AdPilot — Reconsidered

What the 2026 market evidence changed in our design — and what we do next.

evidence: 4 research files · 30+ sources
The one-line verdict

Control over black-box AI is the product. Our typed-approval layer already is that layer — and no analyst names an SMB agentic ads operator anywhere.

What changed

4 beliefs flipped · old on top, evidence below
FLIP 01 Creative
We thought

AI creative quality retains users

Now

Accountability retains users

Icon.com pivoted from AI admaker to $1,000 human UGC; IAB: 37% fear audience distrust of AI ads.

Icon.com pivot IAB · 37% distrust
Do human-review tier in paid plans
FLIP 02 Competition
We thought

Platform automation is the competition

Now

Platform automation created our market

Every leader shipped default-on AI, then retrofitted controls after backlash — AI Max auto-upgrade, Advantage+ ROAS drops.

Google AI Max Meta Advantage+
Do approval-first is the pitch, not a feature
FLIP 03 Category
We thought

Analysts will validate the category

Now

The SMB agentic operator is unclaimed

Forrester renamed it Revenue Marketing and named only enterprise leaders — Salesforce, Adobe, HubSpot.

Forrester Revenue Marketing
Do define the category ourselves
FLIP 04 Acquisition
We thought

Reporting is table stakes

Now

Reporting is the acquisition hook

It is the #1 most-installed marketing workflow — 3,800+ n8n templates, and the reason people install anything at all.

n8n · 3,800+ templates
Do templates gallery first

The reconsidered sequence

6 steps · effort S/M/L · 2 shipped, 1 in flight
01

Templates gallery, incl. reporting

Most-installed workflow pattern — Optmyzr rules-library.

Shipped S
02

URL → campaign onboarding

Creatify / Meta lowest-friction first run — paste a link, get a campaign.

Shipped M
03

Cross-platform digest

Demand cluster #2 — Motion's measure-first.

Now M
04

Policy pre-check + appeal drafter

Loudest unserved need in the official forums.

Next M
05

Creative scorecard + make-next

Motion raised $30M on this layer.

Next M
06

Human-review tier

Icon lesson + IAB trust stat — the retention layer.

Later L

Where we deliberately say no

3 standing refusals
Not doing

No default-on automation

We are the control layer. Every leader shipped default-on AI first, then had to retrofit controls after the backlash.

Instead Approve-before-spend on anything that touches budget.
Not doing

No HN-first launch

Our buyers are owners and operators, not developers. A developer-audience spike would misprice the whole category.

Instead Reddit and LinkedIn, where the demand threads already are.
Not doing

No seat pricing

A seat prices the tool. Our buyer is paying for an outcome, and per-seat math makes the agent look like software to license.

Instead Outcome framing: spend managed, not logins sold.