× Midas Analytics
01 / 18
FO × Midas Analytics
Foundation · Phase 1 Build · Proposal v2
v2 · revised on your feedback · 5 Aug 2026

A tech-first family office,
built on one source of truth.

We start where everything else stands: a central, cloud-based data layer that turns your custodians into one place you can see — and slice — your entire book. Get that right, and every vertical becomes a module on top.

12 weeks 5 → 15 custodians built on v2 · 5 Aug 2026
Michele De Filippo
Prepared by Michele De Filippo, PhD
AI Lead · Midas Analytics
What changed since v1
your feedback, folded in

Four updates — and one yes.

Everything marked with the v2 pill in this deck is new or updated. In short:

Data lives in Singapore
Hosted in AWS Singapore (ap-southeast-1), per your preference — locked across the design.
The data-infrastructure diagram you asked for
How data is organised — and exactly how the dashboards retrieve it.
Your dashboard list — mapped Day 1 / Day 2
All six views from your email, item by item. Blue = Day 2 — your own marking.
Pricing & timeline, updated for the fuller scope
12 weeks (was 10) and a revised fixed fee — Day-2 items gated so you only commit to what's defined.
And yes to the working session — let's walk Day 1 vs Day 2 together this week. Last slide books it.
Where this is heading

One platform. Five verticals.
Built over time.

The destination is a single, scalable platform that spans the family office. We don't build all of it at once — we build the layer that lets us add each piece as a module.

Client profiling
Portfolio insights
Fund management
Client reporting
KYC
This is my understanding — correct me where I'm wrong.
The objective · why foundation-first

Get the data foundation right — and every vertical becomes a module on top.

Every stalled data project we've seen starts the same way — teams rush to build use cases before the foundation exists. So it wobbles, then it falls. We build the base first; everything else stands on it.

vs Modules first No shared base — silos, rework, then it topples. DATA FOUNDATION Foundation first One base — every module stands on it.
The foundation · scope of this first block

Three things this block delivers.

01 · Source of truth
All data
Land all custodian data into a secure raw database on the cloud — one clean place it all flows into.
02 · #1 priority
Visibility
The moment data lands, you can slice and dice it freely — see the whole book, on day one.
03 · First AI layer
Ask it
A simple AI assistant — ask questions about your data in plain language and get clear answers back.
Proposed setup · built on · Singapore region
v2 · Singapore

From custodian feeds to insight — one pipeline.

01
Custodian feeds
Your custodians and banks — 5 to start, scaling to ~15.
02
Secure raw store
All feeds land untouched — the raw source of truth.
03
Curated layer
Cleaned, modelled, reconciled — trusted and query-ready.
04
Custom dashboard
Designed & owned by us, tailored to how you work — not off-the-shelf BI.
05
AI layer
Sits on top of trusted data — questions, checks, prediction.
Step 1 / 5
Custodian feeds
Your custodians and banks — 5 to start, scaling to ~15. Every position and transaction flows in, automatically.
The architecture · live data flow
v2 · Singapore

How your data actually moves.

Your custodians The data foundation aws Your family office Native expansion — 5 vertical modules on top profiling · insights · funds · reporting · KYC Dashboard Custom · visibility 4 Your team Sees it all AI layer Ask · check · predict Answers Alerts · memos 5 Bank A Bank B Custodian C + up to 15 Ingest 1 Raw store Source of truth 2 Curated Clean · modelled 3

Custodian feeds land raw, get curated once, then surface two ways — a dashboard your team reads and an AI layer your team asks. All on Singapore (ap-southeast-1), built to add custodians and modules without re-architecting. Visibility — the dashboard surfaces performance vs. benchmarks, SIPO risk & breach alerts, dividends & upcoming earnings, and a view per custodian. The whole book, the moment data lands. AI layer — sits on the curated data: flags portfolios in / out of line with each mandate, answers plain-language questions, and examines price distributions. The answers flow straight to your team.

Data infrastructure · how it's organised
new in v2 · your question 2

How data is organised — and how dashboards get it.

Sources
Custodians ×5 → ~15Addepar or direct feeds
Market dataprices · FX · benchmarks · CA
House dataclients · products · business · reg.
Raw · Singapore
Landing zoneuntouched · immutable
Audit trailwhat arrived, when, from whom
Curation · daily
Pipelinevalidate · reconcile · master · CA · price checks
Exceptions queueheld for human review
Curated + computed
Five domainsClients · Products · Portfolio · Business · Reg.
Calculation enginereturns · risk · analytics — hover for the full list
Serving API → consumers
Serving APIone access path · fully logged
Dashboardsevery view, from the API
AI layersame numbers, asked in English
Partner APIsper-partner, least-privilege
hover any node for the detail — including exactly what the calculation engine computes
One rule holds the design together: every consumer — dashboard, AI, partner — reads curated & computed data through the serving API. One source of truth, one access path, fully logged.
Trust & security
v2 · Singapore locked

How your data stays yours.

A family office runs on confidentiality. The foundation is built so your data is protected by design — and never leaves your control.

Built on
Encrypted in transit & at rest, hosted in AWS Singapore (ap-southeast-1) — your preference, locked in. Your data never leaves the region.
Read-only — credentials you hold
We only read, on credentials you control and can revoke. We never trade or move money.
Least-privilege access
Scoped to this engagement, limited to two selected Midas team members.
NDA first, your policy throughout
An NDA before any data reaches us — aligned to your own infosec policy at kickoff.
Midas is ISO 27001-certified
ISO/IEC 27001:2022 · view PDF
What "visibility" looks like — day one

See the whole book, the moment data lands.

Performance vs. benchmarks
Where you're out- or under-performing, at a glance.
Risk & SIPO monitoring
Allocation bands, targets, concentration limits — with alerts.
Dividends & earnings
Income flows, plus upcoming earnings to watch.
A view per custodian
Drill into each custodian / bank on its own.
Risk, against your SIPO

Your investment policy, watched automatically.

Your SIPO — Statement of Investment Policy and Objectives — sets the rules. Parameters are yours; we set them together. Here are four live examples of what the monitoring catches as the book moves; step through them.

1 2 3 4 Example scenarios — press → to step through
Asset allocation — Equities52%
0%band 35–60%100%
Single position limit6.5%
0%cap 10%15%
Counterparty / bank limit22%
0%cap 30%50%
Return & volatilityon target
vol targetwithin tolerancereturn target
All within policy — every limit is green; the book sits comfortably inside your SIPO bands.
Example 1 / 4
Visibility, made real

See it working — not a screenshot.

We've already built this for another family office. Click through to a live example you can explore yourself — it's the real product, running on entirely made-up data.

Click here
Open the live dashboard interactive example · all data fictional · opens in a new tab
Performance vs. benchmark SIPO risk & alerts Dividends & earnings Per-custodian view
Your dashboards · client portfolio views · 1 of 2
new in v2 · your list

Your list, mapped — exactly as you marked it.

Day 1 — in this buildDay 2 (blue) — next phase, your markingtbd — we define it together at the working session

Client portfolio · high-level view

Asset classes
EquityFixed IncomeCommodityPrivate CreditPrivate EquityHedge FundsStructured ProductsCash · true vs MMOptions
Geographies
EM / DMNA · Europe · APACCountries
Plus
FXBenchmarks tbd / base ccyIndustries tbdLiquidity: daily → >1 quarter

Client portfolio · performance

Abs & % change · MoM · QoQ · YoY · YTDSince inceptionRolling 12 monthsDividend / yieldOut / under vs benchmarksFX contributionCosts: mgmt + performance feesSharpe ratio

Individual investment performance

Abs & % · per asset class / geographySince inclusion · rolling 12mMoM · QoQ · YoY · YTDOut / under vs benchmarksCosts: mgmt + performance feesReturn standard deviation
Your dashboards · risk & firm views · 2 of 2
new in v2 · your list

Risk & firm-level — the full picture.

Day 1 — in this buildDay 2 (blue) — next phase, your marking

Portfolio risk analysis

Over / under-weight vs benchmarksConcentration vs threshold tbdCorrelation · classes / geos / industriesBetaFactor: Value · Momentum · GrowthMarket cap: L / M / SMax drawdown / time to recoveryStress · replays: 2008 · Mar-20 · 2022 · Aug-24Liquidity: time to full unwindVaR · scenario simulationsHypothetical stress tbdLock-ups / notice vs cash needs tbdFI: duration · mod. duration · DV01 · YTM/YTW

Firm-level view

AUMRevenues / yieldNet new money · MoM → YTDConcentration: counterparty / issuer / underlyingReturn dispersion · similar mandates

Counterparty & custodian

Concentration per counterparty / custodianStructured-product issuer riskDeposits & cash placement
Day-2 items are scoped & priced at the working session — you only commit to what's defined.
Where AI comes in · once the foundation is live

Ask your book a question. Get an answer.

what happened overnight? press → to send
Midas AI
Beyond the dashboard

Post-earnings memos, on tap.

Another deliverable we've produced for a family office: concise memos that turn an earnings print into a decision. The same data layer feeds them.

Post-earnings memo
AURORA GLOBAL GROWTH FUND · Q3 FY25
GENERATED · T+0 · 06:14
The print

Q3 revenue $2.41bn, +14% YoY — a 3% beat. EPS $1.32 vs $1.18 est. Operating margin +180bps to 24.6%. FY guidance raised to high-teens growth.

What it means for the book

The position is 6.3% of the equity sleeve — a top-5 holding. The beat strengthens the 12-month thesis; quality and momentum screens both improve. No mandate or concentration limit breached.

Action

Hold. Trim only if it approaches the 10% single-position cap. Watch FX — ~38% of revenue is EUR-denominated.

Illustrative example · fake data — your real memos drop in here.
The plan
v2 · 12 weeks for the fuller scope

Day 84: the foundation is live — with your Day-1 dashboards on top.

Workstream
W0W1W2W3W4W5W6W7W8W9W10W11W12
Week 0 · scoping & prep
Scope
Custodian connectivity
5 feeds → ingestion
Raw + curated layers
Raw store
Curated layer
Custom dashboard
Core visibility
Full Day-1 suite
AI layer
AI assistant · mandate checks
~Week 7 — visibility liveWeek 12 — full Day-1 suite live
What we'll need from you, to start

A short list to get moving.

An NDA
We need to know your business better in order to help — the NDA lets us in.
Read access to your custodian feeds
Read-only, on credentials you hold and can revoke — exact mechanics TBC at kickoff. Nothing leaves your control.
Your SIPO parameters
The bands and limits we'll monitor against — so alerts reflect your policy.
A few hours of your team
To walk us through how you work, so the dashboard fits you — not the reverse.
The investment
v2 · repriced for your Day-1 scope

One fixed price — now covering your full Day-1 list.

12-week build · foundation + Day-1 suite
USD 48,500
Fixed fee · USD 29,500 foundation (unchanged from v1) + USD 19,000 for the extended Day-1 dashboard suite from your list.
Effective cost after HK R&D credit≈ USD 24,500
You recover≈ 50%
Then — running & maintenance
AWS Singapore hosting, monitoring, upkeep & support — you recover about half here too. Market-data subscriptions passed through at cost (depends on benchmark choices).
USD 920 /mo
≈ USD 460 /mo effective

300% enhanced R&D deduction via Midas as a certified DLRI — next slide. Fees shown are placeholders to confirm.

What's included
  • Your custodian feeds wired into Singapore — direct or via Addepar
  • A secure raw store + a curated layer across your five data domains
  • The full Day-1 dashboard suite — all six views from your list
  • A first AI layer — ask your data questions
  • Risk & SIPO monitoring with portfolio-level alerts
Day-2 (blue) items — options, VaR, hypothetical stress, FI analytics, liquidity ladder, industries, dispersion — are not in this fee. Scoped & priced at the working session, so you only commit to what's defined.

You own your data and every curated layer we produce; export terms are set in the SOW.

Financial incentive · HKSAR R&D
v2 · math updated to the new fee

A 300% R&D deduction — recover about half.

Building with Midas qualifies for Hong Kong's 300% enhanced R&D deduction — bringing the effective cost of the foundation down to roughly half the fee.

Effective cost after the credit
≈ USD 24,500
From a USD 48,500 fee — you recover ≈ USD 24,000 (~50%) through the HK R&D credit.
Build fee · foundation + Day-1 suiteHKD 378,300 (USD 48,500)
300% tax deductionHKD 1,134,900
Tax saving @ 16.5%− HKD 187,300
Effective cost≈ HKD 191,000 (USD 24,500)

Midas is a certified DLRI (No. D041). 300% on the first HKD 2M; 200% above — confirm filing with your tax advisor. Source: IRD (Cap. 112). Fee is a placeholder.

Certified DLRI · No. D041 — click to enlarge
Get started
v2 · the session you proposed

Three steps to kickoff — starting this week.

Step 01

Working session — this week

Call or face-to-face, as you suggested: we walk your Day-1 / Day-2 list, confirm feasibility item by item, and capture what's still to be defined.

Step 02

NDA & agreement

We draft and circulate the NDA and a simple statement of work. You sign — about a week.

Step 03

Kickoff & Week 0

Scoping, data-access requests, and the foundation starts building.

Thank you.

midasanalytics.ai
navigate · space reveal