Explore our latest articles and insights.
Showing 1-9 of 52 articles
A solo UX designer at Holmes Corporation transformed research from slow, hard-to-gather interviews into fast, evidence-based decision-making by integrating AskRally into product development. With synthetic testing powering rapid validation, the team now iterates faster, presents data-backed recommendations, and builds with confidence—without growing headcount.
Wisdom of the Agents Newsletter 🤖 lvl 23
Ask Rally evolved its AI research tool from purely generative models to a system grounded in real human data. Facing issues like AI-sounding responses, bias, and a lack of nuanced preferences, they refined AI personas using diverse data and real human examples. Key developments include "survey boosting" and "GenPop," a virtual panel built on authentic human interviews. This approach ensures realistic responses, reduces hallucinations, and makes market research more accessible and insightful. Real people are the "secret ingredient" for authentic AI personas.
It's possible to feed AI personas different pieces of content to measure how exposure to competing narratives shifts purchase intent in real time.
Most teams overcomplicate audience building by starting at the wrong level of sophistication. This guide reframes the problem—showing how AI personas can evolve from quick prototypes to calibrated virtual panels that explore real conversational realities, not predict behavior.
When AI personas fail to predict consumer behavior, it’s not because the models are broken, it’s because we’re asking them to do something humans can’t even do for themselves. The real value isn’t in predicting what people *will* do, but in understanding how they *talk* about what they might — because that’s where decisions, and markets, are actually made.
A technical deep-dive showing how to automate audience uploads to AskRally using its API instead of manually editing JSON files. This guide walks developers and agencies through a reliable two-step process to create and update audiences programmatically across multiple AskRally accounts.
At the London Agentic AI Meetup, Mike Taylor showcased how to train an **LLM-as-a-Judge** and use it to optimize creative generative tasks through the **Evaluator–Optimizer pattern in DSPy**. His live demo built a *comedian AI* that learns to tell better jokes using **GEPA evolutionary optimization**, illustrating how to systematically create, evaluate, and improve AI programs.
Listen to this behind the scenes impromptu training between Mike and the Every.to team. Mike gives an introduction to the DSPY framework.
