July 2026 Summaries

In a conversation with Austin Hay, an operating partner at Khosla Ventures, the focus was on how founders can leverage AI to enhance their go-to-market (GTM) strategies by prioritizing the automation of mundane, administrative tasks rather than attempting to create an all-encompassing growth machine from the start. Austin emphasizes that AI excels in automating tasks that are process-driven and non-probabilistic, such as managing expenses and cleaning lists, but struggles with generating high-quality content, which should involve human creativity and refinement. He advocates for "skillifying" tasks—breaking them down into discrete, manageable skills—before automating them, and stresses the importance of maintaining human oversight in any automated process to avoid errors that could harm customer relationships. Austin also highlights the distinction between business reporting, which should be standardized, and directional reporting, which allows for flexibility and speed, while cautioning against the misconception that GTM processes can be entirely automated without human intervention. The conversation underlines the need for founders and teams to focus on essential tasks, gradually automate repetitive processes, and develop their tool stack organically based on practical needs.

Knock recently rebuilt its template editor, a crucial component for customers to design message templates with personalized content using Liquid, a templating language. The new editor, based on Tiptap, mimics document editors like Notion, allowing users to insert dynamic values effortlessly. A significant challenge was developing autocomplete functionality for Liquid variables, particularly within loops, which are complex because they lack a defined schema. Unlike the previous CodeMirror-based editor that parsed Liquid into a tree structure, Tiptap relies on a document model that does not inherently understand Liquid's structure. This necessitated a new approach: flattening the document into a Liquid string for parsing, which provides loop scope and variable information. This method efficiently manages autocomplete suggestions, even when documents are not fully formed, by utilizing regex scans to handle incomplete Liquid syntax. The new system successfully replicates the old editor's behavior and accommodates the typical incomplete states of documents during typing, enhancing user experience while maintaining reliability.

In March 2026, Knock introduced the Knock Agent, an AI tool designed to manage customer messaging resources by creating workflows, templates, and user segments, and assessing messaging performance. This agent operates through the Knock dashboard, Slack, the API, or MCP server, and was developed using bash, a virtual file system, and Knock's management API. The development team aimed for a scalable solution, ultimately building a virtual bash environment in Elixir, which allowed the agent to utilize a filesystem instead of multiple tools to access and modify data. For observability, they implemented OpenTelemetry traces and a homegrown evaluation system. The system supports learning new skills, including customer-defined ones, by accessing information via a CLI and a debugging skill, allowing it to adapt to various resource types beyond its initial focus on workflows and templates. The agent is intended to be efficient, avoiding performance costs associated with a full sandbox and instead utilizing a virtual, in-memory option. Future plans include refining session summarization, expanding evaluation cases, and exploring subagents for specialized tasks, ensuring the agent remains adaptable and responsive to evolving customer needs.