastructure-as-code approach on Google Cloud, we can optimize our cloud resources, streamline our operations, and accelerate time to market. This enables us to introduce innovative AI features while maintaining a unified view of our API and insights workloads.
Kunal Dawn
Senior Vice President for Engineering, Bobble AI
Google Cloud Partner Niveus played a pivotal role in Bobble AI's successful migration to Google Cloud, with zero impact on latency and error rates. Its expertise in cloud migration strategies, coupled with its deep understanding of Bobble AI's specific requirements, enabled a smooth and efficient transition.
Prior to its migration to Google Cloud, Bobble AI hosted multiple microservices in individual VMs, and the failure of one often led to cascading failures within the entire VM. This had a significant impact on user experience, leading to delays, errors, and disruptions in services, such as content delivery and personalized recommendations.
At the same time, the distributed nature of microservices across multiple VMs presented additional challenges for developers. Identifying and resolving issues within this environment required significant effort, as developers had to navigate the complexities of debugging across different VMs and understanding the interdependencies between microservices.
Moreover, scaling microservices to handle sudden increase in traffic during holidays and campaigns was costly, when using VMs. Bobble AI had to allocate resources such as CPU and storage to the VM instead of optimizing individual microservices. This approach, while effective in addressing immediate performance concerns, ultimately contributed to higher cloud costs and wasted resources.
By migrating our microservices to Google Kubernetes Engine, we achieved significant improvements in scalability and performance. GKE's ability to independently scale microservices, coupled with its cost-effective pricing, enabled us to optimize our cloud resources and reduce overall cloud costs by 15%.
Kunal Dawn
Senior Vice President for Engineering, Bobble AI
This approach, while effective in addressing immediate performance concerns, ultimately contributed to higher cloud costs and wasted resources.
"By migrating our microservices to Google Kubernetes Engine, we achieved significant improvements in scalability and performance," said Dawn. "GKE's ability to independently scale microservices, coupled with its cost-effective pricing, enabled us to optimize our cloud resources and reduce overall cloud costs by 15%."
Furthermore, GKE's managed infrastructure and streamlined development workflows enabled developers to reduce scaling and infrastructure provisioning time by 87%. Bobble AI achieved a 99.95% API success rate by deploying its APIs on Google Cloud's robust and highly-available infrastructure. This allowed Bobble AI's developers to focus on innovation, accelerating development, and enhancing user engagement.
As Bobble AI grew in popularity, user activity grew significantly. In addition to bobblehead stickers and memes, users are also using features such as voice typing or Sports Widget to track their favorite games. During peak hours before and after work, Bobble AI experiences more than two to three times its average traffic.
To ensure real-time delivery for users around the world, Bobble AI uses Cloud Load Balancing to distribute traffic across multiple servers, preventing any single server from becoming overloaded. Cloud Load Balancing also helps to ensure high availability by routing traffic to backup servers in case of outages or cyberattacks.
For improved performance, Bobble AI brings content closer to its users by distributing its frequently accessed content and API responses on Cloud CDN, a global network of content delivery servers. This means users can access Bobble AI's entire library of more than one million stickers and GIFs, without using device space. Only stickers generated by the user are stored on their own device to protect user privacy.
Data is a cornerstone of Bobble AI's operation, driving both positive user experience and business success. The recommendation engine relies heavily on data to accurately analyze user input, understand intent, and deliver highly personalized content. For example, if a user types, "Do you want a coffee?," the app might suggest relevant stickers without spamming the user with ads related to coffee.
Beyond its role in recommendations, data also empowers product managers with actionable insights. By analyzing user behavior and preferences, Bobble AI can identify areas for improvement, uncover new business opportunities, and refine marketing campaigns.
To support these critical functions, Bobble AI stores 750 terabytes of data in Cloud Storage. This data, such as images and raw user-activity data, is essential for the recommendation engine and other data-driven initiatives.
For real-time data access, Bobble AI leverages Cloud SQL to retrieve and process user information in real-time, ensuring a seamless user experience. BigQuery helps Bobble AI extract insights from large datasets to analyze trends and patterns in user engagement or retention. To manage and orchestrate the API and insights, Bobble AI employs Dataproc with Cloud Composer for flexible data processing pipelines and batch jobs.
Finally, Cloud Pub/Sub optimizes performance by ensuring reliable and scalable messaging between different components of Bobble AI's architecture.
Adopting Prometheus has streamlined our microservice management. We can easily review and automate code without making changes to the UI. This shift has significantly improved code consistencies and enabled seamless automation, a major achievement for our team.
Kunal Dawn
Senior Vice President for Engineering, Bobble AI
Improving the microservices environment not only boosts app performance and insights but also the developer experience at Bobble AI. The shift from VM to containers and cloud-native tools has simplified development and reduced the risk of errors.
Developers can now review and modify microservices directly through code, ensuring code consistency and facilitating automation. By leveraging tools such as Managed Service for Prometheus for monitoring and Terraform for infrastructure management, Bobble AI significantly enhanced its operational efficiency and reduced manual overhead.
"Adopting Prometheus has streamlined our microservice management. We can easily review and automate code without making changes to the UI," said Dawn. "This shift has significantly improved code consistencies and enabled seamless automation, a major achievement for our team."
Additionally, Google Cloud Console has transformed Bobble AI's debugging process. Previously, developers took a full day to manually inspect individual VMs to identify microservice issues. New developers relied on knowledge transfer from more experienced developers to understand the app's architecture, components, and dependencies. Now, they can seamlessly stream and analyze logs from all microservices in real time, enabling rapid troubleshooting and resolution.
To ensure timely responses to incidents and anomalies, Bobble AI leverages Cloud Logging to establish customized alert parameters. This proactive approach enables developers to swiftly address issues, minimizing downtime and maintaining optimal app performance. By improving monitoring and alerting, Bobble AI can effectively identify underutilized cloud resources and make adjustments to reduce costs.
Generative AI is poised to revolutionize the way we converse and Bobble AI is at the forefront of the transformation. By integrating a third-party generative AI solution into its keyboard as a premium feature, Bobble AI empowers users to effortlessly convert voice into AI-generated text, even for those who struggle with typing. To deliver such personalized experiences, Bobble AI analyzes vast amounts of data to ensure that its AI-generated content and responses are highly relevant and engaging for users.
To safeguard user information, Bobble AI adheres to stringent privacy policies and employs robust encryption measures. All communications are end-to-end encrypted on Google Cloud, ensuring that data remains confidential and protected from unauthorized access. Additionally, privacy data is stored in an anonymous format, further safeguarding user information and mitigating the risks associated with potential data breaches.
Generative AI empowers users to express themselves more effectively, fostering deeper connections. This opens new avenues for businesses to engage with their customers on a more personalized level. By leveraging Google Cloud, we're well-positioned to handle the increasing demands of AI-driven applications while respecting user privacy, ensuring exceptional user experiences.
Kunal Dawn
Senior Vice President for Engineering, Bobble AI
"Generative AI empowers users to express themselves more effectively, fostering deeper connections. This opens new avenues for businesses to engage with their customers on a more personalized level," said Dawn. "By leveraging Google Cloud, we're well-positioned to handle the increasing demands of AI-driven applications while respecting user privacy, ensuring exceptional user experiences."
Founded by a dynamic duo of brothers, Bobble AI is the world's first Conversation Media Platform.
They are on the mission of enriching everyday conversations by empowering expressions for users with an amazing suite of Keyboard applications.
With expressive and personalized content including stickers, GIFs and emojis, deep localization with over 100 languages, AI-based contextual recommendations, speech-to-text, and much more, Bobble AI is transforming the conversations of over 50 million users and counting.
Bobble AI's flagship product Bobble Indic Keyboard allows real-time content creation and personalization through its leading-edge AI technology. It is the highest-rated, most engaging, and retaining keyboard in the world.
Industry: Media & Entertainment
Location: India
Products: BigQuery, Cloud CDN, Cloud Composer, Cloud Load Balancing, Cloud Logging, Cloud Monitoring, Cloud SQL, Cloud Storage, Pub/Sub, Dataproc, Google Kubernetes Engine, Managed Service for Prometheus
About Google Cloud partner- Niveus Solutions
Niveus Solutions is an award-winning Google Cloud partner that works with customers to solve complex business problems.