LeGaxy · prepared for Carret Private Capital

Three weeks before he died, the succession plan was still being written.

Zong Qinghou, who built Wahaha from a school-run business he took over in 1987, died on 25 February 2024. The court record describes instructions for three offshore trusts of US$700 million each. They were not in place when he died.

Three plaintiffs later brought trust claims. A Hong Kong court preserved US$1.799 billion in a Jian Hao Ventures account at HSBC while Hangzhou decides the underlying claim. The court has not decided who owns the assets.

The arrangements were unfinished at death. That is the only point we take from it.

Jacky Zong & Ors v Kelly Fuli Zong & Anor [2025] HKCFI 3355; renewed leave to appeal refused [2026] HKCA 1292. Interim orders only; no merits decision as at 5 August 2026.

The quieter version

The firm knows the principal. It barely knows the next generation.

Mr Chan is 74, garment trims, production in Dongguan. In 2017 he named his younger son general manager in a staff memo. He never explained it to his daughter Ka-man, who works in London and believes it meant he did not trust her.

Raymond can see the structure. He cannot see the misunderstanding underneath it, and nothing in the file records it.

The intervention

One guided session surfaces the decision the file cannot.

Not a retreat, not a family constitution. One hour, in Cantonese, with Ka-man joining from London. Mr Chan's conditions: recorded consent from everyone present, and nothing enters the record until he has reviewed it.

At this point it is a candidate entry and nothing more.

Authority

Nothing becomes the family record until Mr Chan approves it.

The turn

A family question becomes a named decision.

Raymond now has a reason to call Ka-man that is not an investment pitch, and the firm has an open item with a named owner instead of an unspoken assumption.

The chain

The product makes every hand-off observable.

Nine steps, three phases. Every step has an owner, and the source is never overwritten.

The product

An iPhone in the room. A console for everything after.

Two surfaces, one record. The app captures under consent; the console is where the principal decides and where Carret's controls live.

The console

The screen your compliance officer will ask to see.

One workspace per firm, RM seats inside it. Every entry resolves to a timecode in an original recording, and every decision is attributed and logged.

The audit trail is the product. If an entry cannot be traced back to a recording and an approval, it does not belong in the record.

What we are building

Two interfaces, one governed record.

Proposed for Carret review. Hosting and integrations not yet selected, and nothing above is validated in production.

Where the machine stops

The system never decides family intent.

Governance

Controls required before any family data is collected.

The pilot

A Carret pilot creates the evidence for the next decision.

If Carret does not name an internal owner by the end of the pilot, the engagement closes and the data is returned and deleted under the agreed schedule. Everything on this slide is proposed for discussion.

Where we start

Start where the trusted relationship already exists.

Hong Kong family offices first. The institution buys, the relationship manager operates, the family benefits. We are not selling to families directly and we are not starting anywhere else.

Insurance agencies, Singapore, and philanthropically funded work with elderly people living alone are all plausible later segments. None is in the base case, and none is being built.

Team

Credible where the pilot is hardest.

Commercials

Revenue begins with one RM and one family.

Worked example — one RM's full book: 250,000 implementation (once) + 180,000 licence + (12 families × 60,000 = 720,000) = HK$1,150,000 in that firm's first year. Of the pilot-year 490,000, HK$180,000 is recurring software and HK$310,000 is one-off. Production hosting beyond the pilot is excluded and will be quoted once Carret selects the hosting arrangement.

Landscape

What the alternatives actually do.

Three named platforms, plus the two substitutes every family office really uses. We have marked ourselves honestly — most of our column is designed, not shipped.

Assessed from public product material in August 2026, not from hands-on use. We would rather you correct this table than accept it.

Three years

Six firms, ten seats, positive in year three.

Year 2 revenue is checkable: (4 seats × 180,000) + (2 new firms × 250,000) + (12 families × 60,000) + (3 reviews × 18,000) = 1,994,000. Year 2 costs less than year 1 because the build is paid for once. Cumulative net across the three years is about negative HK$1.75 million, so this path does not quite reach self-sufficiency on HK$2.7 million — we would rather show you that than curve-fit it. Philanthropically funded work is excluded from the base case entirely.

Use of funds

HK$2.7 million buys a shipped product and one validated deployment.

Individual compensation is available in diligence and is not shown here. This is capital sought, not a decision ready for signature — there is no instrument, valuation or governance term attached to it yet, and we would rather settle those separately than imply they are agreed. Everything on this slide is proposed for discussion.

What we are asking

Three decisions, kept separate.

We are not asking you to endorse a market story. We are asking you to identify where this workflow would fail inside a real private-bank relationship.

Kenta Ng · [email protected]  ·  Dhawal Doshi · [email protected]  ·  Bosco Ng · [email protected]