From a live product
to paying customers.
A grounded plan to take SALVO — already running in production — from “no way to charge” to first paid customers with agentic ad management. Built on a full codebase audit and current market & pricing research.
A production-grade engine with no meter on it — yet.
The thesis. A real, live, multi-tenant SaaS with deep Google Ads + Meta integrations and a differentiated AI agent — far more mature than most pre-revenue products. The path to revenue is commercial plumbing and trust, not more ad features.
The reality check. Two hard blockers: no billing system at all, and the flagship launch paths have never run against a real ad account. Both fixable in a quarter — both must close before we market.
A steady market meets a hyper-growth wave.
- Agentic AI is the wave. ~$7B in 2025 → $33–57B by 2030–31 (30–46% CAGR); AI ad-ops startups raised ~$700M in 2025.
- Marketing automation is the floor. A durable double-digit grower — up to $81B by 2030 (11.5% CAGR).
- AI commands a premium. AI-native tools price 20–37% above peers; credit pricing grew 126% YoY.
- SALVO's lane: an agentic, multi-platform ad manager — exactly what buyers are funding.
Sources: MarketsandMarkets · Mordor · Grand View · Adweek · Tropic (2025–26).
Not a prototype. A live, operated SaaS.
Real infrastructure
GKE, Cloud SQL (private VPC), Memorystore, 18 Terraform files, WIF auth, CI/CD with migrate-on-deploy. Prod + dev, both live.
Mature SaaS backbone
Redis-session auth, orgs → projects → members, RBAC, org/user-scoped connections, audit logging, AES-256-GCM, Semgrep CI.
Differentiated AI agent
Mastra + Vertex orchestrator, Ed25519 tokens, per-run budgets, draft→approve flow, live chat & streaming. Production-grade.
Deep on two platforms. Connect-only on five.
Bar = real campaign-launch depth. Tier-2 platforms connect via OAuth but have no launch wizard.
- Google + Meta carry launch wizards, audiences, conversions, reporting & experiments.
- The other five do basic CRUD only — creative upload throws “not implemented.”
- So “multi-platform” is today a two-platform reality.
Decision: fund tier-2 depth, or launch as “Google + Meta, done exceptionally well.”
A guarded agent: propose → human approves → apply.
- Mastra agent on Vertex Gemini 2.5 Pro (one-file swap to Claude).
- Security: Ed25519 tokens, per-project flag + kill switch, allowlisted tools, per-run budgets, circuit breaker, GDPR redaction.
- Google-Ads-only, single-project, propose-only.
- No cross-platform tools or autonomous optimization loops — that's the post-GA growth track, not a rebuild.
How SALVO works right now.
Requests flow left → right. The agent acts on the user's behalf through the same guarded adapters — every write still goes via the API's ownership & permission checks.
Breadth + an agent + prediction — in one place.
| Capability | Point tools Madgicx, Revealbot |
Enterprise suites Smartly, Skai |
SALVO |
|---|---|---|---|
| Manage Google + Meta + 5 more from one place | ~ | ✓ | ✓ |
| Agentic assistant — it proposes, you approve | ✗ | ~ | ✓ |
| SMB-friendly entry price | ✓ | ✗ | ✓ |
| One-prompt ad launch — AI builds the creative | ✗ | ✗ | soon |
| AI success prediction before you spend | ✗ | ~ | soon |
| Learning brain across your campaigns | ✗ | ~ | soon |
✓ full · ~ partial · ✗ none · “soon” = on the SALVO roadmap (see the Vision section).
~55–65% to GA — but very unevenly.
Foundation
Infra, auth, multi-tenancy, agent security, Google/Meta depth. The hard, slow part — largely done.
Trust & compliance
Security strong, but GDPR self-service erasure & real-account validation are incomplete.
Commercial layer
Billing, plan enforcement, metering, multi-platform breadth — mostly missing.
Lopsided in a good way: what remains is faster-to-build commercial scaffolding, not the expensive foundation.
There is no way to take money today.
- Zero Stripe (or any processor) anywhere in the codebase.
- No billing models — no subscription, plan, invoice, quota, or entitlement among 43 models.
- The billing page is a hardcoded mock: a static “Free Plan,” usage bars pinned to “0,” and an “Upgrade” button with no action.
- No paywall, no metering, no plan-gating across 149 procedures.
Build from scratch: data model → Stripe → checkout → webhooks → lifecycle → entitlements → metering → billing UI.
What stands between us and a paying customer.
| # | Gap | State today | Effort | Gates paid GA? |
|---|---|---|---|---|
| 1 | Billing & monetization — Stripe, subscriptions, checkout, webhooks, UI | None — greenfield | L | Hard blocker |
| 2 | Real-account validation of all Google + Meta launch paths | Mock-tested only | M | Hard blocker |
| 3 | Plan-tier entitlements & usage metering across procedures | None enforced | M | Required |
| 4 | GDPR self-service erasure + 90-day purge job | Export only; delete owner-only | M | Enterprise gate |
| 5 | Tier-2 platform depth — or scope-down | Connect-only, no wizard | M–L | Scope decision |
| 6 | Meta new-surface test coverage (video, lead forms, pixels) | Shipped with zero tests | S–M | Risk |
| 7 | Agent breadth + autonomy (Meta, cross-platform, loops) | Google-only, propose-only | M–L | Post-GA growth |
Effort: S ≈ days · M ≈ 2–4 weeks · L ≈ 4–6+ weeks (small team, AI-assisted velocity).
The market prices on spend tiers — $39 to $35K+/yr.
| Tool | Category | Entry | Mid | Top / Enterprise | Model |
|---|---|---|---|---|---|
| Madgicx | AI Meta/Google opt. | ~$39–55 | by spend | $30K+ spend | % of ad spend |
| Revealbot | Rule automation | $49 | $99–299 | $899–2,999 | Spend-banded flat |
| Optmyzr | PPC management | $249 | $499 | Custom | Spend-banded flat |
| AdEspresso | SMB Meta/Google | $49 | $99 | ~$499 | Spend-banded flat |
| Adzooma | SMB/agency PPC | Free | ~$69 | ~$179 | Flat, spend-agnostic |
| AdCreative.ai | AI creative gen | $39 | $249 | $599+ | Flat + credits |
| Smartly.io | Enterprise creative+media | ~$4–5K/mo | — | Custom | % of media spend |
| Skai | Enterprise omnichannel | ~$95K/yr | ~$210K/yr | $420K+/yr | Annual contract |
Public/calculator-gated pricing, May 2026. Skai figures are analyst estimates. Sources: vendor pages, Vendr, G2, First Page Sage.
Spend-banded tiers win. Pure % loses.
- Spend-banded flat tiers (winner). Flat price per tier, capped by managed spend. Predictable, scales with value, self-segments SMB→enterprise.
- Pure % of spend (avoid). 1–3% invites resistance, unpredictable bills, hard to meter across 6 APIs — feels like an agency fee.
- AI credits (layer on top). Meter the heavy autonomous actions — protects 50–60% AI margins, creates expansion.
Three tiers, anchored to the comparable set.
- Up to ~$10K/mo ad spend
- Google Ads + Meta
- 1 seat · core AI
- Bundled AI credits
- Up to ~$100K/mo ad spend
- All platforms · multi-account
- Multi-seat · branded reports
- Agentic automation · big credit pool
- $500K+/mo ad spend
- SSO · SOC 2 · audit · API
- Dedicated support
- Custom credits · SLAs
Enterprise starts below Smartly.io and far under Skai — we undercut incumbents while capturing enterprise budgets. ~20% annual-prepay discount.
Spend-banded subscription + metered AI credits.
How a customer's bill is built — a predictable subscription spine, plus usage-priced credits on the heavy autonomous actions:
A credible path to ~$150K MRR within ~18 months.
- Q3 '26 — paid GA opens with design partners; prove the launch + billing loop.
- Q4 '26 — Growth tier becomes the workhorse as agencies adopt; ~$12K MRR.
- 2027 — breadth + agent autonomy unlock upmarket; first enterprise logos.
Blended ARPU climbs ~$130 → ~$280/mo as the mix shifts up. Conservative vs the 30%+ card-trial benchmark.
Five focused workstreams, run in parallel.
Lock decisions & foundations
Scope lock (Google + Meta), billing data-model, validation harness, spend-capped sandbox account.
Monetization core L
Stripe, subscription & plan models, checkout, webhooks, lifecycle, real billing UI replacing the mock.
Entitlements & metering M
Plan-tier gating across 149 procedures, spend-band tracking, AI-credit metering & balances, overages.
Trust & validation M
Validate all launch paths on a real account, Meta test coverage, GDPR self-service erasure + purge job.
Launch readiness M
Pricing page, card-required trial, billing QA, security pack, private beta with design partners.
Paid GA → marketing
Open paid signups, then turn on marketing. Post-GA: agent breadth toward full agentic.
Parallel tracks compress this into ~14 weeks.
Run sequentially these gaps take ~5+ months; parallelizing billing, validation & compliance lands paid GA near week 13–14.
First paid in ~Q3 2026. Full agentic, Q1 2027.
Be precise: first paying customers get a strong assisted agent; the autonomous, cross-platform vision lands ~Q1 2027 — sequenced after revenue.
What must be true before a single ad runs.
- Billing works end-to-end — a real card buys a real plan; webhooks update state.
- Plans are enforced — advertised limits gated server-side.
- Every launch path validated live — all 6 Google types + Meta on a real, spend-capped account.
- GDPR self-service erasure + purge shipped — needed for enterprise reviews.
- Meta new surface tested — video, lead forms, pixels before paid traffic.
- Scope is honest — market “Google + Meta” unless tier-2 ships.
- Trial → paid loop — card-required trial, onboarding, dunning.
- Support & status — incident path, status page, billing inbox.
Non-negotiable: a bad live launch spends a customer's real budget incorrectly. Real-account validation is a launch gate, not a nice-to-have.
Card-required trial for SMB, demo-led for enterprise.
Self-serve · card-required trial (30%+ vs 9–18% no-card). 14–30 day cycle.
Sales-assisted. 45–90 day cycle. Win on multi-account, branded reporting & agentic automation.
Demo-led, annual. 120–210 day cycle. SOC 2 / GDPR readiness is the gate.
What could bite — and how we de-risk it.
| Risk | Why it matters | Mitigation |
|---|---|---|
| Live launch spends budget wrong | Flagship paths never hit a real account | Spend-capped sandbox; validate every path in Phase 3; gate GA on it |
| Billing slips (greenfield L) | It gates 100% of revenue | Start Phase 1 day one; use Stripe Billing primitives; thin first cut |
| Multi-platform overclaim | 5 of 7 platforms can't launch | Market “Google + Meta”; ship tier-2 wizards post-GA before claiming breadth |
| AI inference margin | 50–60% GM vs 80–90% classic SaaS | Meter heavy agent actions as credits; bounded bundled allowance |
| Enterprise compliance gate | GDPR erasure/purge not shipped | Ship erasure + 90-day purge in Phase 3; prep SOC 2 pack |
| Small-team bandwidth | Parallel tracks need focus | Hard scope lock; defer agent autonomy & tier-2 depth to post-GA |
From assisted today to one-prompt, self-improving ads.
- Google + Meta, deep
- Assisted AI agent (propose → approve)
- 149 procedures in production
- Billing in build → paid GA
- Cross-platform agent (Meta + more)
- AI success prediction on drafts
- Learning brain v1
- Tier-2 platform launch wizards
- In-platform AI video + image generation
- True one-prompt ad launch
- Autonomous optimization loops
- Brain that learns across the network
Vision / roadmap — sequenced after first revenue, not before it.
New AI services wrap the same guarded core.
Highlighted = new AI layer. The Learning Brain turns every live campaign into sharper predictions & suggestions — a compounding data moat single-network tools can't match. Vision — not yet built.
Score an ad before you spend a cent.
AI verdict on your draft
- ✓ Strong hook & offer match
- ! Add a clear call-to-action
- ! Narrow the audience ~20% to cut waste
- ✓ Budget pacing looks healthy
🧠 A brain that learns from every campaign
Draft → predict → fix → launch. The more campaigns SALVO runs, the better the predictions — a data moat single-network tools can't match.
Three decisions unlock
the 14-week clock.
Approve “Google + Meta” for launch; defer tier-2 & agent autonomy.
Approve $59 / $249 / from $1.5K tiers + AI-credit metering.
Greenlight Phase 0–1 now (scope lock, sandbox, start billing).
Bottom line: SALVO is operationally far ahead of most pre-revenue products. The work to first revenue is commercial plumbing & validation — a focused quarter. First paying customers: ~late August 2026.