Build AI employees that compound.

A senior AI engineering team for companies that need working systems, not slideware. We design, build, and operate AI employees around your data, brand, workflow, and architecture.

Discovery callAccess Your Proposal

Tell us the role

[email protected]

Discovery call

30 minutes with founders

Proposal

Custom scope in 48h

DBS Bank

Enterprise AI

OCBC Bank

First Gen AI rollout

Sonzai Labs

Mind Layer platform

Pocket Souls

Consumer AI

Agency model

A senior team, not a vendor on the side.

We work like a forward-deployed product and engineering team. You get founder attention, senior AI architecture, production code, and an operating loop that keeps improving after launch.

$0M+

Value delivered through AI at DBS Bank

0

AI systems shipped in 2026

0+

Years building production AI systems

What we build

Three disciplines. One production system.

01

AI Employees

Operating roles, not demos.

We build agents that qualify, triage, follow up, report, and improve inside your real workflow. They use your tools, your data, and your guardrails.

  • Customer operations agents for tickets, cases, and follow-up
  • Lead qualification agents that score, route, and nurture demand
  • Decision-support agents that synthesize messy internal context

02

Mind Layer

State that compounds.

Most AI resets every session. Sonzai gives an agent operating context across people, accounts, teams, workflows, and time.

  • Persistent memory, mood, personality, and relationship state
  • Model-agnostic context assembly under a 200ms retrieval target
  • Lower model spend by moving cognitive load into the runtime

03

Custom Platforms

Your brand, data, and architecture.

When a generic tool cannot fit, we build the product surface and backend around the way your business already works.

  • White-labeled front ends and internal workbenches
  • CRM, database, comms, and API integrations
  • Enterprise security, audit trails, and deployment support

How we work

Architecture, then code, then operating cadence.

The first version is intentionally focused. We prove the workflow with real users, then turn every deployment signal into better state, stronger routing, and a more useful AI employee.

01

Discovery

We map the workflow, users, systems, failure modes, and the business outcome worth automating.

1-2 weeks

02

Architecture

We design the agent loop, state model, integrations, human escalation, and production guardrails.

1-2 weeks

03

Build

We ship a focused first version, validate on real work, and expand only where the data supports it.

4-12 weeks

04

Evolve

We monitor, tune, and keep improving the system as the agent learns the operation.

Ongoing

The team.

Ryan Foo

CEO

Product and strategy

Ryan has shipped products across AI, gaming, and consumer tech. At Limit Break, he designed progression systems for a studio that raised $200M. At Delphi Digital, he invested in consumer products and emerging technology markets.

  • Ex-Game Designer, Limit Break
  • Ex-Consumer Investor, Delphi Digital

Nasrudin Salim

CTO

AI architecture and engineering

Nasrudin has architected production AI systems at Singapore's largest financial institutions. As Head of MLOps at DBS Bank, he built infrastructure that delivered $70M in annual value. At OCBC, he led Singapore's first generative AI rollout.

  • Ex-VP Engineering, Almanak
  • Ex-Head of MLOps, DBS Bank
  • Ex-Lead AI Architect, OCBC Bank

Examples

Vertical agents built around real work.

Every business is different. These examples show the shape of bespoke AI employees we can design, launch, and operate with you.

Start here

Tell us what you want the AI employee to own.

The form sends your request to [email protected]. We review the workflow, then reply with a path to a founder-led consultation and proposal.

[email protected]