Louis Knight-Webb is a software engineer and founder who co-created Vibe Kanban, an open-source workspace for coordinating AI coding agents, and previously co-founded Who Targets Me, a political-advertising transparency project. His central concern is how people retain oversight when automated systems increasingly determine what they see, how software gets built, and which changes reach production.
Before Britain’s 2017 general election, Knight-Webb and Sam Jeffers created Who Targets Me, browser software that helped voters monitor political advertisements on Facebook. In 2019, Knight-Webb submitted evidence to the UK Parliament arguing that restrictions on advertising data obstructed independent scrutiny of political campaigns.
In 2021, he and Gabriel Gordon-Hall established bloop. Their open-source code-search platform combined natural-language queries, semantic search, code navigation, and AI-assisted changes. Knight-Webb also founded the London chapter of AI Tinkerers, holding early gatherings at bloop’s office before handing hosting responsibilities to Luke Harries.
His independent lou-eval benchmark addressed long-context model evaluation by testing whether models retrieve relevant facts scattered through lengthy inputs while avoiding misleading near-matches.
Knight-Webb and collaborators launched Vibe Kanban in 2025, giving developers isolated agent workspaces, code-diff reviews, inline feedback, previews, and pull-request workflows. Vibe Kanban Cloud, introduced in February 2026, added shared projects, issue assignment, and synchronized team workflows. As concurrent agent output strained SQLite, Knight-Webb moved execution logs into per-process files while retaining the database for structured application data.
His defining ideas
- Planning and review become the engineering work. Coding agents shift human effort toward defining requirements, checking generated changes, and deciding what can safely ship.
- Supervision should fit the task. Backend features, migrations, and refactoring lend themselves to detailed specifications and tests; interactive frontend work often requires closer review because visual states, styling, and user behavior are harder to specify upfront.
- Parallel coding-agent orchestration protects human attention. As agents spend longer running tests, type checks, and browser automation, developers need interfaces that coordinate multiple independent workstreams without constant context switching. Automated frontend quality assurance remains promising but not yet routine.
- Open-source popularity does not guarantee commercial viability. Orchestration products can attract users while most spending flows to model providers and coding-agent vendors, making enterprise distribution and positioning commercially decisive.
In April 2026, Knight-Webb announced bloop’s shutdown during AI Engineer Europe, using Vibe Kanban itself to prepare the announcement. The company’s closure ended its commercial operation while leaving Vibe Kanban available as open-source software.
