×

Discussion Document · For Brainstorming · 14 May 2026 · Strictly Private

Building HK's AI-native
finance talent — together.

A three-track exploration: where Midas Analytics and DMO can build a system that upskills finance professionals, generates leads for both sides, and compounds with every cohort.

Prepared for

Ravi Dillon
James Oshea
Sujan Melwani
DMO Group

Prepared by

Michele De Filippo, PhD
CEO, Midas Analytics

Format

Working document

Status

Not a proposal — yet

AI theoretical capability vs observed usage across job categories

Anthropic Economic Index · 2025 · radar chart

§ 01 The opportunity at scale

The market is massive — and barely tapped.

Blue area: what AI could already do across job categories today.
Red area: what people are actually using it for.

Business & Finance

~95% vs < 15%

Capability vs observed usage today.

Legal · Office & Admin · Education

90%+ vs < 20%

Same story across knowledge-heavy verticals.

What this means

Finance sits at the bullseye. Highest theoretical reach. Lowest observed usage. The runway for AI upskilling is years long.

Global AI literacy distribution — 84% have never used AI

Each dot = ~3.2M people · 2,500 dots = 8.1B humans · Feb 2026

§ 02 The bar today

AI literacy is on the floor.

8.1 billion people. The fraction who pay for AI is rounding-error small. The fraction who use it in real work is smaller still.

~6.8B (84%) have never used AI

~1.3B (16%) use a free chatbot occasionally

~15–25M (0.3%) pay USD 20/mo for AI

~2–5M (0.04%) use AI in actual coding / workflow

And in HK finance?

Even worse. Senior analysts walking into interviews having never opened Claude. Hiring managers asking AI-fluency questions to candidates who don't know what an MCP is.

Received by us on 13 May 2026

A real inbound to Midas Analytics from a HK professional

Real inbound · Midas Analytics · HK

§ 03 The signal from the street

HK Market is screaming for help.

An unsolicited inbound that landed in our inbox this week — and it isn't the first.

DMO sees this from the hiring side. Strong analyst CVs. Falling short on AI fluency in the interview room — and losing offers because of it.

Midas sees the same gap from the training side. Senior operators across asset managers, banks, and corporates asking how to upskill teams fast.

Same gap. Two vantage points. Worth combining.

Working hypothesis

DMO + Midas close the loop on AI talent in HK finance — from hiring to training and back.

§ 04 The shared opportunity

What each side brings to the table.

Press → to reveal — DMO first, then Midas, then the joint product.

DMO brings

20 yrs in HK finance hiring
Deep candidate flow
Trust at career inflection points
Live signal on hiring-manager asks

AI curriculum tied to finance use cases
Engineering depth — agents, dashboards
HK-based · DLRI-certified (D041)

+

Joint product

Diagnose the AI gap
Close it with fit-for-purpose programs
Place upskilled talent back into market
Sell upstream to whole teams

§ 05 The three tracks at a glance

One funnel, three doors.

Three offerings, each a different entry point into the same talent system. Same colors used across the deck.

"I want this 1-on-1"
← drops back to Track 01

Track	Who it's for	Format	Indicative ticket	Revenue model
Track 01 Career Reset	Individual analysts who didn't close the offer	1-on-1, ~6 weeks	Mid · HKD	Candidate-paid + revenue share with DMO
Track 02 Open Workshops	Anyone in HK finance, open enrolment	2–3 hour public sessions, 20–40 people	HKD 2k–3k / seat	Per-seat + lead-gen for Track 3
Track 03 Enterprise Programs	Whole teams inside asset managers, banks, corporates	Tailored, 4-week cohort	High · per-team	Direct fee + referral kick-back to DMO

Loved the 1-on-1?
Brings Midas into the org →

Workshop landed →
"do this for my team"

§ 06 Track 01 — Career Reset

Don't lose the candidate after a "no". Reset them.

Today, a strong analyst who fails an interview disappears from the loop. Track 01 turns that exit into the start of a longer relationship — for both DMO and the candidate.

Step 01

Interviews via DMO

Analyst goes through DMO's standard hiring loop.

Owned by DMO

→

Step 02

Doesn't get the role

AI gap surfaces in the room. Offer slips.

DMO sees this

→

Step 03

DMO debriefs & routes to Midas

Candidate isn't lost — they're handed off.

+

Joint hand-off

→

Step 04

Midas tailored 1-on-1 reset

~6 weeks. Each candidate ships an AI artefact tied to the target role.

Delivered by Midas

→

Step 05

Re-enters the market — stronger

Back into DMO's pipeline. New interviews. Higher hit rate.

DMO places them

Loops back · stronger candidate, next round

Why it works · #1

Revenue stays inside the partnership

Candidate-paid fee, split between Midas and DMO. DMO keeps the relationship, monetizes the failed interview, and books the future placement fee.

Why it works · #2

Candidate is the buyer

Career-driven, motivated, willing to invest. They're also the easiest audience to close — the gap was just named on the debrief call.

Why it works · #3

Data dividend

Every reset surfaces what hiring managers are actually testing for. That signal feeds Tracks 02 and 03 — and DMO's own hiring intel.

§ 07 Track 02 — Open Workshops

Land low, expand high.

A small ticket gets people in the door. The room is the lead-gen for Track 03.

Top of funnel

Public session · 2–3 hours · 20–40 attendees

HKD 2k–3k ticket. Co-branded invite from DMO + Midas. Topic-driven (one finance workflow per session).

Value lands in the room

Hands-on with a real finance use case

Attendees leave with a working AI artefact tied to their job — and the recognition that they could use this on their team.

Conversion · A

"Do this for my team"

Senior people in the room ask Midas in. Lead handed to Track 03 Enterprise.

Conversion · B

"Sign me up for the next one"

Returning attendee. Compounds the brand. Feeds the warm-list for the next round.

We've done this before

Midas Analytics running a workshop at the Italian Chamber of Commerce, Hong Kong

Italian Chamber of Commerce · Hong Kong · AI workshop run by Midas

§ 08 Track 03 — Enterprise Programs

Whole teams, four weeks, real outputs.

Mini case: an asset manager wants their investment, risk, compliance, marketing, and BD teams AI-fluent — without the disruption of a re-org.

Week 1

Foundations

Map the team's top 3 workflows. Identify where AI gives leverage, where it doesn't.

MilestoneWorkflow audit + AI-fit scorecard, per role.

Week 2

Hands-on

Build the first agent or automation on the highest-leverage workflow, live with the team.

MilestoneOne shipped agent, owned by the team.

Week 3

Cross-function

Connect outputs across investment / risk / compliance / marketing / BD. AI that knows the org, not just the desk.

MilestoneCross-functional brief, signed off by leads.

Week 4

Embed

Adoption plan, internal champion identified, scoring of the next wave of workflows.

MilestoneRoadmap for the next 6 months, signed.

Cross-functional audience

Investment Risk Compliance Marketing Business Development …and more

Commercials: per-team pricing. Working-hours or evening cohorts. Scope of customization decided at kickoff.

DMO's role: warm intros + co-sell — DMO is already inside these institutions on the hiring side. Referral kick-back or a co-branded engagement.

§ 09 The flywheel

Three doors, one system.

Each track feeds the other two. This is what makes the partnership compound rather than just a menu of services.

Track 01 → candidate-side data that sharpens Tracks 02 and 03.
Track 02 → fills the room with future Track 03 buyers.
Track 03 → produces alumni — future Track 02 attendees and DMO placements.

Track 01

Career Reset

Track 02

Open Workshops

Track 03

Enterprise

Track 01 data
sharpens curriculum

Workshop attendees
convert to Enterprise buyers

Enterprise alumni
become DMO placements

§ 10 A pilot to start with

Move fast. 3 weeks. Two options.

The window to own AI upskilling in HK finance is open today and won't stay open. Two viable starting points — Track 03 (Enterprise) follows naturally as the lead-gen output of either.

Track 01

Career Reset · 3 candidates · 3 weeks

Wk 1

Wk 2

Wk 3

Wk 1: DMO routes 3 candidates who recently missed offers. Midas runs intake.
Wk 2: Tailored 1-on-1 reset. Each candidate ships an AI artefact tied to their target role.
Wk 3: Re-interview, joint review, decision on scaling.

Why: lowest activation cost. Candidate-paid. The data feeds Tracks 02 and 03.

Track 02

Open Workshop · 1 session · 3 weeks to ship

Wk 1

Wk 2

Wk 3

Wk 1: Topic locked, venue booked, co-branded invite from DMO + Midas.
Wk 2: Audience curation (DMO's book) + content build (Midas).
Wk 3: Run the session. Convert leads into Track 03 pipeline.

Why: wider reach. Brand signal. Plants the seeds for Track 03 in the same room.

Either path lights up Track 03 (Enterprise) afterwards — that's the premium service, sold off the leads these two pilots generate.

§ 11 Financial incentive

Engaging Midas cuts cost in half.

Phase 1 of any track qualifies for the HKSAR Inland Revenue Department's enhanced R&D tax deduction. Midas Analytics is a certified Designated Local Research Institution (DLRI) — No. D041.

0% tax deduction on qualifying R&D expenditure under the Inland Revenue Ordinance, Cap. 112.

Path A · Candidate pays Midas

Candidate-paid → 50% effective discount

Candidate pays Midas directly for Track 01 / Track 02 / Track 03. Midas issues the DLRI invoice. Candidate (or their employing entity) claims the R&D deduction at filing — effective cost roughly halves.

Path B · DMO pays Midas

DMO-paid → 50% off for DMO

DMO pays Midas for the training, bundling it into the candidate package or selling it on as a service. DMO claims the R&D deduction on the Midas invoice — effective cost to DMO roughly halves.

Why this works

Midas is a certified DLRI (No. D041) — Computer & Information Sciences. The training and tailored upskilling work qualifies as R&D under Cap. 112.

The math (per HKD 100k spend)

HKD 100,000 invoice → HKD 300,000 deduction → HKD 49,500 tax saving @ 16.5% profits tax. Effective cost: HKD ~50,500.

A working document.

Everything in here is up for debate. The goal of this deck is a shared starting point for the next conversation, not a proposal. Push back, edit, co-own.

midasanalytics.ai