Skip to content
Relay
12 findings

Margaret Ellison, advisor (synthetic persona)Senior advisor, founder and executive practice, New YorkAbout $820M across 183 households; four relationships above $50M; 7 of those households are in this prototypeWho's who

Discovery

Opportunities found in what clients said, not in what the feeds show. Each cites its sentence. You accept it onto today's list.

Discovery · ran day 0, 06:40

I read 21 messages, notes and contact summaries for what clients said, not what the feeds show. 4 look like opportunities and wait on you, each cited to its sentence with a confidence. 1 have no citable document, so the evidence screen will refuse them.

  • ClientBrandvold: Retirement or pension decision, "Can we find time next week to talk through the pension transfer? Evenings UK tim..."
  • ClientThornbury: Company sale, vesting or funding round, "Discussed the company's late-stage funding round"
  • ClientBrandvold: Relocation or cross-border move, "Discussed a possible move back to the UK in two years"

Decide the first one

In production a model would read free text for these events; here a set of extractors does, and either way a person accepts each one.

Found, waiting on you

Retirement or pension decisionBrandvoldLife event80 percent

“Can we find time next week to talk through the pension transfer? Evenings UK time work best.”

Read from a message (msg-005), 4 days ago · Margaret Ellison, New York · A retirement date fixes the Longevity number and starts the withdrawal plan.

Would cite: Account review and documentation procedure 0.33

Company sale, vesting or funding roundThornburyOutside event70 percent

“Discussed the company's late-stage funding round”

Read from a contact (contactHistory[0]), 14 days ago · Margaret Ellison, New York · Vesting or a sale creates concentration or cash, and a tax bill, on a date.

Would cite: Managing single-name concentration 1.00Exchange fund, 7-year lockup 0.23

Relocation or cross-border moveBrandvoldLife event65 percent

“Discussed a possible move back to the UK in two years”

Read from a contact (contactHistory[0]), 40 days ago · Margaret Ellison, New York · A move across a border changes tax treatment of what is held.

Would cite: Non-US pooled funds held by US taxpayers 0.67

School or university feesRennerLife event60 percent

“Quarterly review; grandchildren's education trust raised for the first time”

Read from a contact (contactHistory[1]), 90 days ago · Margaret Ellison, New York · A fees commitment is a Legacy goal with a date on it.

No document in the corpus supports this kind of conversation. Accepted, it lands on the list refused until one exists.

By kind of event

  • Retirement or pension decision1
  • Relocation or cross-border move1
  • Company sale, vesting or funding round1
  • School or university fees1

Where they were read

  • message1
  • note0
  • contact3
  • request0

A colleague's note or a contact summary is read at ten points less confidence than the client's own words.

What is autonomous here, and what a model would do
Reading is autonomous: 9 extractors over every captured message, team note and contact summary, each a pattern with a confidence, each result cited to the sentence and the record it came from, each matched to the corpus documents that would support the conversation. Adding an event kind is one entry in the extractor list. In production a typed extraction model would do the reading and read it better; it would still not put a candidate on today's list, and a candidate with no citable document would still arrive saying so.