Skip to main content
Pin the scope — In every question below, name the entity and the source-of-truth tables. A plausible answer from the wrong (summary) table is worse than no answer — if two sources could answer, run both and let your SME rule which is truth.
Run each of these against your validated warehouse. For every answer, note that Ana cites the governed surface and shows the rendered SQL. The values below are the starter’s pinned golden values from validation/golden-queries.md — verified against a synthetic wealth warehouse on Databricks (horizon 2026-05-31, window 2025-06-01…2026-05-31). Yours will differ; the point is that the same call gives the same number every time.

​
5.1 · AUM (the firm-wide number)

Prompt
You’ll see: AUM from the account_value series at one valuation date (golden: **60,372,722,173∗∗across∗∗102,340∗∗openaccounts),notasumofpositionsnapshots—andtheaverage−12mbasis(≈60,372,722,173** across **102,340** open accounts), not a sum of position snapshots — and the average-12m basis (≈58.6B) used for billing. point_in_time ≠ average_12m; the surface pins both (notes/aum-definition.md).

​
5.2 · Net flows (organic growth)

Prompt
You’ll see: the organic-growth number isolated from market beta (golden: net **−1,672,009,009∗∗=in1,672,009,009** = in 648,148,543 + out −$2,320,157,552), sourced from account_value.net_external_flow (notes/aum-definition.md).

​
5.3 · Return (TWR vs MWR)

Prompt
You’ll see: the GIPS-standard TWR (golden: +10.15%), with client flows geometrically removed, and MWR (Modified Dietz) exposed explicitly — never silently swapped. They answer different questions: manager skill vs. the client’s dollar experience (notes/return-definition.md).

​
5.4 · Allocation and concentration

Prompt
You’ll see: allocation weights that sum to 1.000 (golden: US_EQUITY 0.4436 · FIXED_INCOME 0.3446 · INTL_EQUITY 0.1571 · CASH 0.0547), and a concentration breach list (golden: 201,120 holdings >10% across 60,701 accounts, max weight 0.8601) — both point-in-time, both vs. the account denominator.
Know what concentration means here — The governed surface flags a holding’s weight in its own account above a threshold (default 10%), at one snapshot. scope="issuer" combines all share classes of one issuer — the more conservative view. Whether diversified funds are looked-through or exempt is a decision you pin (notes/concentration-definition.md); on the synthetic set, issuer == position because dim_security has no issuer grain.

​
5.5 · Effective fee rate (revenue quality)

Prompt
You’ll see: realized yield as fee revenue / average AUM × 10,000 (golden: 306,644,253/306,644,253 / 58,582,267,503 = 52.34 bps), with the billing-basis average AUM denominator — not point-in-time (notes/fee-definition.md).
Why everyone gets the same number — Metrics like AUM, return, and fee yield can be computed several ways (point-in-time vs average AUM, TWR vs MWR, gross vs net). The ontology pins one governed definition — with the decision recorded in ontology/notes/ — so Finance, the CIO office, and the advisors stop disagreeing.

​
5.6 · When the answer isn’t governed yet — watch the model grow

Now ask something from your shortlist that the starter doesn’t already cover. This is the important beat: a starter pack is a head start, not the finished model. (Even the synthetic dataset has honest gaps — no benchmark-return series, and undated closures — recorded openly in golden-queries.md and schema-mapping.md.)
Prompt
You’ll see: Ana explore only the frontier (not re-derive the whole warehouse), answer, and propose a write-back — a new metric committed to your repo with provenance. Review and merge it, and the next person who asks gets the governed answer for free. That’s the malleable loop: the ontology you ship is the one you grow, and it gets more complete every time you use it.
You ratify; high-stakes definitions get review — Ana proposes; humans ratify via normal git review. Anything in governance-mnpi-pii.md scope or a core metric (AUM, return, a benchmark assignment) should require review before merge — see STANDARDS.md. The point isn’t to let an agent rewrite your model unsupervised; it’s that discovered knowledge is captured instead of re-discovered next time.

​
✅ Checkpoint

  • All five metric families answered through governed surfaces, SQL shown
  • You can point to the notes file explaining at least one metric’s definition decision
  • Ana proposed a write-back for a not-yet-governed question — and you saw it land as a PR