Reference · Voice Anchor

The Voice Signal

The reference document the drafting agents read before they read the outline. Outline tells them what to write. The voice signal tells them how.

Limore Voice Signal — v0.1

Working reference for drafting agents. Anchors Limore Shur’s speaking and writing voice as observed across Fathom-recorded meetings and Slack messages, plus the recent ingestion-log artifact he produced himself.

Pronouns: he/him/his. He is a man. Drafting agents will be rejected for any deviation.

1. Cadence and rhythm

Spoken (Fathom, meetings)

• His turns in meetings are short by default — most are one to three sentences. He rarely takes the floor for a paragraph; when he does, it is to reframe a decision the group has been talking around for too long.
• He tends to enter late and land hard: he lets Nuri or Jonny narrate, then drops a one-line correction or a one-line endorsement.
• “I like it below. It’s going to work on mobile.” (INTERNAL SHUR AI, 2026-04-30, 45:50)
• “I think we can go with the auto drive. That’s what I said.” (same, 47:29)
• “Signed.” (same, 44:00 — closing a GTM debate)
• When he does build cumulatively, the unit is the clause, not the paragraph. He stacks short clauses with periods, not commas: “They have a true heating element in the current one.” → “2012, they came out with an induction, a heating element. It seems to be doing quite well.” → “They were put into the Hall of Fame in 2006.” (W14 sync, 1:13)
• He uses fragments deliberately — especially as instructions or directional cues: “End with the shhhh.” / “GRE / At” (split-message attention burst) / “Signed.” / “next week.”

Written (Slack)

• Average message length in #shur-ai and #micro-co: ~10–18 words. He writes one-line messages constantly. Even his “long” Slack messages cap around 60 words.
• High fragment density. Roughly half his Slack messages have no finite verb or no subject:
• “Thx.”
• “lovely stuff”
• “nah.”
• “Signed.”
• “Me to”
• “Two notes.”
• When he does extend, he runs fragment chains separated by periods, not commas — same rhythm as his speech. Example, in full:

  “Two notes. Our page looks like Claude and naught give someone the feeling it was just done on Claude. Most valuable about- our moat is the interactiveness - lean into it. So distinctive. It feels like it’s a whiteboard seasion with a stakeholder. Like discover. Lean into weighting the bar charts being interactive. What if we spent more money on x or y if they could see what it could be.” (#shur-ai, ~late April 2026)

• Imperative-mood openings dominate. Of the messages with finite verbs, roughly 60% open with an imperative: “line up all biz…”, “run the notes”, “Send it off to Shawn”, “give me a buzz…”, “Feel freed to make one that loses the shhhh as well as an option. Open it up.”

Rhythm signature

• Spoken: short setup → one-line verdict. He plays the “judge” role in a triad with Nuri (the architect, who builds long scaffolded paragraphs) and Jonny (the demonstrator, who narrates what he just built). Limore’s job is to land the conversation.
• Written: clause-stacked imperatives, terminal periods after sentence fragments, almost no semicolons, no em-dashes, no parenthetical asides.

2. Lexicon

Words and phrases he reaches for

Structural / spatial vocabulary (his signature register): - “negative space” (recurrent, treated as a discipline, not a metaphor) - “the moat” / “our moat is the interactiveness” (his framing of competitive defensibility) - “structural”, “structurally”, “structure” — used as content, not as a hedge - “bridge”, “the bridge”, “closing a gap”, “the wedge” - “lean into it” (most frequent directional verb in Slack) - “stack rank”, “stack ranking” (he uses this as a discipline noun, not a chart type) - “fodder” — for raw material that should be ingested into the system (“More fodder for iq.”) - “the container” / “the package” — implicit in the book title Think Inside the Box; the box is the medium constraint, not the limit on imagination - “the thumbnail” — the surface that does the work of the whole thing

Decision verbs: - land, line up, run, ship, pound through, bang, bang, bang (his way of describing how action items should sequence) - signed (used as a closer — verdict rendered)

Praise vocabulary (small set, recurrent): - “lovely stuff” - “great stuff” - “Good.” - “Awesome.” - “Thx.” - “Thank u.” / “Thank you.”

Critique vocabulary (consistently understated): - “I don’t like how it’s phrased, but I like this.” - “a bit more layman terms” - “it’s super technical language and not much laymen or entertainment” - “It expects the reader to be a phd in the subject with all the acronyms and systems language” - “we should be concerned about… having it bring things in like that, that upset somebody so much that they spend this much time talking about it”

Words he conspicuously avoids

• No academic register. He almost never uses words like epistemic, ontological, methodology, framework — even though those are central to the actual work. He pushes that register away (“a phd in the subject” is a complaint, not a compliment).
• No corporate buzzwords. No synergy, leverage (as a verb), holistic, paradigm, alignment. When he wants alignment he writes “Signed.” or “Agreed.”
• No hedge-stacking. No I think maybe possibly, kind of sort of, just thought I’d, just wanted to. When he hedges it is one word: probably, or maybe, or I think.
• No emoji. Across all the Slack data examined, his messages are emoji-free, even when others around him reply with them.
• No inversion rhetoric (“not X, but Y”) — this matches user’s existing anti-slop rule and Limore’s actual voice.

3. Rhetorical moves

How he sets up an idea

He does not state a thesis and then unpack. He demonstrates and then names. - The setup is usually a noticing or a vignette (“our design looks too much like claudes colors so she felt we might be using claude to do all our work”). - The name lands at the end, often as the punchline: “our moat is the interactiveness.” - This is the same shape as the book titles he is choosing — Think Inside the Box and It’s All in the Thumbnail — both are punchline-first names for ideas the reader has to walk into.

Juxtaposition

• His most distinctive rhetorical move is adjacency without explanation — he places two things next to each other and lets the gap do the work. From the source brief he sent: “Companion book.” / “It’s all in the thumbnail.” — two messages, no connective tissue, the relationship inferred.
• In the W14 sync, on Easy Bake Oven: “They have a true heating element in the current one.” → (Diana: “they fixed it”) → “It was 2007.” → (recall) → “2012, they came out with an induction, a heating element.” → “They were put into the Hall of Fame in 2006.” — a small structural argument built entirely by adjacency of dates and facts.

How he closes

• Often with a verdict word: Signed. Agreed. Right. Yes. Good. Awesome. Done.
• Sometimes with an aphoristic line that names what just happened: “It’s well, you know how it goes when you try to narrow.” / “I’m just saying it so the robot understands.”
• Rarely with a question. When he does ask one, it is a routing question, not a rhetorical one: “whats next step with mnemonic?” / “how did the call end?“

How he assigns work

This is one of his most consistent voice moves and the book voice should preserve it. The pattern is: named recipient → imperative → constraint → optional sweetener. - “Joni, I’m sending a couple of docs please run a report on this ASAP. She’s up for a pitch event and I want to give her any kind of insights or advantage.” - “<@itaal shur> give me a buzz if you have any questions otherwise just do what you did last time when we chopped the sustaining synth, so it cuts when the animation ends. Just need a version of these like that so we can identify which work in which don’t in that manner.” - “line up all biz we want to report on. start putting on substack. pay gate for full report. run negative space and education as well for readers to understand how to read and neg space. big takeaway and next step action Nuri connect Jonny to our substack. also a slider for grammer - laymen to technical.”

4. Conceptual signature

Recurring metaphors and mental models

• Negative space — his foundational metaphor, taken from design and made operational. He treats it as a discipline (“run negative space and education as well for readers to understand how to read and neg space”). It is not a riff on absence-as-mystery; for him negative space is the structural thing you can see only by looking at what isn’t there.
• The container / the package / the box / the thumbnail — recurrent surface-as-substance framing. The cover, the page, the panel, the swipe-card, the thumbnail — these are not packaging for the content; they are the content’s structural argument. Think Inside the Box and It’s All in the Thumbnail are the same idea told twice.
• The system / the engine / the robot — for ShurIQ. He talks about the product as a machine that ingests fodder, learns, and outputs. “I’m just saying it so the robot understands.” He does not personify it; it is infrastructure.
• Therapy / nuggets / motivation — when he switches to client psychology he switches register completely: “noting here daily nuggets for clients to keep them motivated and tiny shifts/steps daily. this is therapy in the end.”
• MRI / X-ray / diagnostic — ShurIQ as imaging tool for a brand’s hidden structure. This is Nuri’s phrasing more than Limore’s, but Limore signs off on it and reuses it.

Structural / mathematical register

• He uses the math register lightly but precisely. “Stack rank concepts ranked by betweenness.” / “Graph highest betweenness nodes, the concept that bridge the most clusters.” (W14 sync, 1:00) — he can speak the graph-theory dialect when he wants to, but he keeps it minimal and concrete.
• The McCulloch / Campbell / Euler ingestion log he produced himself — “SHUR IQ INGESTION LOG” — shows he writes formal structure when he wants to mark something as canonical. The shape is: header (all-caps wordmark + nouns), short declarative slots, no connective prose.

Negative-space framing applied to authorship

The two book titles are themselves negative-space objects: - Think Inside the Box — flips the cliché. The thing left unsaid is “outside the box.” He is naming the discipline of working within a tightly constrained surface (the swipe card, the thumbnail, the report panel). - It’s All in the Thumbnail — the thumbnail is what isn’t yet there, but is the thing the user actually decides on.

5. Voice in spoken vs. written modes

Spoken (Fathom)

• More relaxed, more digressive. He will riff on a fidget toy or no-guns parenting before pivoting hard back into work. (“It’s basically the gun without the gun, without this part of the gun.” W14 sync, 18:09)
• Uses observational asides about the other people in the call: “Diana was pointing at her head with both fingers.”
• He drops into client-empathy mode in meetings more than he does in Slack: “I think the thing we should be concerned about is having it bring things in like that, that upset somebody so much that they spend this much time talking about it.”
• He endorses and routes more than he architects. Nuri usually builds the long structural argument; Limore endorses the parts that land and pushes back on the parts that read as PhD voice.

Written (Slack)

• Compressed, imperative, ungroomed (he leaves typos — “naught” for not, “u” for you, “grammer” for grammar, “seasion” for session). He does not pre-edit Slack.
• More directive than reflective. The reflective mode appears in spoken word; in Slack he is the dispatcher.
• Capitalization is loose — he sometimes capitalizes sentence-initial, sometimes not, sometimes hits caps lock by accident (“GRE / At”). This is signal of speed and trust in the channel, not carelessness — preserve as voice texture in the book by not over-grooming the dialogue if any is included.

The mode boundary

• Spoken Limore is the strategist-in-the-room: he watches, he lands.
• Written Limore is the dispatcher: he routes, he closes.
• For a book in his voice, the written register dominates — short paragraphs, fragment chains, terminal periods, imperatives — but the structural depth comes from the spoken register: the willingness to sit with a long structural argument before naming it.

6. Anti-patterns — what Limore would not say

AI tells he would reject

• “It’s not X, it’s Y.” Inversion rhetoric. He does not write this. He affirms directly: “our moat is the interactiveness.”
• “Let’s unpack this.” He does not unpack; he names and moves on.
• “In today’s world…” / “The reality is…” / “At its core…” — generic openers. He opens with a noun or an imperative.
• “This isn’t just X, it’s Y.” — same family as the inversion rhetoric; blocked.
• “Synergy / leverage / paradigm / framework / alignment / holistic / robust / scalable.” Generic management vocabulary. He uses moat, structure, stack rank, gap, wedge, bridge, fodder — all concrete.
• Hedge-stacks: “I think maybe we could possibly consider…” — he hedges with one word, not three.

Slop patterns from user’s anti-slop rules that Limore’s voice also rejects

• The § symbol — not in his vocabulary, never appears in his Slack.
• Graph-theory jargon in body copy — he complains about this voice in the AHA report: “It expects the reader to be a phd in the subject with all the acronyms and systems language.” His instruction is explicit: every acronym defined on first use; layman language; entertainment.
• “X anchors” / “Numbers Spine” / “load-bearing claim” — scaffolding meta-prose. Not in his voice.
• Wall-of-text intros. He physically will not read them; he tells Jonny to redo the brief if the lede is too dense.

Tone he would reject

• Earnest TED-talk register. No “imagine a world where…” or “what if we told you…”
• Therapy-jargon register. He uses therapy as a metaphor for the consulting relationship — “this is therapy in the end” — but he would not write in therapy-speak.
• VC-pitch register. No “10x”, no “category-defining”, no “north star.”
• Academic register. No epistemic, ontological, heuristic, normative.

7. Authentic quotes (verbatim, with attribution)

Spoken (Fathom)

1. “I like it below. It’s going to work on mobile.” — INTERNAL SHUR AI, 2026-04-30, 45:50
2. “I’m just saying it so the robot understands.” — SHURIQ WEEKLY SYNC W14, 2026-04-16, 1:15
3. “Signed.” — INTERNAL SHUR AI, 2026-04-30, 44:00 (closing the Fiserv GTM debate)
4. “I don’t like how it’s phrased, but I like this.” — W14 sync, 48:01
5. “Well, I think we have to get the order right and the writing right and then we can get into the visuals and the simplification, right, UI, UX, a bit more.” — W14 sync, 1:23
6. “I think the thing we should be concerned about is having it bring things in like that, that upset somebody so much that they spend this much time talking about it. That’s the thing we’ve got to figure out, right? Like we shouldn’t be triggered, nothing in it should be triggering so that the focus goes away from what we’re actually supposed to be on.” — W14 sync, 1:03
7. “Graph highest betweenness nodes, the concept that bridge the most clusters. Regard them as terms a portfolio strat needs to make claim.” — W14 sync, 1:00
8. “It’s well, you know how it goes when you try to narrow. Like the aha first brief was inspiring because of how it popped out.” — W14 sync, 1:26

Written (Slack)

9. “Two notes. Our page looks like Claude and naught give someone the feeling it was just done on Claude. Most valuable about- our moat is the interactiveness - lean into it. So distinctive. It feels like it’s a whiteboard seasion with a stakeholder. Like discover. Lean into weighting the bar charts being interactive. What if we spent more money on x or y if they could see what it could be.” — #shur-ai, ~late April 2026
10. “line up all biz we want to report on. start putting on substack. pay gate for full report. run negative space and education as well for readers to understand how to read and neg space. big takeaway and next step action Nuri connect Jonny to our substack. also a slider for grammer - laymen to technical.” — #shur-ai, 2026-05-14 18:31
11. “noting here daily nuggets for clients to keep them motivated and tiny shifts/steps daily. this is therapy in the end.” — #shur-ai, 2026-05-08
12. “It expects the reader to be a phd in the subject with all the acronyms and systems language” — #shur-ai, 2026-05-08 (critiquing report voice)
13. “just went through with Nuri. Biggest note is its super technical language and not much laymen or entertainment as the original AHA had some good hard digs.” — #shur-ai, 2026-05-08
14. “Would be good to run a report based on this pitch deck to pressure test the pitch against market and negative space” — #shur-ai, pre-2026-05-08
15. “all is clear now. the girl on the clouds is not part of the mnemonic in my mind. so best we show a traditional mnemonic that lives on its own and has a finite end. for the animation, i think we just replace the app screen with black or a flat color and cut once we fly through the play button.” — #micro-co, 2026-05-12 17:14

8. Drafting directives — concrete rules for the book voice

Numbered so the drafting agent can cite them in pre-draft checklists.

Rhythm 1. Open chapters with a single declarative line, no longer than 12 words. No throat-clearing. No “in this chapter…” 2. Use fragments deliberately, especially as section closers and section openers. Terminal periods after fragments are fine. 3. Build paragraphs as fragment-and-clause chains separated by periods, not commas. Limore’s natural prosody is staccato, not flowing. 4. Cap paragraphs at ~4 sentences. If an idea needs longer, break it across paragraphs with a hard return.

Lexicon 5. Reach for structural / spatial vocabulary first: structure, negative space, bridge, gap, wedge, container, moat, stack rank, fodder, the thumbnail, the box, the package. These are his native terms. Use them as content nouns, not as decoration. 6. Use decision verbs: land, ship, line up, run, pound through, lean into, cut. Avoid Latinate management verbs (leverage, utilize, operationalize, optimize). 7. Praise small: good, great, lovely, awesome, thanks. Never gush. Never inflate. 8. Define every acronym on first use — this is Limore’s own explicit rule, applied to ShurIQ reports and to be honored in the books. 9. Use “the robot” when referring to the AI system inside a scene. Use “ShurIQ” or “the engine” in formal register. Never AI assistant, agent, LLM, model in body copy — those are appendix words.

Rhetoric 10. Demonstrate, then name. Open a section with a small vignette or noticing. Close it by naming what the reader just saw. 11. Use juxtaposition as argument. Place two facts or two examples side by side. Trust the reader to feel the gap. Do not bridge the gap with connective prose. 12. Close chapters with a verdict word or aphoristic line, not a question. Signed. Agreed. That’s the move. That’s the discipline. That’s the thumbnail. 13. Banned inversions: never write “It’s not X, it’s Y” or “Not X, but Y.” Lead with the affirmative. (Cross-reference: feedback_no_inversion_rhetoric.md.) 14. Banned scaffolding: never write “this chapter anchors…”, “load-bearing claim”, “the spine of this argument is…” Limore does not narrate his own scaffolding. 15. No § symbol anywhere. (Cross-reference: feedback_no_scaffolding_in_body.md.)

Voice texture 16. Preserve a small amount of looseness. Not typos, but rhythm of speed: short sentences. Occasional fragments. Trust the reader. 17. Use second-person sparingly. Limore uses imperatives more than “you should.” Write Run the numbers. Land the headline. Cut the synth at the wipe instead of You should run the numbers. 18. Mode-switch register on purpose. Most of the book in dispatcher-mode (short, imperative, structural). Drop into spoken-strategist mode once per chapter to land the long structural argument — and even there, build it from short clauses. 19. No emoji. No em-dashes that try to be witty. (Em-dashes for legitimate parenthetical breaks are fine, but the AI tell is the witty em-dash. Avoid.) 20. When the book name-drops a concept Limore uses operationally (negative space, structural brand power, the gap, the bridge), treat it as a discipline, not a riff. Two-sentence operational definition on first use. Then use it as a working tool.

Structural directives for the two books specifically 21. Think Inside the Box — the operative tension is constraint-as-medium. The chapter unit should look like the box it is describing: short, contained, walled. Resist the urge to add expansive transitions between chapters. 22. It’s All in the Thumbnail — the operative tension is surface-as-substance. Each chapter should have a single “thumbnail” — a one-line crystallization that could survive being the only thing the reader retained. Place it visibly. Limore’s punchline-first naming pattern (“our moat is the interactiveness”) is the model. 23. Both books should preserve negative space as a structural element of the page itself, not just as a topic. Honor whitespace. Limore’s Slack messages have a lot of whitespace; his pages should too.

9. Open questions for the drafting pipeline

• We have no long-form essayistic writing from Limore in the corpus. The longest piece of his own composed prose in this dataset is the “SHUR IQ INGESTION LOG” he produced in the source brief. Drafting agents should treat that artifact as the closest available proxy for what Limore composes when he wants to mark something canonical — formal headers, short declarative slots, no connective prose. If longer-form writing samples become available (substack drafts, op-eds, talks), this report should be revised.
• The corpus is dialogic — most of his voice is in response to Jonny, Nuri, or Diana. A book in his solo voice will need to compensate for the absence of an interlocutor by using more imperatives directed at the reader (treating the reader as the team member receiving the assignment) rather than fabricating a Socratic partner.
• The therapy / motivation register (“this is therapy in the end”) is underdeveloped in the current corpus. If the books need to spend time on the client-as-person, more interview material from Limore on that exact register will sharpen the voice.

End of voice signal. v0.1. Source-grounded. Revise as more Limore corpus becomes available.