MARK·GRID·CELLS·SEED·

// THE AGENTIC AD AGENCY PLATFORM

Every brand has a sigil. Sigil refines it.

Generates · publishes · ingests engagement, per brand.

// POWERING16 brands · 4 agencies · 12,400 posts shipped
X·LINKEDIN·INSTAGRAM·FACEBOOK·THREADS·TIKTOK·YOUTUBE·MASTODON
01Play with it

Feel the current.

Every piece below is live. Type, drag, reseed, and flip. No signup, all client-side.

Real outputs, in their real feeds. Instagram.
Mmarlowandvine
Golden hour, done right. Our house margarita, poured fresh.

marlowandvine Golden hour, done right. Our house margarita, poured fresh.

  • Length is in rangepass
  • Banned phrase: revolutionaryblock
  • No clear call to actionwarn
  • Hashtags look cleanpass
Mix the kit.

Every job, one timeline. Your customers always know what is next.

Book a demo
Week 1 of 1240% on brand
Sample output at week 1
sigil

02 // HOW IT WORKS

Brief in. Tuned posts out. Engagement back in.

STEP I · BRIEF

VOICE · lockedKIT · loadedREFS · 4 chars

STEP II · FLEET

Curator routing references
Generator drafting per-platform
Critic checking 18 rules
Auto-corrector regen-to-fit
CRITIC · 18 rulesAUTO-CORRECT · onCOST · $0.04/post

STEP III · OUT

X@crossbarcactus
Aug 14 · 09:00
LINKEDINCrossbar Cactus
Aug 14 · 09:05
IG@crossbar.cactus
Aug 14 · 09:10
QUEUE · 3 scheduledMODE · auto-executeRETRIES · 0
↻ engagement loops back
03 // REAL OUTPUTS · One brief. Tuned per platform.

Same brand. Same logo. Different voice per channel, written and visually composed by Sigil from one 14-token brief.

X · @crossbarcactus
Crossbar & Cactus
@crossbarcactus · 8h
Three new blends on tap Tuesday. Maria's Saguaro Margarita is back. No fanfare, just come hang. #DallasEats
Two craft cocktails on a polished oak bar at dusk, Crossbar & Cactus
CROSSBAR& CACTUS
MAY 7 · DALLAS@CROSSBARCACTUS
2:14 PM · May 7, 2026
LinkedIn · Crossbar & Cactus
Crossbar & Cactus
Hospitality · Dallas, TX · Posted May 22, 2026
Six months in and the room finally feels like itself. We rebuilt the menu around what people actually order, not what a consultant thought they should. The May Tuesday menu launches this week: three blends, three plates, all of them on the wall for a reason. If you're in Dallas and you've been waiting for us to settle in, this is the week. From all of us at Crossbar & Cactus, thanks for the patience.
Polished bar interior at golden hour with craft cocktail foreground, Crossbar & Cactus
CROSSBAR& CACTUS
MAY 22 · DALLAS, TXHOSPITALITY
👍412 reactions · 38 comments · 12 reposts
Instagram · @crossbar.cactus
crossbar.cactus
Dallas, TX
Close-up craft cocktail in a coupe glass, Crossbar & Cactus editorial
CROSSBAR& CACTUS
JUN 11 · CROSSBAR.CACTUSDALLAS, TX
Liked by maya.iglesias and 2,847 others
crossbar.cactus Tuesday's our quietest hour, and our favorite. Saguaro Margarita, Cactus Fries, the Dwayne corner of the bar still warm. Come find us.
#CrossbarCactus #DallasEats #TuesdayMenu
JUN 11 · 2026
Also from one brief · Long-form

Articles and reports. Cited, charted, export-ready.

Sigil writes blog articles and 5 to 10 page reports with a research pass behind every claim: live web sources plus academic papers from Europe PMC, OpenAlex, and arXiv, screened by a relevance judge that would rather cite nothing than cite junk. Pick the angle: brief-led, both sides, for, against, or data only. Reports ship with an executive summary, numbered references, stat callouts, and charts drawn from verified numbers. Export as Markdown or print-ready PDF.

Report · Summit Air & Heating
Heat Pumps in 2026: The Cold-Climate Truth About Performance, Costs, and Rebates
Executive summary

The old claim that heat pumps can't keep up in freezing weather is about a decade out of date. Modern units, correctly sized and installed, keep homes warm in cold climates, according to real field data...

Average installed cost
Heat pump
$8,350
Gas furnace
$5,750
[S11] Carroll, Chesser & Lyons (2020). Air Source Heat Pumps field studies: a systematic review · Renewable and Sustainable Energy Reviews

04 // THE CRITIC

The brand-safety gate every post passes before it goes live.

BEFORE · flagged draft

CC

Crossbar Cactus

@crossbarcactus

Before: flagged draft post with brand-safety violations

Crushing it this Tuesday Best cocktails in Dallas, period. Don't sleep on this!! Tap in below (link in bio)

CRITIC FINDINGS

BLOCKvoice.no-superlatives

BEST, period, DON'T MISS OUT!!!

WARNimage.text-overflow

overlay clipped at right edge

WARNimage.contrast-low

cream text on busy bright bg

BLOCKvoice.emoji-cap

3 emojis, max 1

BLOCKvoice.no-slang

'tap in', "don't sleep"

AFTER · all 18 checks pass

CC

Crossbar Cactus

@crossbarcactus

After: corrected post with all 18 checks passing

Three new blends. Tuesday onward. Crossbar & Cactus, Dallas. Reserve a stool below.

ALL 18 CHECKS PASS

Approved & scheduled · Aug 14 · 09:00

18 rules · full taxonomy

VOICECHAR-CAPIMAGE-TEXTCONTRASTCTA-PLACEHASHTAG-CAPEMOJI-CAPTONEBANNED-TERMSGLOSSARY-USEREF-MATCHNSFWFACTUALNO-COMPETITORSCLAIM-EVIDENCELINK-CHECKALT-TEXTMENTION-RULES
05 // EVERY BRAND HAS A SIGIL. The math is different. The discipline is the same.
CROSSBAR & CACTUS
sigil · 5×5 grid · seed 4f8a
MARIGOLD LABS
sigil · 5×5 grid · seed 7e2c
ASTER PROTOCOL
sigil · 5×5 grid · seed e2a1
LAPIS HEAVY
sigil · 5×5 grid · seed 1d7b

06 // COMPOUNDING. Engagement per post across 52 weeks (Crossbar Cactus sample)

BRAND·crossbarcactusRANGE·1yN·52 postsΔ·+180%
WK1WK13WK26WK39WK52¹WEEK 1 · AI GUESSES VOICE²WEEK 12 · 50+ POSTS FEEDBACK³WEEK 52 · BRAND VOICE IS A MODEL
¹WEEK 1 · AI guesses your voice from brief alone.
²WEEK 12 · Sigil has 50+ approved posts as signal.
³WEEK 52 · Your brand voice is a learnable mathematical fingerprint.

Sigil compounds.

Every approved post is a training signal. Every engagement metric refines the model. By week 12, "AI guess-the-voice" becomes "AI knows-the-voice." By week 52, your brand has a learnable mathematical fingerprint.

→ what Sigil learned in 12 weeks

Week 1: generic draft post output
WEEK 1 · generic draft 12 2 380
Week 12: sharpened voice post output
WEEK 12 · sharpened voice 1.2K 84 28K

07 // ARCHETYPES

Who Sigil is for.

IN-HOUSE TEAM

5-50 brands. One team.

Centralize ops; let team members focus on quality not pasting. Multi-tenancy, role gating, per-brand budgets keep stakeholders aligned without surveillance overhead.

WorkspacesComposerCalendarMembers
See use case
BOUTIQUE AGENCY

100 brands. 10 clients. Same team.

Scale the team without scaling the headcount. CSV bulk import, per-brand brand kits, audit log let you run 10× more brands than you have FTEs.

CSV ImportComposerBrand KitAudit
See use case
SOLO CREATOR

Sigil is your CMO.

1-3 brands. You're the operator. Sigil's auto-execute + scheduler means your brand voice ships even when you're shipping product. The composer becomes your morning ritual; the calendar your accountability.

ComposerCalendarAuto-executeAnalytics
See use case

08 // QUESTIONS

FAQ.

How does Sigil learn from my brand?+
Sigil ingests your brand kit (voice, glossary, references, banned terms) once. Every published post's engagement (likes, comments, shares, click-through) feeds back into the next brief's context. After roughly 30-50 posts the voice tightens measurably; by week 12 you'll notice Sigil anticipating your tone rather than approximating it.
What platforms can it publish to?+
X, LinkedIn, Instagram, Facebook, Threads, TikTok, YouTube. New platforms ship monthly. You can also use Sigil as a draft-only studio and publish manually. The Critic still runs; you just approve and post yourself.
What does the Critic actually check?+
18 rules across three domains: voice (no superlatives, tone match, emoji cap, slang blocklist), platform (character limits per tier, image-text overflow, hashtag count), and brand-safety (banned terms, factual claims, reference-asset match, competitor mentions). Every rule can be toggled per brand.
What if the Critic blocks a post?+
Two paths: auto-correct (Sigil regenerates with the violation flagged inline, usually 1 retry, costs ~$0.003) or human review queue. Composer defaults to manual review so you stay in the loop. Bulk-import via CSV defaults to auto-correct, which you can override per row.
Can Sigil write blog articles and long reports?+
Yes. Blogs and reports are a first-class format next to social posts. Articles run 800 to 1500 words; reports run 2500 words and up with an executive summary, numbered references, stat callouts, and charts drawn only from verified research numbers. Both export as Markdown, plain text, or print-ready PDF. Customers can request them from the portal with an angle (both sides, for, against, or data only) and optional academic citations.
Where do the citations come from?+
A research pass runs before writing: live web search plus academic databases (Europe PMC, Semantic Scholar, OpenAlex, arXiv) routed by topic. A relevance judge screens every paper and drops keyword-only matches. If nothing genuinely supports the topic, the article ships without citations rather than inventing them. Every kept source is listed in the review panel with authors, venue, year, and citation count.
How much does it cost to run?+
Generation cost depends on your model mix. With Sigil's default Config D (DeepSeek text + gpt-image-2 visuals), a typical 3-platform post runs ~$0.04 in direct API cost. Platform fee scales from a pilot tier to full enterprise. Contact us for current pricing during alpha.
Can I export my data?+
Yes. Full CSV export of posts, drafts, engagement, brand kit, and audit trail. Standard tenancy controls apply (workspace + brand membership with owner/admin/member/viewer roles). SOC 2 Type II is in progress.

09 // SCHEDULE A DEMO

See your brand's sigil.

30-minute walkthrough. We'll show Sigil working with your real brand voice. No canned demo.

Insignia Labs · MMXXVI · prefer email? sales@insigniaconsultancy.com