Over the past couple of years, PhonePe’s journey toward operationalizing AI began with a search for external platforms capable of automating complex, cognitive business processes. However, as we tested various AI tools for our engineering and operational requirements, we consistently encountered friction, eventually leading us to architect our own internal application layer: Agent Hub.
\n\n\n\nOur initial objective was to use AI to achieve efficiency on internal processes. However, off-the-shelf products repeatedly imposed hard constraints. We frequently hit request limits in standard AI studios and strict document caps while indexing our internal knowledge bases. For our technical teams who need specialized integrations (for example: 1TB+ GitLab repositories), the expenditure and effort involved proved insurmountable due to multiple factors, both technical and non-technical.
\n\n\n\nTechnical issues we faced included context window overflow, slow processing/ storage limitations and so on.
\n\n\n\nGiven the domain PhonePe works in, a large part of the organization is governed by strict regulatory frameworks that not only govern data sharing and hosting, but also require us to scrape sites like the RBI and SEBI for up-to-date information. Standard web scrapers failed to provide the high degree of accuracy and reliability these workflows demanded.
\n\n\n\nAdditionally, many of our operations required human-in-the-loop interventions, forcing teams to toggle between their existing operational consoles. Also, external AI vendor environments would have introduced unacceptable change management friction.
\n\n\n\nWe built Agent Hub to provide a highly specialized, secure, and integrated platform tailored to our unique engineering needs. By focusing on deep infrastructure capabilities, we have created an environment that powers complex operational workflows at scale:
\n\n\n\nThese engineering choices enable a flexible, mixed-model stack. By avoiding vendor lock-in, we process routine document tasks cost-effectively via open-source models hosted on our own infrastructure, while reserving paid frontier models for tasks demanding complex reasoning.
\n\n\n\nBy focusing our engineering efforts on owning the workflows, integrations, and application layer rather than just the underlying models and data, Agent Hub has profoundly impacted operations across our business units. Today, the platform runs over 200 internal agents that handle more than 1,600 daily queries for 759 Weekly Active Users, saving an average of 5 minutes per automated task.
\n\n\n\nThe depth of these automations has transformed our turnaround times:
\n\n\n\nThanks to Agent Hub, our internal teams have been able to overcome constraints of external tools to create a secure, highly integrated platform that efficiently streamlines workflows and slashes costs and turnaround times across the company.
\n","date":"07 July, 2026","uri":"/engineering-agent-hub-how-phonepe-built-a-production-grade-internal-ai-platform/","categories":{"nodes":[{"name":"AI","slug":"ai"}]},"author":{"node":{"name":"Muktha Tavane","slug":"mukthatavane","description":"Muktha is our Corporate Communications wordsmith with over 7 years of experience in the fintech domain, complemented by additional expertise in retail and entertainment content industries. She crafts in-depth articles, data-driven insights, and strategises content campaigns. She’s also closely involved in PR and manages key media partnerships for PhonePe and its group companies."}},"featuredImage":{"node":{"altText":"","localFile":{"childImageSharp":{"fluid":{"base64":"data:image/png;base64,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","aspectRatio":1.5873015873015872,"src":"/static/42471d19da744bc64c7bb6fc06442382/bc8e0/Agenthub-scaled.png","srcSet":"/static/42471d19da744bc64c7bb6fc06442382/8ac63/Agenthub-scaled.png 200w,\n/static/42471d19da744bc64c7bb6fc06442382/3891b/Agenthub-scaled.png 400w,\n/static/42471d19da744bc64c7bb6fc06442382/bc8e0/Agenthub-scaled.png 800w,\n/static/42471d19da744bc64c7bb6fc06442382/6050d/Agenthub-scaled.png 1200w,\n/static/42471d19da744bc64c7bb6fc06442382/00405/Agenthub-scaled.png 1600w,\n/static/42471d19da744bc64c7bb6fc06442382/a76e0/Agenthub-scaled.png 2560w","sizes":"(max-width: 800px) 100vw, 800px"}}}}},"subtitleAcf":{"text":"PhonePe Team"}},"moreCategoryBlogs":{"nodes":[{"id":"cG9zdDoxMTMwOA==","excerpt":"Over the past couple of years, PhonePe’s journey toward operationalizing AI began with a search for external platforms capable of automating complex, cognitive business processes. However, as we tested various AI tools for our engineering and operational requirements, we consistently encountered friction, eventually leading us to architect our own internal application layer: Agent Hub. Navigating… Continue reading Engineering Agent Hub: How PhonePe Built a Production-Grade Internal AI Platform
\n","uri":"/engineering-agent-hub-how-phonepe-built-a-production-grade-internal-ai-platform/","date":"July 07, 2026","title":"Engineering Agent Hub: How PhonePe Built a Production-Grade Internal AI Platform","categories":{"nodes":[{"slug":"ai","id":"dGVybTo0MjE2","name":"AI"}]},"author":{"node":{"id":"dXNlcjoyMg==","name":"Muktha Tavane"}},"featuredImage":{"node":{"altText":"","localFile":{"childImageSharp":{"fluid":{"base64":"data:image/png;base64,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","aspectRatio":1.5873015873015872,"src":"/static/42471d19da744bc64c7bb6fc06442382/bc8e0/Agenthub-scaled.png","srcSet":"/static/42471d19da744bc64c7bb6fc06442382/8ac63/Agenthub-scaled.png 200w,\n/static/42471d19da744bc64c7bb6fc06442382/3891b/Agenthub-scaled.png 400w,\n/static/42471d19da744bc64c7bb6fc06442382/bc8e0/Agenthub-scaled.png 800w,\n/static/42471d19da744bc64c7bb6fc06442382/6050d/Agenthub-scaled.png 1200w,\n/static/42471d19da744bc64c7bb6fc06442382/00405/Agenthub-scaled.png 1600w,\n/static/42471d19da744bc64c7bb6fc06442382/a76e0/Agenthub-scaled.png 2560w","sizes":"(max-width: 800px) 100vw, 800px"}}}}},"subtitleAcf":{"text":"PhonePe Team"}}]},"moreBlogs":{"nodes":[{"id":"cG9zdDoxMTMwOA==","excerpt":"Over the past couple of years, PhonePe’s journey toward operationalizing AI began with a search for external platforms capable of automating complex, cognitive business processes. However, as we tested various AI tools for our engineering and operational requirements, we consistently encountered friction, eventually leading us to architect our own internal application layer: Agent Hub. Navigating… Continue reading Engineering Agent Hub: How PhonePe Built a Production-Grade Internal AI Platform
\n","uri":"/engineering-agent-hub-how-phonepe-built-a-production-grade-internal-ai-platform/","date":"July 07, 2026","title":"Engineering Agent Hub: How PhonePe Built a Production-Grade Internal AI Platform","categories":{"nodes":[{"slug":"ai","id":"dGVybTo0MjE2","name":"AI"}]},"author":{"node":{"id":"dXNlcjoyMg==","name":"Muktha Tavane"}},"featuredImage":{"node":{"altText":"","localFile":{"childImageSharp":{"fluid":{"base64":"data:image/png;base64,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","aspectRatio":1.5873015873015872,"src":"/static/42471d19da744bc64c7bb6fc06442382/bc8e0/Agenthub-scaled.png","srcSet":"/static/42471d19da744bc64c7bb6fc06442382/8ac63/Agenthub-scaled.png 200w,\n/static/42471d19da744bc64c7bb6fc06442382/3891b/Agenthub-scaled.png 400w,\n/static/42471d19da744bc64c7bb6fc06442382/bc8e0/Agenthub-scaled.png 800w,\n/static/42471d19da744bc64c7bb6fc06442382/6050d/Agenthub-scaled.png 1200w,\n/static/42471d19da744bc64c7bb6fc06442382/00405/Agenthub-scaled.png 1600w,\n/static/42471d19da744bc64c7bb6fc06442382/a76e0/Agenthub-scaled.png 2560w","sizes":"(max-width: 800px) 100vw, 800px"}}}}},"subtitleAcf":{"text":"PhonePe Team"}},{"id":"cG9zdDoxMDg4Mg==","excerpt":"When processing billions of transactions in a system as large as PhonePe, underlying databases go through constant querying, repairs and management. While a set of these tasks are automated, some are managed by human users. This leaves a potential risk of unauthorised/untracked human activities being executed in the databases. PhonePe addresses this risk with a… Continue reading Hoshō: A Database Activity Monitoring Tool
\n","uri":"/hosho-a-database-activity-monitoring-tool/","date":"May 15, 2026","title":"Hoshō: A Database Activity Monitoring Tool","categories":{"nodes":[{"slug":"engineering","id":"dGVybTo3","name":"Engineering"}]},"author":{"node":{"id":"dXNlcjoyMg==","name":"Muktha Tavane"}},"featuredImage":{"node":{"altText":"","localFile":{"childImageSharp":{"fluid":{"base64":"data:image/png;base64,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","aspectRatio":1.5873015873015872,"src":"/static/fd5350b7061977085d7c523b6f5a309f/bc8e0/Hosho-Final-Header-Icon-scaled.png","srcSet":"/static/fd5350b7061977085d7c523b6f5a309f/8ac63/Hosho-Final-Header-Icon-scaled.png 200w,\n/static/fd5350b7061977085d7c523b6f5a309f/3891b/Hosho-Final-Header-Icon-scaled.png 400w,\n/static/fd5350b7061977085d7c523b6f5a309f/bc8e0/Hosho-Final-Header-Icon-scaled.png 800w,\n/static/fd5350b7061977085d7c523b6f5a309f/6050d/Hosho-Final-Header-Icon-scaled.png 1200w,\n/static/fd5350b7061977085d7c523b6f5a309f/00405/Hosho-Final-Header-Icon-scaled.png 1600w,\n/static/fd5350b7061977085d7c523b6f5a309f/a76e0/Hosho-Final-Header-Icon-scaled.png 2560w","sizes":"(max-width: 800px) 100vw, 800px"}}}}},"subtitleAcf":{"text":"Pratik Anurag, Manisha C, Shubham Sharma, and Krishna Prasanth"}},{"id":"cG9zdDoxMDczOA==","excerpt":"Infrastructure as Code (IaC) is often treated as the holy grail of Site Reliability Engineering (SRE). We use tools like SaltStack and Ansible, alongside golden images and post-install scripts, to define our ideal state. But here is the hidden secret of infrastructure engineering: IaC dictates how things should be, but it doesn’t guarantee what actually… Continue reading Beyond IaC: Engineering a solution to Guarantee What Actually Is in Production
\n","uri":"/beyond-iac-engineering-a-solution-to-guarantee-what-actually-is-in-production/","date":"April 17, 2026","title":"Beyond IaC: Engineering a solution to Guarantee What Actually Is in Production","categories":{"nodes":[{"slug":"engineering","id":"dGVybTo3","name":"Engineering"}]},"author":{"node":{"id":"dXNlcjoyMg==","name":"Muktha Tavane"}},"featuredImage":{"node":{"altText":"","localFile":{"childImageSharp":{"fluid":{"base64":"data:image/png;base64,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","aspectRatio":1.5873015873015872,"src":"/static/63436876543243cc7eb59bd61c85f7c1/bc8e0/Beyond-IaC-HEADER-scaled.png","srcSet":"/static/63436876543243cc7eb59bd61c85f7c1/8ac63/Beyond-IaC-HEADER-scaled.png 200w,\n/static/63436876543243cc7eb59bd61c85f7c1/3891b/Beyond-IaC-HEADER-scaled.png 400w,\n/static/63436876543243cc7eb59bd61c85f7c1/bc8e0/Beyond-IaC-HEADER-scaled.png 800w,\n/static/63436876543243cc7eb59bd61c85f7c1/6050d/Beyond-IaC-HEADER-scaled.png 1200w,\n/static/63436876543243cc7eb59bd61c85f7c1/00405/Beyond-IaC-HEADER-scaled.png 1600w,\n/static/63436876543243cc7eb59bd61c85f7c1/a76e0/Beyond-IaC-HEADER-scaled.png 2560w","sizes":"(max-width: 800px) 100vw, 800px"}}}}},"subtitleAcf":{"text":"Aniruddha Thombre, Udeshya Giri, Shivam Pansy Satyendra Tyagi, and Ghanshyam Gupta"}},{"id":"cG9zdDoxMDY5MA==","excerpt":"At PhonePe Limited (“PhonePe”), the top banner on our platform’s homepage (the “masthead”) isn’t just an ad slot—it’s the important real estate we have. However, delivering real-time, personalized recommendations at a scale of hundreds of millions of users presents a complex problem to solve for two reasons: Whether a user is brand new or a… Continue reading Optimizing Merchandising: Machine Learning for Real-Time App Experience and LTV Maximization
\n","uri":"/optimizing-merchandising-machine-learning-for-real-time-app-experience-and-ltv-maximization/","date":"February 16, 2026","title":"Optimizing Merchandising: Machine Learning for Real-Time App Experience and LTV Maximization","categories":{"nodes":[{"slug":"engineering","id":"dGVybTo3","name":"Engineering"}]},"author":{"node":{"id":"dXNlcjoyMg==","name":"Muktha Tavane"}},"featuredImage":{"node":{"altText":"","localFile":{"childImageSharp":{"fluid":{"base64":"data:image/png;base64,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","aspectRatio":1.5873015873015872,"src":"/static/53b94435b8d140eec0522e246b3cec75/bc8e0/Machine-Learning-System-FINAL-1-scaled.png","srcSet":"/static/53b94435b8d140eec0522e246b3cec75/8ac63/Machine-Learning-System-FINAL-1-scaled.png 200w,\n/static/53b94435b8d140eec0522e246b3cec75/3891b/Machine-Learning-System-FINAL-1-scaled.png 400w,\n/static/53b94435b8d140eec0522e246b3cec75/bc8e0/Machine-Learning-System-FINAL-1-scaled.png 800w,\n/static/53b94435b8d140eec0522e246b3cec75/6050d/Machine-Learning-System-FINAL-1-scaled.png 1200w,\n/static/53b94435b8d140eec0522e246b3cec75/00405/Machine-Learning-System-FINAL-1-scaled.png 1600w,\n/static/53b94435b8d140eec0522e246b3cec75/a76e0/Machine-Learning-System-FINAL-1-scaled.png 2560w","sizes":"(max-width: 800px) 100vw, 800px"}}}}},"subtitleAcf":{"text":"Rohit Walimbe, Pramod Patil, Kunal Bhadra, Nrutya Doshi, Kedar Swadi"}},{"id":"cG9zdDo5NzM2","excerpt":"At PhonePe’s scale, handling millions of transactions and user interactions every hour requires an analytics platform that’s not just fast, but instant. When every millisecond counts — both for user experience and incident detection — batch processing simply doesn’t cut it. Enter Garfield, PhonePe’s custom-built, high-throughput event analytics platform, engineered around the power of Apache Druid. Garfield is the… Continue reading Powering Real-Time Decisions with PhonePe’s Garfield Platform and Apache Druid
\n","uri":"/powering-real-time-decisions-with-phonepes-garfield-platform-and-apache-druid/","date":"November 04, 2025","title":"Powering Real-Time Decisions with PhonePe’s Garfield Platform and Apache Druid","categories":{"nodes":[{"slug":"engineering","id":"dGVybTo3","name":"Engineering"}]},"author":{"node":{"id":"dXNlcjoyMg==","name":"Muktha Tavane"}},"featuredImage":{"node":{"altText":"","localFile":{"childImageSharp":{"fluid":{"base64":"data:image/png;base64,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","aspectRatio":1.5873015873015872,"src":"/static/0a2eb436f64553e845347bea12e6331a/bc8e0/Garfield-Tech-Blog-Icon-02-scaled.png","srcSet":"/static/0a2eb436f64553e845347bea12e6331a/8ac63/Garfield-Tech-Blog-Icon-02-scaled.png 200w,\n/static/0a2eb436f64553e845347bea12e6331a/3891b/Garfield-Tech-Blog-Icon-02-scaled.png 400w,\n/static/0a2eb436f64553e845347bea12e6331a/bc8e0/Garfield-Tech-Blog-Icon-02-scaled.png 800w,\n/static/0a2eb436f64553e845347bea12e6331a/6050d/Garfield-Tech-Blog-Icon-02-scaled.png 1200w,\n/static/0a2eb436f64553e845347bea12e6331a/00405/Garfield-Tech-Blog-Icon-02-scaled.png 1600w,\n/static/0a2eb436f64553e845347bea12e6331a/a76e0/Garfield-Tech-Blog-Icon-02-scaled.png 2560w","sizes":"(max-width: 800px) 100vw, 800px"}}}}},"subtitleAcf":{"text":"Nitish Goyal"}}]},"previous":null,"next":{"uri":"/hosho-a-database-activity-monitoring-tool/","title":"Hoshō: A Database Activity Monitoring Tool"},"categoriesData":{"nodes":[{"id":"dGVybTo0MjE2","name":"AI","count":1,"uri":"/category/ai/"},{"id":"dGVybToxOQ==","name":"Architecture","count":15,"uri":"/category/architecture/"},{"id":"dGVybTo0NDI=","name":"Cricket","count":null,"uri":"/category/sports/cricket/"},{"id":"dGVybTo3","name":"Engineering","count":15,"uri":"/category/engineering/"},{"id":"dGVybToyNQ==","name":"Mobile","count":10,"uri":"/category/mobile/"},{"id":"dGVybToyMg==","name":"Open Source","count":null,"uri":"/category/open-source/"},{"id":"dGVybToxNg==","name":"People Stories","count":3,"uri":"/category/people-stories/"},{"id":"dGVybToyOA==","name":"Product","count":3,"uri":"/category/product/"},{"id":"dGVybTo0MzM=","name":"Sports","count":null,"uri":"/category/sports/"},{"id":"dGVybTozMQ==","name":"Tech Stories","count":null,"uri":"/category/tech-stories/"},{"id":"dGVybTox","name":"Uncategorized","count":null,"uri":"/category/uncategorized/"}]},"recentPosts":{"nodes":[{"id":"cG9zdDoxMTMwOA==","title":"Engineering Agent Hub: How PhonePe Built a Production-Grade Internal AI Platform","uri":"/engineering-agent-hub-how-phonepe-built-a-production-grade-internal-ai-platform/"},{"id":"cG9zdDoxMDg4Mg==","title":"Hoshō: A Database Activity Monitoring Tool","uri":"/hosho-a-database-activity-monitoring-tool/"},{"id":"cG9zdDoxMDczOA==","title":"Beyond IaC: Engineering a solution to Guarantee What Actually Is in Production","uri":"/beyond-iac-engineering-a-solution-to-guarantee-what-actually-is-in-production/"},{"id":"cG9zdDoxMDY5MA==","title":"Optimizing Merchandising: Machine Learning for Real-Time App Experience and LTV Maximization","uri":"/optimizing-merchandising-machine-learning-for-real-time-app-experience-and-ltv-maximization/"},{"id":"cG9zdDo5NzM2","title":"Powering Real-Time Decisions with PhonePe’s Garfield Platform and Apache Druid","uri":"/powering-real-time-decisions-with-phonepes-garfield-platform-and-apache-druid/"}]}},"pageContext":{"id":"cG9zdDoxMTMwOA==","categories":[{"id":"dGVybTo0MjE2","name":"AI","slug":"ai","uri":"/category/ai/","count":1},{"id":"dGVybToxOQ==","name":"Architecture","slug":"architecture","uri":"/category/architecture/","count":15},{"id":"dGVybTo3","name":"Engineering","slug":"engineering","uri":"/category/engineering/","count":15},{"id":"dGVybToyNQ==","name":"Mobile","slug":"mobile","uri":"/category/mobile/","count":10},{"id":"dGVybToxNg==","name":"People Stories","slug":"people-stories","uri":"/category/people-stories/","count":3},{"id":"dGVybToyOA==","name":"Product","slug":"product","uri":"/category/product/","count":3}],"categoryId":"dGVybTo0MjE2","previousPostId":null,"nextPostId":"cG9zdDoxMDg4Mg=="}}, "staticQueryHashes": ["4132566406"]}