Convictional

Announcing that Convictional is shutting down and the path for customers to migrate their data

Convictional is reintroducing itself and launching the first product built to solve goal alignment in organizations, keeping teams mission-aligned by design as AI agents take over task work.

This writing explores the concept of organizational health and how it has changed with AI, from its definition to its importance, and how it can be applied tangibly within a real organization in this day and age.

Exploring multi-agent collaboration and goal alignment by varying their communication tools. We introduce our work on GoalBench, a turn-based game played by a team of agents operating a software startup with shared and individual goals.

I ran an experiment to measure the 'noise' in our AI deep research pipeline: why the exact same question rarely gives you the same report twice. Here is why that variance is a fundamental limit of computing, and observing that human subjectivity created 2

On humanity, the role of humanity, and why being human first is important for companies in the future.

I ran an experiment to see if 'knowledge bundles' assembled on-the-fly could replace pre-computed knowledge graphs. Here is why this dynamic approach beat our baseline, and how smaller models kept the costs in check.

On the value of human-authored communication in an AI-saturated world, and why companies need new ways to signal that the words came from a person, not an AI model.

Explores the difference between motives and goals and their relationship to AI agent alignment.

An essay that discusses the notion that in order for groups to preserve knowledge, they must struggle with decisions and record that struggle somewhere widely accessible.