Raul LeeBHASKARAN

Raul Lee Bhaskaran · Advocate & solicitor. Founder. Producer.

Most things get built in one lane.
I build between the lanes.

This is the operator door. raul.my is where you buy or hire me; this is where you build alongside me. I open one or two of my builds a year to someone who carries the half I don't: distribution, capital, a route to market. If that's you, the two doors below are the audition: build with me, or take a project I've opened up and run. Same operator, same estate; narrow on purpose.

See open projects
7
plays I'd run if there were more of me
hook + revenue
on the table
mechanism
redacted until you're in

01 / The thesis

One bet. Many proofs.

Not a scatter of side projects; a single thesis with a working build engine behind it. AI-native products at the seams of law, education, and culture. Each of these is a door. Walk through whichever one fits.

AI
Proxi
AI WhatsApp operator for businesses
Education
Saraf
Autism education for late-diagnosed adults
Legaltech
MatSelamat
AI court-document generator
Legaltech
Setara
Legal protection plan for Malaysians
Product
HEXACO Protocol
Personality assessment + AI oracle
Civic
ALIM
Art & Law Initiative
Music
Cliff Coffin
Dark techno / production
Advisory
Consult
Work with me directly

Why I give the ideas away

Ideas are cheap. I share them on purpose.

Chances are you won't do anything with them; almost no one does. Execution is the moat, and it's wider than people like to admit. So take the idea, it's yours. The only thing worth protecting is the conversation with the person who actually ships. That sits behind the one filter that means anything: skin in the game. If that's you, come through the door, not the inbox.

02 / Open projects

Biz plans in a box.

Real plays I'd run if I had more of me. The hook, the wedge, and a conservative revenue read are on the table. The mechanism, the numbers, and the people stay redacted until you're actually in.

Ledger
MVP-readyWeeksNear-zero capital

Done-for-you, compliance-safe content for Malaysian regulated professionals.

Advocates, licensed financial planners and doctors can't market like normal businesses; the Bar's advertising rules, the SC, and the MMC are strict and almost nobody understands them. Compliant content is a moat, not a commodity, and the rules change slowly enough to encode once.

Conservative: RM 8k-17.5k MRR at a RM 390/mo retainer, 20-45 professionals.
  • + Legal grounding (RLB Advocates)
  • + A working voice-locked content pipeline
  • + Proof-of-concept running on Raul's own channels
Redacted until you're in
  • The compliance rule-set encoded into the generator
  • The retainer ladder above RM 390
  • The Bar-Council-safe positioning that keeps clients clear of trouble
Wayfinder
ValidatedMonthsModerate capital

White-label AR heritage trails, sold to Malaysian state and municipal tourism bodies.

States have the heritage and the budgets; they lack the tech and a credible builder. Niantic open-sourced 8th Wall, so the cost floor for web-AR just collapsed. First mover writes the template every other state copies.

Conservative: RM 80k-115k per deployment; 3-4 realistic state/municipal targets inside 24 months.
  • + A heritage catalogue already in hand
  • + A working multi-tenant AR/AI engine
  • + A credible government-facing route to the boards
Redacted until you're in
  • The per-deployment pricing build-up
  • The channel partner introducing the boards
  • The post-8th-Wall cost reduction that protects the margin
Briefcase
ScopedWeeksLow capital

Plain-language contract risk reports for Malaysian SMEs, sold through their accountants.

Every SME signs contracts it never reads, and a lawyer is too slow and dear for a quick gut-check. Accountants and company secretaries already hold the trust and the client list; this is a paid add-on they resell, not a cold product.

Conservative: RM 9k-20k/mo at RM 79/report, ~120-250 reports a month across a handful of partner firms.
  • + A retrieval + risk-scoring pipeline proven on MatSelamat
  • + A B2B2C channel through accountant and co-sec networks
  • + Domain candidates held
Redacted until you're in
  • The liability-scoring mechanism and its guardrails
  • The revenue split with partner firms
  • The two firms already warm to reselling it
Sirah
ScopedWeeksLow capital

An intelligence layer that reads anonymous heritage and pilgrimage foot-traffic as intent, then hands tourism boards and site operators the one thing they have never had: provable attribution for who moved, why, and what it was worth.

States and trust bodies own the largest daytime crowds and the weakest measurement. Start at a single high-volume heritage precinct or pilgrimage corridor already running an ExplorAR-style guide; instrument the anonymous flow, model the visit psychographically, prove which routes and sponsors actually changed behaviour. The privacy posture is the inverse of the nightlife data plays: consent-led, no facial ID, PDPA-grounded. Land one precinct, expand to the trail, then to the board.

Conservative: RM 12k-30k per precinct attribution engagement, plus recurring intelligence retainers with boards and site operators; 3-4 realistic precinct/trail targets inside 24 months.
  • + ExplorAR foot-traffic and venue tech, plus the existing 207-place catalogue
  • + Applied-neuroscience and psychometric modelling (HEXACO, MSc) for reading a crowd, not counting it
  • + PDPA and privacy grounding for anonymised, consent-led data (advocate; RLB)
  • + The MatSelamat AI pipeline for the board-facing reporting layer
Redacted until you're in
  • The psychographic inference model and its feature set
  • The attribution and incrementality methodology
  • The anonymisation and consent mechanism that keeps it lawful
  • Named tourism boards, site operators, and the sponsor pipeline
Vellum
ScopedWeeksLow capital

A document-AI appliance that runs entirely on the firm's own machine; client files never leave the building, and every extraction carries a citation back to the page it came from.

The only way a regulated professional can use document AI without breaching confidentiality is if the model runs on-device and every output is auditable back to source. Both halves are now cheap. The buyers are professionals who are forbidden, not merely unwilling, to send matters to the cloud; you start where confidentiality is a rule, not a preference. Sell the box, not a seat: the firm owns the appliance and the audit trail.

Conservative: RM 18k MRR at RM 600/mo across ~30 small firms, plus RM 6k setup per firm; the box is owned, not rented.
  • + Legal grounding on privilege and confidentiality rules (RLB Advocates)
  • + A retrieval and extraction pipeline proven on MatSelamat
  • + The build and licensing pipeline; the product itself runs on the firm's own hardware
Redacted until you're in
  • The local-inference stack and which open models run on commodity firm hardware
  • The provenance-audit schema that makes outputs privilege-defensible
  • Per-box unit economics and the hardware bill of materials
  • Named pilot firms and Bar Council contacts
Custody
IdeaWeeksNear-zero capital

A flight recorder for AI agents; it sits between any agent and the tools it touches, logs every action with the reason, and produces a record you can hand to a client or a regulator.

Agents now write their own skills and drive real machines at runtime, so the dangerous gap is no longer capability but evidence of what the agent did and why. Don't compete on building agents; sit underneath whatever agent the operator already runs as the layer that records and can replay it. Sell first to agencies and MSPs who must answer to a client, then to anyone deploying agents into regulated work where a dispute is foreseeable.

Conservative: USD 99/mo per deploying team (~USD 4k MRR at 40 teams), plus USD 1,500 per incident-replay engagement.
  • + Evidentiary and chain-of-custody framing: what a defensible record must contain (advocate; RLB)
  • + A production agent already in the wild that needs this layer (Proxi; dogfood and channel)
  • + Claude and Gemini API tooling
Redacted until you're in
  • The broker interception point and how the action-plus-reason log is captured tamper-evidently
  • The replay mechanism and storage format that makes a record admissible-grade
  • Per-seat and per-engagement unit economics
  • Named MSP and agency design partners
Greenroom
IdeaMonthsModerate capital

A booking, rider and settlement layer for Southeast Asia's underground music circuit.

Promoters and collectives run this on WhatsApp and spreadsheets; there's no Resident-Advisor-grade tool tuned for SEA's informal, cash-heavy, multi-currency reality. The edge here is first-hand scene access, not a deck.

Conservative: RM 2k-5k MRR at RM 89/mo across 25-60 promoters and collectives, KL out to the region; ticketing take-rate is the upside.
  • + First-hand scene access through Cliff Coffin and the label
  • + A reusable multi-tenant architecture
  • + A short list of promoters who'd pilot it
Redacted until you're in
  • The take-rate model on ticketing and settlement
  • Which promoters committed to pilot
  • The payments workaround for cash-first markets

About Raul Lee Bhaskaran

Raul Lee Bhaskaran, also known as Raul Lee, is a Malaysian advocate and solicitor and technology founder. AI-native products at the seams of law, education, and culture. He builds and ships products across Artificial intelligence product development, Legal technology, Malaysian law, Advocacy and solicitation, Autism and neurodivergence, HEXACO personality assessment, Electronic music production, Augmented reality, Startup strategy, Arts rights and freedom of expression, Applied neuroscience. Ventures and projects include Proxi, Saraf, MatSelamat, Setara, HEXACO Protocol, ALIM, Cliff Coffin, Consult. He runs Axon Avenue PLT for technology ventures and RLB Advocates & Solicitors for legal practice, and founded ALIM, the Art & Law Initiative.