Real experts. Digital twins.

The accountability of a human. The speed of a machine.

Work with AI digital twins of real experts. The twin does the work. The expert reviews, refines and puts their name on it.

Request a demo See how it works

Illustrative workflow Expert twin workspace

Daniel's twin Corporate legal work

Singapore · Expert playbook and sources

Example task

Review a shareholder agreement

Twin reviews and redlines

Twin-generated work, using the expert's defined approach.

Verification requested

Risky points are escalated for Daniel to examine and refine.

The missing layer

AI has scaled output. Not accountability.

GENERIC OUTPUT

A plausible average

General-purpose AI can produce more, faster. But a polished answer is not the same as expert-specific judgement.

EXPERT-SPECIFIC JUDGEMENT

A defined way of working

Fraction twins work with an expert's scope, knowledge, procedures and voice—with expert verification when requested.

What changes

Expertise designed into the work.

A twin extends the expert's way of working. Human-approved work remains clearly distinct from output produced by the twin alone.

An expert, not an average answer

Specific knowledge, preferred sources and expert-defined judgement shape the response.

AI speed. Human judgement.

The twin handles the work; the expert can examine, refine and approve it when verification is requested.

Built around the job.

Tasks follow playbooks and procedures designed for the work—not a single prompt trying to do everything.

How it works

From task to expert review.

Bring the work to a twin, then bring in the human expert where judgement and accountability matter.

Choose your expert

Select the expertise and task that fit the work you need to do.

Put the twin to work

Ask the twin to draft, review or redline using the expert's defined knowledge and playbook.

Request verification

The expert examines risky parts, refines the work and approves it within the agreed engagement and scope.

Move forward with confidence

Know clearly when output has been reviewed and approved by the expert, rather than generated by the twin alone.

Twin-generated work is not human-verified until the expert has reviewed and approved it.

Daniel Kung, corporate lawyer and Fraction co-founder

Daniel Kung Corporate lawyer · Singapore

The first twin

Starting with legal. Built for expertise.

Daniel's twin brings his corporate legal approach into an AI-assisted workflow, while escalating work for Daniel's review when human verification is requested.

Reviews against the expert playbook
Drafts and redlines documents
Advises with sources
Escalates for Daniel's review

Example areas

Shareholder agreements Convertible notes Data protection

Meet the first twin

Planned expertise

One model. More fields of judgement.

Fraction is starting with Daniel Kung's corporate legal twin. Additional professional domains are planned as the network develops.

Planned expertiseHR

Planned expertiseTax

Planned expertiseFinance

Planned expertiseCompliance

Planned expertiseEngineering

Starting fieldLegal

Availability will expand over time. Contact the team for current access.

Inside a twin

More than a prompt. An expert's way of working.

Each layer makes the twin more specific to the expert and the job it is designed to support.

Scope

Defines the work the twin is designed to handle and when it should escalate.

Knowledge

Uses expert-selected material, context and sources relevant to the task.

Procedure

Follows playbooks and steps that reflect how the expert approaches the job.

Voice

Communicates in a style suited to the expert, audience and work.

Our team

The people behind Fraction.

Dan Kung, Sam Kung and Veljko Vasic are building Fraction around how experts actually work.

CEO & Co-founder

Dan Kung

Singapore and New Zealand-qualified lawyer. Founded LDU in 2018, with in-house experience at IBM, Grab, Binance and ByteDance.

CTO & Co-founder

Sam Kung

Design and engineering experience at IBM, SEGA and EPAM. Builds the AI systems behind the twins.

Co-founder

Veljko Vasic

Founder of OTLRS, based in Singapore, with a background in payments and fintech.

Dan and Sam are twin brothers.

For companies

Put expert judgement to work.

Talk to the Fraction team about your workflows, the expertise you need and current access.

Send an enquiry

Interested in building your digital twin? Use the form to tell us about your expertise and how you work.

FAQ

The important questions.

What is a digital twin?

A Fraction digital twin is an AI system built around a real expert's scope, knowledge, procedures and voice. It can carry out defined work and involve the human expert for review and approval.

Is every AI output verified?

No. You can request expert verification. Work is only human-verified after the expert has reviewed and approved it.

Which expertise is available?

Fraction is starting with Daniel Kung's corporate legal twin. Other domains are planned; contact the team for current access.

Can I connect a twin to my own AI?

Agent-to-agent access is part of Fraction's product direction. Contact the team to discuss current integrations and your workflow.