“Spend limits, approval gates, and an audit trail on every write.”
That sentence was on Synter’s homepage on August 26 — the first time in fifteen weeks of tracking that any competitor had used Ott’s exact compliance vocabulary on a product surface. We verified it live, wrote the framework for what a real audit trail requires, and scored their claim against it.
Nine days later, it was gone.
Verified live this week: Synter’s MCP page now contains zero governance language — no “audit trail,” no “spend limits,” no “approval gates,” no FAQ line, nothing. The homepage was redesigned around a new hero (“The AI Agent Operator for Ads”), and the audit-trail claim was re-scoped from “on every write” to “on every major platform.”
Two things are true at once. The market now sees that “audit trail” is the feature to fight over — and borrowing the words was never the same as shipping the product.
The shortest compliance lifecycle on record
Here’s the timeline, every date verified:
- Aug 22 — “Spend limits, approval gates, and an audit trail on every write” appears on Synter’s site. First tracked competitor use of Ott’s moat vocabulary.
- Aug 24 — the weekly competitive brief flags it CRITICAL.
- Aug 26 — we verify it live across four surfaces (homepage hero, homepage feature list, MCP page FAQ, pricing row) and publish the framework post: what a real audit trail requires.
- Aug 31 — the walk-back:
/mcpno longer contains “audit trail,” “spend limits,” or “on every write” anywhere in served HTML. - Sep 1 — zero governance terms on the MCP page. Homepage re-scoped. Pricing row unchanged.
Adopted, promoted, verified, and deleted — nine days, start to finish. That’s not a product iteration cycle. That’s a content cycle.
What the redesign kept, what it cut
| Surface | Aug 26 (verified) | Sep 1 (verified) |
|---|---|---|
| Homepage hero | “READ ACCESS FREE · 21+ platforms · Spend limits · Audit trail” | “The AI Agent Operator for Ads.” |
| Homepage feature list | “Spend limits, approval gates, and an audit trail on every write” | “Spend limits, approvals, and an audit trail — on every major platform” |
| Homepage audit section | — | “Audit log — Every action is recorded — who, what, when, on which account” |
| MCP page | FAQ: “Synter executes, with approval gates on every change” | Zero governance language |
| Pricing table | “Audit Logs & Governance”: 7-day / 90-day / Unlimited-SIEM | Same row, untouched |
Read the table top to bottom. Every product surface that carried the compliance claim was scrubbed or re-scoped. The one place the words survive is the pricing table — where “Audit Logs & Governance” is a paid-tier feature with retention limits: 7-day history on the entry plan, 90 days on the mid tier, “Unlimited / SIEM Export” on the sales-led plan.
An audit trail that’s a pricing tier isn’t an audit trail. It’s an upsell. And when the redesign came, the upsell survived while the claim didn’t.
“On every write” became “on every major platform”
The re-scope is the tell. “An audit trail on every write” is a per-action promise: every campaign change, every budget edit, every AI-agent write gets recorded, attributable, and exportable. “An audit trail on every major platform” is a coverage claim: the platform family has the feature somewhere. Those are different promises — and the weaker one is the one that survived.
Score the surviving language against the six properties of a real audit trail: actor attribution, immutability, before/after values, independent storage, jurisdiction awareness, retention and export.
The new homepage says “every action is recorded — who, what, when, on which account.” That’s a logging intent, not a trail. No immutability claim. No before/after values. No jurisdiction awareness. No export path. And the pricing page still tells the real retention story: seven days of history on the entry plan.
Seven days of history is not record-keeping. It’s browser history.
Why the words were always the easy part
The walk-back made visible what the Aug 26 post documented: the claim had no fixed address and no documentation behind it.
- No audit-trail documentation ever shipped. No spec, no compliance docs, no jurisdiction feature. The claim lived in hero copy and a pricing row.
- Zero finance or regulated-vertical content across the site’s 237 blog posts — down from 243 in late August, with nothing new published since the Prospector post on Aug 25.
- Prospector “GA” was announced via blog post —
/prospectorstill returns 404. Product pages come after press releases, or not at all.
A compliance claim with no documentation, no vertical content, and no product page is a sentence looking for a product. When the redesign came, it was the first thing cut.
The regulator’s bar doesn’t do redesigns
The week the words disappeared, the FCA published research showing 44% of young investors believe AI-generated financial information is regulated. It isn’t — the FCA explicitly warns general-purpose AI chatbots are unregulated. And 38% said they’d invest based on AI output alone.
That’s the demand side of the accountability standard the FCA has been setting all year:
- Rathi (June 24): “Accountability for regulated activities and outcomes must remain clear.”
- AAMP 2.3 (July 30): agentic autonomy “that respects a hard approval boundary is a critical requirement for regulated advertisers.”
- FCA Handbook API (Aug 6): the FAQ admits the API “does not provide past versions of Handbook content.” The regulator ships current rules; the historical record of what your agency actually did is your job.
Regulators don’t check whether your marketing page still says the words. They ask what ran, who approved it, and what you can export when they come calling. A claim that survives a content refresh was never the deliverable — the record was.
Four questions that would have caught this in August
- Is the audit-trail claim on a product surface or in hero copy? (It was in the hero.)
- Is it infrastructure or a pricing tier? (It was a pricing row with 7-day retention.)
- Is the promise per-action or per-platform? (It said “every write” — then quietly became “every major platform.”)
- Can you export the record, and does it name the actor, the before/after, and the jurisdiction? (No spec ever shipped.)
Run those four questions against any vendor’s compliance claim before you trust it. The words are cheap. The record is the product.
What Ott ships
Ott answers all four questions with shipped infrastructure:
- Activity Logging — timestamped, actor-attributed (humans, tools, and AI agents by name), before/after values, stored independently of Meta, jurisdiction-tagged, exportable in about 90 seconds.
- Agency Hierarchy — client → brand → account, with FCA, MiCA, ESMA, and Alberta jurisdiction tags per client.
- Budget Ledger — prepaid client balances, real-time deduction, overdraft alerts, and approval gates on spend changes.
- Flat pricing — $29–$199/mo, no per-client or spend-based fees. Compliance infrastructure, not a tier you unlock by booking a demo.
Synter said “an audit trail on every write” for nine days. Then a content refresh deleted it.
Saying it was always the easy part.