We introduce \textbf{Directed Information $γ$-covering}, a simple but general framework for redundancy-aware context engineering. Directed information (DI), a causal analogue of mutual information, measures asymmetric predictiveness between chunks. If $\operatorname{DI}_{i \to j} \ge H(C_j) - γ$, then $C_i$ suffices to represent $C_j$ up to $γ$ bits. Building on this criterion, we formulate context selection as a $γ$-cover problem and propose a greedy algorithm with provable guarantees: it preserves query information within bounded slack, inherits $(1+\ln n)$ and $(1-1/e)$ approximations from submodular set cover, and enforces a diversity margin. Importantly, building the $γ$-cover is \emph{query-agnostic}: it incurs no online cost and can be computed once offline and amortized across all queries. Experiments on HotpotQA show that $γ$-covering consistently improves over BM25, a competitive baseline, and provides clear advantages in hard-decision regimes such as context compression and single-slot prompt selection. These results establish DI $γ$-covering as a principled, self-organizing backbone for modern LLM pipelines.

Have you ever joyously stepped out to your backyard garden, freshly brewed coffee in hand, only to find your meticulously cared-for plants and herbs wilted and dying? Was the soil too dry? Did pests find their way in? During times like these, some frustrated gardeners may wish their fickle ficus would just tell them what it needs. A new Microsoft-partnered project in the UK is trying to see if that concept can be demonstrated in the real-word.

Hardcore gardeners sometimes, when no one else is listening, talk quietly to their prize blooms. But at next year's Chelsea flower show, visitors will be encouraged to have a chat with its first ever AI-powered garden. The garden designer Tom Massey has partnered with Microsoft to create the Avanade "intelligent" garden. Sensors in the soil are partnered with an AI trained on Royal Horticultural Society plant data and gardening advice, meaning visitors can ask the garden: "How are you?" Massey said: "It could answer: I need a bit more water, I can do with a haircut, maybe."

They say talking to plants helps them grow - with King Charles a keen believer in the theory. And soon plants will'talk' back thanks to AI that will allow gardeners to converse with their plot. Budding horticulturalists will be able to ask what their gardens need - and even how they are feeling. The development is to be showcased at next year's Chelsea Flower Show, which will include a garden that can talk back. The Avanade garden designed by Tom Massey and Je Ahn will be able to tell gardeners whether it is sensible to water - as it is going to rain later in the day - or that the water is moist enough already.

Mathematical models have been used to augment or replace human decision-making since the invention of the calculator, bolstered by the notion that a machine won't make mistakes. Yet many people are averse to using algorithms, preferring instead to rely on their instincts when it comes to a variety of decisions. New research from Cade Massey and Joseph Simmons, professors in Wharton's department of operations, information and decisions, and Berkeley J. Dietvorst from the University of Chicago finds that control is at the core of the matter. If you give decision-makers a measure of control over the model, they are more like to use it. Massey and Simmons spoke to Knowledge@Wharton about the implications of their research.