Recommendations that reach the client only once the adviser approves them.

For an advisory firm, we built a system that reads client records, check-ins and activity, flags patterns for the adviser, and keeps every recommendation under their judgement.

Client
Advisory firm
The work
Client signals to recommendations
What we built
Rules, a weekly summary and an approval step
Status
Documented system

230 recommendations reviewed in the first month.

From the system’s review records for its first month in use.

The briefSupport more clients in the same review time, and see warning signs sooner.

The adviser’s review time set the limit on how many clients could be supported. The aim was to support more clients within that time and bring warning signs to the adviser earlier.

The advisory firm supplies client records, check-ins and activity information for review. Whatever the system found, the adviser had to stay responsible for what the client was told.

What we builtRules flag the patterns, and the model drafts the summary.

We built a state engine: a set of fixed rules that looks across client records, check-ins and activity and flags patterns in each review area. A model uses those flags to draft a weekly summary for the adviser. Recommendations then pass an approval step before any client sees them.

From client records to approved recommendations.

From client records to approved recommendations.Client records, check-ins and activity feed a rule-based state engine. Its flags supply a model-drafted weekly summary. An adviser approval gate stands between the summary and recommendations released to the client.
From client records to approved recommendations.Client records, check-ins and activity feed a rule-based state engine. Its flags supply a model-drafted weekly summary. An adviser approval gate stands between the summary and recommendations released to the client.
Key
The system in words

Client records, check-ins and activity feed a rule-based state engine. Its flags supply a model-drafted weekly summary. An adviser approval gate stands between the summary and recommendations released to the client.

Select a stop to see what it does.The adviser approval step stands between the summary and the client.
  1. Source records

    Client records, check-ins and activity information.

  2. State engine

    Rules assess each review area and raise flags when they see a pattern.

  3. Weekly summary

    A model drafts a weekly summary for the adviser from those flags.

  4. Adviser approval

    The adviser approves each recommendation before the client sees it.

  5. Approved recommendations

    The client sees only the recommendations the adviser approved.

The adviser approves what reaches the client.

Professional judgement has a defined place in the system, and it sits right before the client.

  • The rules produce the pattern flags. The model only writes the summary.
  • Recommendations need the adviser’s approval before clients see them.
  • The adviser stays responsible for every recommendation released through that step.

Who does what.

The study documents the pattern flags, the weekly summaries and the adviser’s approval. Each part has one job.

The rules
Assess each review area and raise pattern flags.
The model
Drafts the weekly summary from the flags.
The adviser
Approves recommendations and is responsible for their release.
The client
Sees approved recommendations only.

Automation that does the reading, with the adviser keeping the decision.

If your team reviews the same kinds of record every week, rules and a drafted summary can bring the important ones forward, with a person approving what goes out.

Next study Session preparation from one client record Read the study