Asset-management content engineered around compliance constraints.

An asset management firm with active retail distribution across the US and UK generates a continuous stream of advertising across every channel — digital display, LinkedIn sponsored content, paid search, email campaigns, webinar promotions, quarterly fund commentary. Every piece goes through compliance review before it runs. In two jurisdictions. Without exception.

Investment advertising is regulated for a specific reason: people make real financial decisions based on it. Moving retirement savings, taking on risk they do not fully understand, investing money they cannot afford to lose. FINRA in the US and the FCA in the UK have each concluded that left unregulated, investment advertising will systematically mislead people by making investing sound safer and more profitable than it actually is.

In practice, this means four hard constraints that apply to every piece of content the firm produces:

• No implied guarantees.

  Phrases like “grow your wealth,” “reliable returns,” or “secure your financial future” imply a level of certainty about future outcomes that does not exist. Markets go down. Investments lose value. Content must reflect that honestly.

• Past performance disclaimers.

  Any reference to how a fund has performed historically requires a specific, prominently placed disclaimer that past performance does not guarantee future results. Burying it in small print does not satisfy the requirement.

• Risk disclosure prominence.

  Risk information cannot be an afterthought. The FCA evaluates the overall impression of a financial promotion — not just whether required language appears somewhere on the page. A piece that spends 90% of its length on upside and one line on risk fails the standard even if the risk line is technically accurate.

• No prohibited forward-looking language.

  Statements about what markets or funds will do are heavily restricted. The distinction between opinion, prediction, and guarantee is one regulators take seriously. Certain phrasings are effectively banned.

Writer — the firm’s existing content tool — is built to produce persuasive, engaging marketing language. In almost every industry, that is the goal. In investment advertising, persuasive and compliant are structurally in tension.

Asked to draft an ad for an equity fund, Writer produces something like:

That draft sounds professional. It fails compliance review comprehensively. “Build the future you deserve” implies a guaranteed outcome. “Consistently outperformed” is a past performance claim with no required disclaimer. “Strong returns” implies certainty of gain. “Growing your wealth” implies certainty of profit. There is no risk disclosure anywhere.

The compliance team does not add a disclaimer at the bottom and move on. They rebuild the piece structurally — removing implied guarantees, reframing the performance reference with required language, adding prominently placed risk disclosure, and ensuring the overall impression is balanced. What the marketing team wrote and what compliance approves are fundamentally different documents.

At a 65% first-draft rewrite rate, this is not an edge case. It is the norm. The compliance team spends most of its time performing structural copywriting on content that was never close to approvable.

The firm had years of approved advertising already in their archive. Every ad that passed compliance review is a labeled example of what compliant looks like for this specific firm, across this specific product range, in both jurisdictions.

That archive contains the practical knowledge of what the compliance team actually requires — more specific than reading the FINRA and FCA rulebooks, because it reflects how those rules are applied to this firm’s particular product mix and marketing style.

Coral Tree built a generation system trained on that archive through RAG. At generation time, the system retrieves the most relevant approved ads from the archive — those most similar to the current brief in terms of product type, channel, and audience — and uses them to anchor the generation. The model does not learn compliance in the abstract. It learns what compliance looks like in this firm’s voice, for this firm’s products, as validated by this firm’s own compliance team.

On top of the retrieval layer, a structured rules component encodes the hard requirements explicitly:

• →

  Past performance disclaimers are injected automatically on any piece that references historical returns

• →

  Prohibited phrase categories trigger flagging before the draft reaches the marketing team

• →

  Jurisdiction-specific requirements are applied separately to US and UK versions of the same campaign, reflecting the differences between FINRA and FCA standards

When the marketing team generates a display ad, the output already includes the past performance disclaimer, already uses risk-balanced language, and already avoids the phrasings the firm’s own compliance history shows get rejected. The compliance team receives a first draft that is structurally correct. Their job becomes editorial confirmation — not structural reconstruction.

The rewrite rate improvement from 65% to 18% is the headline. But what it represents operationally is the more important story. A compliance team that was spending the majority of its time rebuilding non-compliant drafts now spends the majority of its time doing what it exists to do — providing genuine regulatory oversight of content that arrives structurally sound.

The campaign cycle time follows directly. Content that previously cycled through multiple rounds of revision before reaching an approvable state now clears review in a fraction of the time. Campaigns that took weeks from brief to live now take days.

The system that produces these results is built on the firm’s own compliance history — which means it gets more accurate over time as that history grows, and it reflects this firm’s specific compliance standards rather than a generic interpretation of the rulebook.