The Companion
Chat on the left, instruments on the right. The Companion is the ·Sum family's signature surface — a slender live window kept beside whichever AI you talk to, bridged to it by MCP — the most broadly useful interface idea I've shipped in years.
It exists to solve a blind spot in the bet the whole family makes. If user-chosen cognitive hosts — chat apps with deep access today, richer harnesses tomorrow — are becoming the primary computing environment, then the conversation is where the work happens — but a conversation is a terrible place to see state. Rosters, indexes, records, what-changed-while-I-was-away: the model can recite them, but recitation scrolls away and goes stale the moment it's printed. The classical fix is to build an app — and now you've built exactly the thing the bet says loses: another surface competing for attention, a second place to live, eventually a second chat.
Two standing surfaces
The Companion is the family's refusal of that fork. The doctrine: a user manages exactly two standing surfaces — their chat and the Companion. Everything deeper is transient, singular, and user-invoked. The Companion itself is deliberately modest: a slender, plain-URL browser window, host-agnostic by design — it sits as happily beside Claude as beside ChatGPT, in any browser, as a popped-out panel or installed as a chrome-less app. Slender is the identity: it's an instrument panel, not a workspace. When you need full-size reading, the escalation opens in one named reader window that every escalation reuses — the third window never multiplies.
The window is the shape; the doctrine is deeper. What the family actually holds constant is two standing roles: a conversation plane for intent, authorship, and reasoning, and an instrument plane for orientation, selection, inspection, and authorization. The Companion is the current portable form of the instrument plane, not an eternal window shape — a future host may seat a website or a persistent rail directly beside chat without changing the division of labor an inch. Conversationally useful does not mean conversationally complete: work that needs sustained visual orientation, structured selection, or consequential approval belongs on the instrument plane, and forcing it into chat would erase the very boundary the family is built to preserve.
The bridge is MCP — and the roles never blur
The division of labor is strict. Your own chatbot holds the MCP connection and does all the writing; the Companion is read-only by construction. It refetches when you focus it and on a slow visible tick, so your agent's writes — and its deletions — simply appear, without ceremony. Reality drives the display. The agent owns the writing; the Companion owns the showing; and the two surfaces never trade places.
The affordance rules are just as strict, and there are only two: pills copy, links navigate. A chip's face is the literal text it copies; underlined text moves you. Copy a chip, paste it into your own chat, finish the sentence in your own words — and your agent performs the act. There is deliberately no send button anywhere in the Companion. We considered one — a quick-send box is the obvious “improvement” — and rejected it, because it would quietly recreate a second chat and route around the agent's judgment. The Companion can show dialogue — in Suminar's seminar view it displays the live record of the conversation itself — but it never joins it. It prepares speech; it never delivers it.
[[Our Cats]] #dave-lisa tell me about Willie — cites a page, names the sum, and asks in plain language; the agent reads the shared Sum and answers. On the right, the sum's pages as copyable chips. Read-only throughout: the chat performs. Click or tap to view full screen.A grammar you can copy
The chips work because the family runs on a three-mark grammar: @ names who, # names where, [[…]] names what. An @handle addresses something that can talk back — a person, or a source agent. A #handle names a container — a sum, a seminar. Double brackets cite a page: you hail an @, you enter a #, you quote a [[…]]. Notation is precision, never a requirement — plain language stays first-class, and “the page about the cats” resolves just fine. But when you want to bind your words to an exact object, the Companion hands you the binding: every chip is a piece of the grammar made copyable. It is deixis for chat — you point at the thing by copying its name.
Why a text grammar at all? Because it is the compromise the present forces. Today's host platforms give an external application exactly one channel into the conversation: text the user pastes. MCP lets the agent reach the product's tools, but nothing lets the product reach the user's composer — there is no way for the Companion to hand your chat a typed reference, no dragging a page into the message box, no host-level binding between a chip and the object it names. Until and unless OpenAI and Anthropic open richer interaction schemes between their platforms and outside applications, the copyable mark is the sturdiest bridge available: notation that survives the clipboard, reads as plain language to the person, and resolves as an exact reference for the agent. It is not ideal, and I would retire it the day the hosts allow something better — a chip you drop into the composer that carries its binding with it. The grammar points at that future; the clipboard is just how it ships today — and the first piece of that future has already arrived, below.
The host declined to ventriloquize
Look closely at what Claude says in that second screenshot, because I didn't script it. Asked about a source, the host answers: best to ask it directly rather than me paraphrase secondhand — and then puts the question to the named agent and relays its cited answer. That is the exact behavior Against the Middleman and Who Are You Talking To? argue for — the assistant refusing to absorb another voice into its own — emerging from the product's structure rather than from a rule. When the architecture makes the speaker legible — named agents, real authorship boundaries, verified quotation — even the host prefers attribution over paraphrase. You don't have to defeat the ventriloquist. You have to build the room where ventriloquism stops being the path of least resistance.
The chat performs. The Companion shows. Neither pretends to be the other.
The sequel: the compromise began retiring itself
When this essay first ran, the bridge carried traffic one way and the copyable marks did all the binding. Two weeks later, the future the grammar points at arrived early — in Mail·Sum, the family's email-fed Sum, now in private alpha. A member-carried pairing code binds one conversation to one Companion window. Say “show my Mail·Sum correspondence with @wes on my Companion” and the adjacent window turns to that person's entry — the handle resolved against your own roster, the intent applied in under a second, over ordinary MCP. And the window reports back what it is showing — as references the agent can look up, never as content that could smuggle instructions into the model. The copyable grammar bootstrapped its own successor — the pairing code you paste is the last clipboard act, and everything after it is typed.
The rules that keep the roles from blurring survived the upgrade intact. View state moves freely — showing you something never needs permission. Proposals stage reversibly. Commitment — a send, a disclosure, a deletion — stays on the product's authenticated surface, behind the owner's unforgeable act. The Companion still never speaks; it now also listens, in a typed vocabulary with named acts and named failures. That vocabulary gets its own essay: The Grammar of the Second Surface.
The second sequel: the window became the app
Added September 2026.
The slender window did not survive contact with a month of real use, and that is the best thing that could have happened to it. Two lessons arrived in quick succession from Mail·Sum. First, the copyable marks retired: once every product surface reports what the member is looking at, “this” resolves without being carried. Nouns rise from the surface; verbs stay in the conversation, where your authority and the record live. Pills that copied citations came out product-wide. Second, slender was never the identity — the role was. Every escalation from the panel to a reader window was the panel admitting it was too small for the work. So in late August the Companion became the app: one installed window, a progressive web app on desktop and mobile in Chrome, a product rail beside a full content pane, every move a real navigation. The assistant's pointing now lands as ordinary in-app navigation. The screenshots above are history — kept because they show the compromise the grammar was, and because the host's unscripted refusal to ventriloquize still stands.
And the app gained a second door. Any assistant still reaches the kernel through an ordinary MCP connector; inside ChatGPT's built-in browser the same page registers WebMCP “site tools,” so the host can work the very page you are looking at, in your signed-in session. That is the future this essay said the hosts would have to open, arriving from the platform side. It changed nothing in the division of labor: the page is an entrance, not a second brain; view moves freely; commitment still passes the gate at the site's edge — now attached to the window, because in an agent browser the member's hand and the assistant's share one input channel, and the site can no longer tell who clicked. Meanwhile the family's two local-custody members read the doctrine literally. Draft·Sum's instrument plane is a pane inside Word; Photo·Sum's is Lightroom itself. The Companion was the first portable form of the instrument plane. The plane is wherever the work already lives (the case study).
What it demonstrates
The Companion is a small thing — a few hundred pixels of read-only browser window — and that is the point. It shows that chat can stay the one conversation you live in without going blind; that a product can be fully present without becoming a persona or a second place to live; and that the interface stays a contract when each surface's authority is explicit: the agent writes, the panel shows, the person speaks in their own chat and holds the only microphone. It is also the working probe for the second of this site's two bets: that the richer UX develops beside the chat, between it and a real website, not inside it as a widget (the grammar). Suminar and Mem·Sum still ship the slender form today, free in public beta; Mail·Sum has grown it into the one-window app — because beside the chat is exactly where the ·Sum family intends to live (the case study).
Essay — building. The ·Sum family's first instrument plane, shown live, with its two sequels. ← All essays