Roadmap
GM3 is built in deliberate layers. Each one is useful on its own, each one makes the next possible, and nothing ships until its numbers can be trusted. Here's where we are and where it goes.
Why this order: mobile UA tooling is being rebuilt around AI-native workflows. Platforms are going headless. Marketers are moving into AI coding tools. GM3's roadmap follows a simple rule: first see everything, then act safely, then measure truthfully, then let intelligence compound. Trust is built in that order, and so is the product.
The unified read surface. One MCP connection in Claude Code or Cursor gives your team the whole ad stack, with numbers labeled by where they came from.
Blended spend, CPI, and campaign performance across every connected network, in your IDE, with per-network data freshness on every answer.
Every number tells you whether it's network-claimed, MMP-attributed, or SKAN-derived. No silent blending, no invented data.
When anything changes outside GM3 — a teammate in Ads Manager, a platform's automation — GM3 detects it, attributes it honestly, and tells you.
Every campaign, ad set, ad, and creative with current status — including the zero-spend ones other tools lose track of.
The full campaign lifecycle from your IDE: launch, creative, bids, budgets. Every action passes through deterministic governance before it touches a live network.
Create campaigns from typed, versionable campaign specs. Adjust bids and budgets, rotate and pause — without opening a dashboard.
Drop assets into your IDE and GM3 validates and prepares each one to every network's format and spec requirements — then keeps a versioned ledger of what ran where, so performance is always traceable to the exact creative.
Spend scopes, circuit breakers, dry-run simulation, and human approval gates. AI proposes; rules dispose; you approve. No model ever touches a budget on its own.
Real network connections (Meta first) replacing demo data — the same product, on your actual accounts.
Every read, proposal, approval, and change recorded. Your compliance story and your trust record, built in from the first action.
The layer that makes everything above trustworthy at scale: reconciling what networks claim against what your attribution actually shows.
MMP data, SKAN postbacks, and network-claimed numbers reconciled into one canonical ledger — with discrepancies surfaced, not hidden.
Continuous checks that GM3's numbers tie out with each platform's own reporting. The goal we hold ourselves to: 99%+.
Cohort-level performance and payback tracking that respects mobile's real complexity: SKAN grain, ATT splits, web-to-app flows.
Holdout design and readouts that tell you what your spend actually caused — not just what got claimed.
With visibility, safe execution, and truthful measurement in place, agents get useful — and get better the longer they work with your team.
Agents that watch your campaigns around the clock, spot fatigue and anomalies, and propose changes with the reasoning attached.
Custom agents shaped by your team's actual decisions, KPIs, and north-star metrics — proposing what your best operator would, with guardrails and your approval.
Beyond ad platforms: attribution, analytics, experimentation, engagement, and revenue tools joining the same governed surface.
Agentic breakdowns of why a creative performs — element-level analysis and plain-language recommendations. Platforms offer pieces of this for their own ads; GM3 brings it together across networks, in the workflow where you act on it.
Buying decisions informed by your entire mix, not any single platform's view of itself.
The long arc: as learned context deepens — your funnel, your KPIs, each platform's behavior — more of the UA loop runs itself, step by step, with your team always holding the dial.
What limits agents today isn't intelligence — it's context. As GM3 learns your team's workflow, your app's event funnel and growth KPIs, and how each platform's optimization actually behaves, agents can safely take on more. Guardrails always on; the dial always yours.
Every metric source-labeled, every day matched to your platform dashboards, every discrepancy surfaced. If we can't verify it, we say so.
GM3 lives where your team works — the IDE and AI workflow — not another dashboard to log into.
Design partners get direct founder access and shape this roadmap — the order above reflects what early teams tell us hurts most.
Join the design-partner beta