Predictive analytics & forecasting
Estimating what might happen, how likely it is, and what would change the outlook. We are interested in forecasts that can be checked against outcomes and improved over time.
AI · Machine learning · Data
We explore how AI, machine learning, and data can help people understand change, challenge assumptions, and make informed decisions. Our interests span markets, business growth, and the way people work.
What changed.
Why it matters.
What to look at next.
The interesting work happens where research, models, and real-world context meet. These are the areas we keep coming back to.
Estimating what might happen, how likely it is, and what would change the outlook. We are interested in forecasts that can be checked against outcomes and improved over time.
Connecting structured data with documents, news, and research. Looking for patterns that hold up when examined in context, with a clear trail back to the evidence.
Exploring what sources such as job postings, product activity, and public records can reveal about a changing business or market. Coverage, bias, and relevance matter as much as novelty.
Building and evaluating ways for language models to find information, connect ideas, and support useful workflows. Keeping sources accessible and human judgment part of the process.
A specific question gives the work direction. Here are a few examples of the problems that interest us.
Organizing evidence, comparing explanations, and assessing how changes in markets or businesses could affect an investment thesis.
Compare a company’s stated priorities with hiring, product activity, and reported results. Look for evidence that supports the thesis, as well as developments that could undermine it.
Research synthesis · Scenario analysisUnderstanding how a market is structured, what customers need, and where further research could reveal an opportunity.
Bring together customer language, changes in a market, and signs of investment. Investigate which needs recur, who experiences them, and whether existing products address them.
Market mapping · Customer understandingHelping people find relevant information, compare options, and spend less time on repetitive analytical work.
Start with a real workflow and its bottlenecks. Try a small change, measure the time saved and errors introduced, and decide where human review remains essential.
Workflow experiments · EvaluationWe start with a question we can investigate and an idea we can test. Small, useful experiments help separate a promising demonstration from something worth relying on.
Understand who is making the decision, what information is missing, and what a useful answer would let them do.
Keep evidence separate from interpretation. Make the sources, assumptions, and limitations easy to inspect.
Check the result against the original question and actual outcomes. Revisit the explanation when the evidence changes.
Have a question about AI, data, or a shared research interest? Send us a message.