Marina Bay skyline painting

By Jam & Raph

Explore AI safely with us

Seamless AI integration for law firms and professional services

Currently working with

KGP LogoDelta LogoHanbridge Institute

Supported by

NVIDIA Inception ProgramSMU LogoTech NationBLOCK71

By the Numbers

0%

less admin time on workflows we automated

0

workshops delivered

0+

firms working with Jam & Raph

“Delta Law found the session really helpful for bringing out Claude and LLM effectiveness”

Joshua

Partner, Delta Law Corporation

Why us

Claude Certified Architects with a law background, on one team

Plenty of people will talk to you about AI. We hold Anthropic's own certification for building with Claude, we come from a law background, and we've won legal tech competitions as the same team.

Certified by Anthropic, not self-taught

We sat and passed Anthropic's own Claude Certified Architect exam (CCA-F). The way we build with Claude is the way Anthropic teaches it.

Law and engineering at the same table

SMU Law and ex-GovTech engineering on one team. The person who understands the matter sits next to the person building the workflow, not a lawyer briefing a vendor.

We've won legal tech competitions together

1st place at the SMU Legal Innovation & Tech Competition 2025 and 1st runner-up in 2024, won by the same team and judged on legal work rather than a generic AI demo.

We have built and deployed real solutions

We've created and maintained bespoke AI software that runs in production: Claude Cowork and Claude Code pointed at real matters and real file stores, and custom software where those can't reach.

Our Team

The team behind Jam & Raph

Law and engineering on one team. Claude Certified Architects who run the workshops and build the workflows themselves.

Awards & recognition

Jam & Raph as Top 6 Finalists at the Bangkok Business Challenge 2026
Finalist$1,300 USD

Bangkok Business Challenge 2026 (Sasin)

Jam & Raph team winning SMU Legal Innovation and Tech 2025
1st place$1,500 SGD

SMU Legal Innovation and Tech 2025

Jam & Raph team winning the Clifford Chance track at Cambridge Hack the Law 2026
Winner

Cambridge Hack the Law 2026 (Clifford Chance track)

Jam & Raph team winning 1st Runner Up at SMU Hult Prize 2026
1st runner up$5,000 SGD

SMU Hult Prize 2026

Jam & Raph team winning BASE Hackathon
1st place$1,000 SGD

NTU x Base Web3

Jam & Raph team at SMU Legal Innovation and Tech 2024
1st runner up$2,500 SGD

SMU Legal Innovation and Tech 2024

What we offer

Fluency first. Working systems second.

Most teams start with the workshop. Advisory and Build are there when one workflow needs more than three hours.

Path 1 · Start here

The Workshop

Not your standard prompt engineering workshop.

3 hours · one team · up to 8 people

01

AI fluency

How LLMs actually work, so you can raise the accuracy of what comes out and be confident.

02

Deep dive into Claude

When a chat window is enough, and when the job needs Cowork, pointed at your SharePoint matter folder, editing in place with formatting intact.

03

Case study

One real job end to end: an email and your firm template in, a finished letter of engagement saved back to the matter folder.

See it in practice

One letter of engagement, end to end

AI works with your current style and resources. A human is always in the loop.

Scope of work lands

A scope of work arrives in the inbox, exactly as it does today.

Template is pulled

Cowork pulls your letter of engagement template straight from SharePoint.

Drafted in house style

The draft comes back in your firm's voice, using your resources and conventions.

Saved for review

It lands in the matter folder for a human to check and sign off.

Path 2 · Put your Claude plan to work

Advisory

Bring us one workflow that hurts. We'll interview the people who run it, then get the Claude plan you already pay for doing the job, with the skills and orientation your team needs to keep it going.

Free 1-hour consult · milestone model

It starts with a free 1-hour consult

No commitment. The hour is for us to learn the firm: your practice areas, how you work, and how a matter generally moves through the office.

Then we interview the people doing the work

We sit with whoever actually runs the workflow and walk it end to end, step by step, exactly as it happens today.

We identify where AI applies

From the interviews we mark the steps AI can take on, and say plainly where it can't, including when the answer is to leave a workflow alone.

You get a product requirements sheet

Once we've scoped what we're building, it's written down before any work starts: what it does, what it touches, and what done looks like.

Path 3 · When it's time to build

Build

When Claude Cowork or Claude Code can't perform the workflow, we build the software that can, and we maintain it afterwards.

Custom AI software · milestone model

For when Cowork and Claude Code can't do it

Some workflows don't fit an off-the-shelf agent. That's the line where Build starts: custom software, written for the job.

Same discovery, aimed at software

The same consult, interviews, and product requirements sheet as Advisory, scoped for something we build and then maintain.

Whatever stack the job needs

Other models where Claude isn't the right fit, a real backend, and custom MCP servers into systems with no official connector: MYOB, LEAP, Talenox, in-house platforms.

Milestone model, maintained after

You pay per completed, scoped milestone, for the work that lands rather than for time spent. We keep it running afterwards on an optional retainer, cancel anytime.

Whether you want a workshop for the team or help with one workflow, we'd love to talk.

Common Questions

FAQ

Quick answers to the questions we hear most often.

Get Started

Leave with a system that runs.

Book a fluency workshop for your firm, or bring us one workflow that hurts.

Contact Us

Work with us

Whether you want a workshop for your team or help with one workflow, share a few details and we'll be in touch.

  • Workshop: 3 hours, one team, your own workflows baked in
  • Advisory: free 1-hour consult, then we interview your team
  • Build: custom AI software on a milestone model
  • PDPA-aware throughout, on SOC 2 compliant infrastructure
Typically responds within 24 hours

Fields marked with * are required.