GO: SERVICE MODE, DEFER STANDALONE SAASComposite: high confidence, Mode AMode B SaaS: 6-12 months out

The agentic website management platform that ships work, not just findings

Monitor, propose, human approves, ship, report. One governed loop across SEO, AEO, CRO and maintenance, fed by data AC already owns, and already proven live on a real client site before this pitch was written.

25
Real findings from one live scan: GSC + GA4 + RankPrompt fused
23
Auto-created as approval tasks in the agency PM system, same run
1
Finding shipped through the full gate: preview deploy + QA to an open PR, same day
$1B
Profound's Feb 2026 valuation, proof the monitoring-only category is real but incomplete
1DiscoverN/A
2ValidatePASS
3FinancialGO
4Pitchthis page
5Namingpending Josie
6BrandingN/A
7BuildPhase 0 next
8LaunchN/A
Executive Summary

Three findings drive the verdict

The problem: monitors tell you, executors fix blind

AEO monitors surface what is wrong in ChatGPT and Perplexity and stop there. SEO executors ship fixes with no visibility into AEO, CRO, or which pages actually matter to ad spend. Nobody combines monitoring and execution and reporting in one retainer.

The mechanic: one loop, five stages, a human at the gate

Monitor, propose, human approves, ship, report, then back to monitor. RankPrompt supplies the AEO intelligence layer no competitor owns outright. Every shipped change passes a named human strategist first, on every tier, with no exceptions.

The moat: proven live, before the pitch, at zero marginal cost

This is not a projection. The loop already ran on nationalmarketingawards.com: 25 real findings, 23 auto-tasked, one shipped through the full gate to an open PR the same day, inside AC's existing stack.

Market Landscape

Two camps exist. Neither does what this does.

AEO monitors tell you what is wrong and stop. AI SEO executors fix a narrow lane blind to AEO, CRO, or ad-spend priority. The whitespace sits exactly on AC's existing book: between $999/mo self-serve tools and $5K+/mo human agencies.

PlatformFocusPricingNoteThe gap
ProfoundAEO / LLM-visibility monitoring, no execution$399+/mo Growth; Enterprise $2K-5K+/mo$1B valuation, Feb 2026 ($96M Series C)Tells you what is wrong in ChatGPT and Perplexity. Fixes nothing itself.
Peec AIAEO monitoring across LLM engines$89-499/mo + $35-165/engine add-onPer-engine meteringBill creep as engines are added; monitoring only, no execution arm
OTTO / SearchAtlasAutonomous SEO execution at scale$99-999/mo50,000+ sites deployedExecutes SEO fixes blind to AEO visibility, CRO, or ad-spend priority
Alli AIAI-driven SEO execution tool$169-1,249/mo (Agency: $599-699 for 15 sites)Execution-only pricingNo monitoring, no human-reviewed approval workflow, no reporting layer
AgencyAnalyticsClient reporting dashboards~$12-18/client blended, per-managed-objectReporting-only categoryReports what already happened. Ships nothing itself.
$100-500/mo
WordPress care plans
$1.5K-15K/mo
SEO retainers (SMB to mid-market)
$2K-15K/mo
CRO retainers
$1.5K-10K/mo
Content retainers

Traditional retainer benchmarks, cited for anchoring context. AC already sells $1.5K-5K/mo retainers today (Clotheshorse: $3.5K/mo).

Annual cost at scale: monitors and executors vs. the combined loop (illustrative, USD)

Figures at annualized list pricing (Profound Growth, Peec AI mid-tier, Alli AI Agency) versus Praetor Maintain and Grow. The point is not that Praetor is cheapest line item by line item; it is that a buyer stacking a monitor plus an executor plus a reporting tool pays more, in more places, for less coordination.

How It Works

One governed loop, five stages

Every stage already exists in AC's MCP stack except the action adapters and the proposal plumbing. The genuinely new build is small; the intelligence layer is already built.

1

Monitor

GSC + GA4 + ads data + RankPrompt fused weekly into the AEO Priority Matrix (Defend, Optimize, Create, Refresh, Amplify) plus a CRO and Core Web Vitals scan.

2

Propose

Findings become a branded monthly strategy doc and tasks in the agency PM system, not raw dashboard noise. Every client-facing item runs through stop-slop before it ships.

3

Human Approves

A named strategist reviews every proposed change before it ships. No direct-to-prod, ever, regardless of client tier or urgency.

4

Ship

Branch or draft-first execution through the CMS adapter (Astro, WordPress, Shopify, Webflow), with a visual QA gate at full resolution before merge.

5

Report

A weekly plain-language change log plus a quarterly RankPrompt QBR closes the loop and feeds directly into next week's monitor pass.

Universal guardrails, regardless of CMS: draft or branch-first execution, never direct-to-prod; a visual QA gate on every change at full resolution; a named human approver per client; scoped low-privilege credentials only, never admin accounts.

MVP Proof

Not a mockup. This already ran, on a real client site.

Live on nationalmarketingawards.com, 2026-07-31. One scan pulled GSC, GA4, and RankPrompt data together, produced 25 real findings, and 23 of them were auto-created as approval tasks in the agency PM system. One finding was executed end to end by an agent, through every gate, the same day.

25
Real findings produced by one scan: GSC + GA4 + RankPrompt fused on a live client site
23
Auto-created as approval tasks in the agency PM system, same run
1
Finding executed by an agent through the gates: preview deploy + QA screenshots to an open PR, same day
0
Direct-to-prod changes. Every shipped edit passed a named human approver first

One scan, one funnel

What's next

  • A RankPrompt-styled dashboard is planned, running locally, surfacing the three kill-gate metrics from day one
  • Weekly delta-scan job (Hermes cron) to make this a standing cadence, not a one-off demo
  • Second pilot on a Shopify site (Clotheshorse) to prove the adapter pattern generalizes beyond one CMS
  • Every number above is from the actual run. Nothing in this section is projected.
Buyers

ICP priority stack and the jobs this replaces

Mode A launches inside the existing AC book. Mode B, a standalone SaaS product, is deliberately deferred until the service motion proves out.

Mode A: service, now

TIER 1: CONVERT

Existing AC clients

Upsell trigger: "are we in ChatGPT?" Low-CAC, already trust the relationship.

TIER 2: WARM

Warm lookalikes

Sourced through AC's existing pipeline. Evidence-led, referenceable proof points.

TIER 3: DISPLACEMENT

Vendors stacking a monitor, an executor, and a dev retainer

The consolidation pitch: one loop, one bill, one accountable strategist.

Mode B: standalone SaaS, deferred

First ICP: 5-30 person agencies

The white-label buyer. Deliberately not pursued until Mode A clears its own kill gates; revisit at the phase-2 timing gate, roughly 2-3 quarters of finding-rate and adapter-durability data from now.

Jobs to be done

1

When my brand is queried inside ChatGPT/Perplexity/AI Overviews and I don't know what it says, I want to see exactly how I'm represented, so I can correct wrong info before a prospect acts on it.

2

When my site underperforms and I have no developer, I want the fix actually shipped, not a 40-item PDF, so visibility recovers without a dev queue.

3

When I approve spend, I want ads, GSC, and AI-visibility trends in one place, so I can judge ROI without reconciling four dashboards.

4

When an agent proposes a change, I want to approve or reject it before it's live, so I don't wake up to a broken page or off-brand copy.

5

When I justify this retainer upward, I want a monthly narrative of found/fixed/lifted, so I can prove ROI without doing the analysis myself.

6

When a competitor starts appearing in AI answers where I used to be cited, I want an immediate alert, so I can respond before losing the deal.

7

When I change CMS or add a channel, I want the same system to keep working, so I'm not re-buying tooling every stack change.

Defensibility

The combination is the moat. No single layer is.

Rated honestly, not optimistically. Two of three layers are converging toward commodity within a year; the value is in how they are governed together, not in owning any one of them.

REAL, NARROWING

RankPrompt ownership

A structural advantage today: we own the AEO intelligence layer end to end instead of licensing it. The category is crowding fast, with Profound at a $1B valuation and five more monitors priced under $500/mo.

TABLE STAKES

Cross-channel data + human gate

GSC, GA4, ads data and AEO fused with a named human approver is genuinely differentiated right now. Every serious competitor is converging on this exact shape within a year, so it will not stay differentiated on its own.

COMMODITIZING FASTEST

Per-CMS adapters

WordPress core's Abilities API 6.9 plus mcp-adapter, Wix Aria, and Shopify Sidekick are all racing toward native agent-callable CMS control. The adapter layer is the fastest-commoditizing piece of the stack and will not stay ours alone.

The honest read: no competitor today closes monitor, propose, human-approve, ship, report in one governed loop. That combination, not any single layer inside it, is what has to stay ahead. Per-CMS adapters will commoditize fastest and should be treated as a race, not a wall.

Pricing

Flat, capacity-bundled tiers for Mode A

Shipped changes bundled per tier, visible on the card as the auditable capacity metric. Add-on packs instead of metering. Mode B pricing exists but is intentionally not launching yet.

Maintain

$895/mo
5 shipped changes/mo bundled
  • Weekly maintenance and technical SEO fixes
  • Monthly RankPrompt AEO snapshot
  • Content strategy report only
  • Human review-gate approval on every change
Most common tier

Grow

$2,495/mo
15 shipped changes/mo bundled
  • Weekly RankPrompt AEO monitoring across 4 LLMs
  • Monthly content strategy plus 2-4 new pages/posts
  • 1 CRO experiment running
  • Ads-data read-only insights informing priority
  • Review gate plus monthly strategy call

Dominate

$4,750/mo
40 shipped changes/mo bundled
  • Continuous maintenance and technical SEO
  • Weekly RankPrompt plus citation opportunities worked
  • Monthly content strategy plus 4-8 new pages/posts
  • 2-4 concurrent CRO experiments
  • Full cross-channel prioritization plus monthly media review
  • Named strategist and quarterly RankPrompt QBR

Add-on packs, never metered per action

Overage pack
+$150 per 5 shipped changes beyond the bundled tier
Extra page bundle
+$500 for 4 additional pages in a given month
Extra CRO experiment
+$350/mo per additional concurrent experiment
Additional site
+40% of the base tier price per extra managed site
AEO hardening
$750 one-time: llms.txt plus structured-data foundation
Astro migration project
$5K-15K one-time, feeds into the cheapest-to-operate playbook

Mode B: future standalone SaaS

DEFERRED 6-12 MONTHS
TierPriceNote
Starter$149/siteor $349 for 3 sites
Growth$899 base+$60/site beyond 5
Scale$2,499 base+$45/site beyond 15
White-label$500-1,500 platform+$35-50/site wholesale, for agency resellers
AEO engine add-on$35-165/enginethe one defensible usage-based meter in the whole model

Flat beats metered

Cursor's credit-billing backlash forced a public CEO apology. Copilot's AI Credits produced documented bill-shock complaints. 78% of IT leaders report surprise consumption charges from usage-based AI tools. Notion's flat $10/seat model sees 65% adoption with no comparable backlash. Agent work is never usage-metered here.

The AI label costs money

A July 2026 SegmentOS pricing experiment measured roughly a 25% willingness-to-pay drop when a price card led with "AI-powered." 36% of consumers report penalizing a brand specifically for AI use. The word AI stays out of every price card. Tier names are capability words: Maintain, Grow, Dominate, never Autopilot or Agent.

Naming

Praetor recommended, Josie picks

33 candidates screened across four lanes (neuro, astro, navigation, mythology). Domain reality was the harshest filter: every bare top-10 name across .com/.ai/.io is squatted except these three finalists.

Praetor

RECOMMENDED
Weighted score: 7.3/10Family fit: 8/10
getpraetor.com open, $12.99

A Roman magistrate who reviews and rules before action, the human-approval gate encoded directly into the name. Risk: Praetorian is an established cybersecurity brand, but it targets a different buyer, so confusion risk reads low. Heavy gaming-lore usage of the word (Warhammer, Halo) cuts both ways: memorable, but noisy in search.

Ganglion

FALLBACK
Weighted score: 6.7/10Family fit: 7/10
getganglion.com, ganglionhq.com, getganglia.com all open

The most literal sibling to Cortex: a nerve cluster coordinating distributed signal. Best domain picture of the three finalists. Ding: the "ganglion cyst" medical association, and it is a harder name to spell cold. Promote this if the Praetor gaming-lore association tests badly.

Argus

REJECTED
Weighted score: 6.25/10Family fit: 6/10
Every viable domain squatted

The hundred-eyed watcher: a self-explaining monitoring metaphor. Killed on domain reality alone, and it sits in an already crowded adjacent monitoring/infosec namespace.

Family context: Praetor would sit beside Cortex (social listening) and Nebulous (agency management) in the AC product family: a single-word, classical-and-neuro naming register that reads as one family across the roster. Register getpraetor.com immediately on selection; also check praetor.ac, since AC already controls short domains on that TLD.

Stress Test

Shark Tank: five challenges, five answers

The exec panel challenged the pitch adversarially. Verdicts recorded verbatim.

S
Skeptical CFOFinance shark

Challenge: "What's cost-to-serve per client-month once RankPrompt and LLM costs scale with monitoring volume?"

Response: Monitoring cost is fixed per brand, not per query; execution work reduces AC's own billable hours.

Verdict: WOUNDED (plausible but unproven; no live cost-tracking data yet).

A
Agency-Owner BuyerBuyer objection

Challenge: "Why not just add AI-visibility tracking to my existing vendor?"

Response: The differentiation is the closed loop: cross-channel proposal plus shipped fixes with a human gate, not another report.

Verdict: SURVIVES, but only if the sales demo shows a merged PR/live change, not just a dashboard.

A
AI-Commoditization BearCompetitive shark

Challenge: "OTTO SEO already auto-deploys fixes at 50,000 sites with no developer. What stops this being a feature, not a company?"

Response: OTTO deploys via a shallow JS pixel layer; the defensible position is depth (real CMS writes, real ads changes, human sign-off).

Verdict: WOUNDED (12-18 month window before the quality gap closes).

P
Platform-Risk BearThe dangerous shark

Challenge: "WordPress standardized agent read/write access in core. Wix Aria and Shopify Sidekick already manage sites natively, bundled with hosting. Why pay you once the platform does this?"

Response: Platform-native agents optimize for generic best practice and platform engagement, not a specific brand's competitive position or ad efficiency, and have no visibility into accounts the platform doesn't own.

Verdict: WOUNDED-TO-FATAL depending on execution speed (the dangerous shark).

C
Churn / Retention SharkRetention shark

Challenge: "Once you fix the obvious backlog in month one, what do you sell in month six?"

Response: Monitoring is inherently continuous (competitor moves, LLM answer drift, algorithm updates), converting from fix-backlog to defend-position.

Verdict: WOUNDED (unproven whether the agent generates believable new findings past month three).

Kill Gates

Three named failure modes, instrumented as dashboard metrics

Each gate is a live number on the planned SiteOps dashboard, not a hope. Failing a gate is a success of the process, not a failure of the idea.

Finding-rate plateau

Dashboard metric: findings per client per month

If fewer than 3-5 new findings surface per client per month by month 3-4, the loop has run out of real work and is manufacturing busywork to justify the fee. This is a service redesign trigger, tracked from week one, not a lagging indicator discovered in a QBR.

The "my website builder does this" objection

Dashboard metric: objection log per sales call

As CMS-native agent features mature across Wix Aria, Shopify Sidekick, and WordPress's Abilities API, this objection appearing in real calls is the earliest read on how fast the wedge is closing. Logged per call, reviewed monthly.

Trust-breaching shipped error

Dashboard metric: incident log, zero-tolerance threshold

Any agent-shipped change that breaks a live client site or damages trust ends the autonomous-ship claim immediately, regardless of how many other gates passed. One incident is enough to force a full review of the ship pipeline before any client resumes.

Timing and Roadmap

Service now. Standalone SaaS at the proof gate.

Mode A launches immediately, absorbed into the existing AC book at effectively zero marginal cost. Mode B is deliberately parked until roughly 2-3 quarters of finding-rate and adapter-durability data justify the standalone build.

Phase 0
1-2 weeks

Wire the loop

  • 2 pilots: one git-native site, plus Clotheshorse (Shopify, freshly closed, no paid media yet)
  • RankPrompt monitor-pack, dry-run preflight first
  • Weekly delta-scan job (Hermes cron) writing findings to PM tasks
Phase 1
Weeks 2-4

Intelligence to proposals

  • Monthly proposal generator: priority matrix + citations + CRO ideas, through stop-slop
  • Weekly change-log email template, the "someone is managing this" artifact
Phase 2
Weeks 3-6, overlapping

Action adapters

  • Astro/Netlify adapter first, the reference implementation
  • Shopify custom-app adapter: GraphQL mutations + theme-draft flow
  • Approval flow wired into the PM system; cost-routed via OmniRoute
Phase 3
Weeks 6-10

Productize

  • WordPress adapter (REST + app passwords) and Webflow adapter (CMS/meta scope)
  • Pricing page + Stripe checkout, cloned from the Clotheshorse flow
  • Convert pilots to paid tiers, then open to the rest of the book

Standing-rule compliance baked in: ads data is analysis, never execution, without explicit authorization. Subscription-first, no silent paid spend. Client copy runs through stop-slop. Anomaly alerts validate against baseline before firing. Per-client scoped credentials only, never admin accounts.