The Feast Field Guide to the Age of Algorithms
Life should be a feast. The algos have other plans.
The industry already has a vocabulary for what's happening to brands. It's all urgency and inevitability — optimize or die, the agent is the new customer, friction is the enemy. This is a different kind of guide. The words here name mechanisms instead of panics. Some of them I coined because nothing else fit; all of them are the language I actually use when I'm in the room with a CMO who can feel the ground moving and can't yet name it. Borrow them.
— Jess Graham, The Feast, 2026
Discovery Debt
The hype: discovery is a marketing problem you solve by buying more reach.
My definition: the compounding cost of being unfindable on your own terms.
Not a tax you pay once — a debt that compounds. Every cycle you rent discovery from a platform instead of building your own, the balance grows, and the agent era just raised the interest rate. (Replaces the older "discovery tax" — a tax is a one-time charge, and this was never one-time.)
Discovery vs. Retrieval
The hype: AI agents make discovery effortless.
My definition: agents don't discover. They retrieve.
Discovery is a human act — wandering, adjacency, the thing you weren't looking for. Retrieval is optimization against a query. Treating them as the same thing is the category error the whole hype cycle is built on.
The Seamlessness Paradox
The hype: removing friction is a gift to the customer.
My definition: the thing sold as a consumer benefit is a consumer and brand loss.
Seamlessness serves the platform and the agent. It costs the brand its last chance to be chosen, and costs the customer the experience of choosing. (Replaces "frictionless trap.")
→ You Call It Friction, I Call It Human Experience
Strategic Friction
The hype: all friction is waste; engineer it out.
My definition: the friction worth keeping — the part of the experience where the choosing actually happens.
Remove every bump and you haven't served the customer, you've deleted the moment they decide. Strategic friction is the friction that earns the choice instead of skipping it.
→ You Call It Friction, I Call It Human Experience
Texture
The hype: if a step doesn't convert, cut it.
My definition: the positive face of friction — the specific, lived grain that ease sands off.
Strategic friction is the principle; texture is what it feels like from the customer's side — the pause where attention slows down and a relationship actually forms. Taste, loyalty, and love are all made in the pause. You can't strike a match against silk.
→ Friction, Working In Your Favor
Preference Compression
The hype: personalization gives people more of what they want.
My definition: the narrowing of taste that happens when discovery is removed and only the existing preference profile gets served back.
Preferences get constructed in the act of finding — "the preference didn't fully exist until I walked in the store." An agent that serves back the profile you already have is optimizing for the person who already exists; the one you were becoming never shows up.
Evaluation Compression
The hype: agents make buying decisions efficient.
My definition: the 20–30 touchpoints of a considered purchase collapse into a single API call.
The agent doesn't browse, compare, or deliberate. Everything that made brand equity matter during the decision gets structurally squeezed out — by design, not accident.
→ Meanwhile, As You Were Optimizing
Agentic Brand Erasure
The hype: you just need to be "visible" to the AI.
My definition: your brand gets erased from the decision entirely.
Not invisible — that's the visibility vendors' framing, and it pulls toward their fix. Erased. Brands get erased by platform dependency, by performance marketing hollowing out the middle, by over-optimization. Agents are only the most efficient mechanism yet. (Replaces "agentic invisibility.")
→ Meanwhile, As You Were Optimizing · The Relationship is the Price
Binaries and Beige
The hype: it's "personalization."
My definition: machines reduce messy human stuff to binaries, and flatten taste toward beige.
Binaries is the logic layer — in or out of the filter. Beige is the taste layer — the statistical mean, the recommended-because-everyone-liked-it. A useful test against any pitch dressed up as "optimization." (With a nod to Kyle Chayka's Filterworld*.)*
→ The Algo In Your Head · My New Favorite Sweater
Beigeville
The place the algorithm sends you when it's optimized for the mean.
If "binaries and beige" is the mechanism, Beigeville is the destination. The beige version of anything is any version an algorithm has ever recommended. You didn't choose it — you were sent there.
The Four-Segment Model
How attention and spend actually sort — and where brand equity still does work.
Committed (locked-in, auto-replenish), Serendipity (browsing, stumbling), Consequential (high-stakes, research-heavy), Commodity (price-driven, agent-optimized). Brand still matters in Serendipity and Consequential — which is exactly where agents threaten to flatten it. (Replaces the three-zone "pickle zone" model.)
A living field guide. These are working terms for brand strategy in the age of algorithms, developed across The Feast in 2026. New words get added as the work names them.