November 2025 Summaries

Pipedrive, a CRM platform, incorporates AI extensively in its engineering processes, as shared by CTO Agur Jõgi. The company has implemented an innovative AI-driven code review system that complements human feedback, fostering a peer-learning environment where juniors write code more swiftly, and seniors refine their coding approaches through AI-generated suggestions. Additionally, Pipedrive's Site Reliability Engineers (SREs) utilize machine learning models trained on historical data to preemptively address potential system outages, enhancing system resilience. The company prefers building AI tools in-house to deepen technical understanding, though it uses third-party models for general tasks, balancing speed and knowledge acquisition. Pipedrive has developed a traffic-light framework to ensure compliance with regulations like GDPR while experimenting with AI, promoting a proactive learning culture within its teams.

Susana de Sousa, who leads Community at Plain, discusses the evolving landscape of customer support, emphasizing a shift from traditional methods towards a more proactive, technical role driven by AI advancements. She highlights the pressures faced by support teams, both from job security concerns due to rapid technological changes and from executive demands for leveraging new tools to enhance customer success. Susana advocates for a focus on foundational elements like documentation and product quality before integrating AI, as it can amplify existing issues. She also stresses the importance of support teams collaborating closely with engineering, suggesting that support roles will become increasingly technical as AI handles routine tasks. Susana encourages support leaders to develop technical skills and systems thinking to effectively address complex issues and adapt to the evolving demands of the industry, predicting that by 2026, support engineers will play a pivotal role in modern customer support.

Jam has introduced six updates to enhance user experience and streamline bug capturing. These updates include MacOS shortcuts that allow users to request Jams with a simple command, and the ability to set custom domains for recording links through dashboard settings. Users can also filter their Jams by various criteria in the dashboard, and enjoy a redesigned Jam app for iOS that facilitates creating bug tickets on mobile devices. The addition of Linear embeds lets engineers view embedded videos with network and console errors directly in tickets or comments. Furthermore, collapsible folders help maintain an organized workspace by allowing users to collapse folders and view the number of Jams contained within each. These enhancements aim to simplify the bug capturing process, making it more efficient and user-friendly.

Built-in privacy for customer support Jams has been launched, providing automatic blurring of sensitive information on customers' screens during recordings, such as credit card details and passwords, without requiring any additional action from the users. This feature is compatible with various customer support tools, including Intercom, and simplifies the process of viewing customer screens by eliminating unnecessary details. Customers can easily add this auto-blur feature to their Jam plan and explore its functionality through demos and documentation. The initiative enhances the experience of capturing and addressing bugs efficiently, offering a format that is popular among developers, and is available for free.

Jam for Customer Support is a tool designed to streamline the process of triaging customer-reported bugs by integrating with various helpdesk platforms like Zendesk and Freshdesk. It allows support teams to provide customers with a Jam link in their support chat, enabling customers to record their screens without installing any additional software. The process involves copying the link into the chat, naming it with the ticket link for easy reference, and receiving detailed debug information, including metadata, developer logs, and video, which is then packaged for developers. This tool simplifies bug reporting and enhances communication between support teams and customers, offering a convenient and efficient solution for bug tracking and resolution.

Sunny Ellis, who leads the support team at Givebutter, a fundraising platform, has successfully integrated AI to manage 60% of inbound support requests, allowing the team to focus on more complex and emotionally nuanced issues. The support team, which has grown from five to thirty members, uses Intercom's Fin for automating responses to repetitive questions, enhancing response times and freeing up human agents for more critical tasks. Before implementing AI, the team revamped their help center documentation to ensure its accuracy and completeness, which proved pivotal for effective automation. Sunny emphasizes hiring for empathy and critical thinking, as AI handles routine inquiries while humans manage connections and complex problems. Additionally, she uses Claude to develop internal tools that assist support agents with centralized information and response drafting. Her advice to support teams exploring AI includes starting with clean documentation, launching small, and continuously refining the process, underscoring that AI integration is an ongoing journey rather than a one-time setup.

Taskrabbit, a marketplace facilitating over 1.6 million home tasks annually since 2008, has evolved into a complex system influenced by thousands of engineers. In a discussion with CTO Scott Porad, the focus was on the impact of AI on mature engineering teams, emphasizing that future software development will be driven more by human confidence in deploying AI-written code than by the volume of code AI can produce. Porad notes that AI is shifting engineers' roles from writing to reviewing code, highlighting the importance of QA skills and code review as essential competencies. He advocates for a new apprenticeship model where junior developers learn by analyzing AI-generated code with experienced engineers, fostering a culture of encouragement rather than enforcement for AI adoption. Porad underscores the potential of AI in enhancing testing processes to reduce risks in complex systems and sees AI's role as an assistant rather than an autonomous developer, especially in reviewing code. He introduces "satellite apps" as a low-risk area for experimenting with AI, and emphasizes the need for cultural adaptation over mere technical implementation, envisioning a future where engineers design, validate, and refine AI-driven systems.