Menu
Portfolio · Engineering in practice

See the problem.
See the solution.
See the audit.
See what changes.

Our projects show how AI, software and operating controls work together—from an employee request to a live trading decision.

Each family illustrates a capability we can bring to a client engagement.

Operational AI

Suri

From “who is handling this?” to a visible workflow.

A claim, leave request or customer issue often starts in a message. Someone then collects the facts, checks policy, finds the approver and updates the application.

How the system helps

Suri brings identity, approved knowledge, case memory and workflow together. It prepares the work, routes decisions to the right person and acts through approved integrations.

What this brings to a client

A practical starting point for a workflow designed around your own people, policies and systems.

Explore Suri’s architecture and use cases
Hands reviewing a receipt beside a laptop in a sunlit office
A receipt claimIllustrative workflow
Employee sends the receiptAcknowledgement and a visible case
Suri prepares and coordinatesIdentity · receipt facts · approved policy · case state
HR or manager makes the decisionEvidence and exceptions arrive with the case
The application records the outcomeThe employee can see where the work stands
People retain decision authority. Applications retain the record.
Configurable agentic workflow

Sura
Agentic Workflow

Your process, configured around how you work.

Stop rebuilding the handoffs around each AI conversation. Define the steps, responsibilities, permitted applications and human decisions in a reusable process.

Configure the way work moves

Choose human, AI specialist or application steps. Set instructions, expected results, review paths and output destinations, then publish the version you want each case to follow.

Find a familiar scenario

Explore SME operations, banking, marketplaces, WhatsApp selling and personal planning ideas. Catalogue coverage is clearly distinguished from tested live integrations.

Explore Sura workflow scenarios
Choose the model
for each task
SuraWorkflow illustration
YOUR REUSABLE PROCESSInvoice to payable draftConfigured once. A separate case each time.
  1. Read the invoiceApproved application
    01
  2. Prepare the exceptionsYour chosen AI specialist
    02
  3. Review and decideThe authorised person
    03
  4. Save the outcomeApplication + result file
    04
Results saved
where you work
Configurable workflows with people, AI and applications working together.
02 · AI decisions in operation

AI Trading

Give AI a clear role—and control what happens next.

Trading brings the engineering challenge into focus: changing data, model judgement, independent strategies and several brokers must work together reliably.

How the systems work

Apex Trader, Vega Trader and Spartan Trader use AI to trade. Multibroker supports Claude trading directly or supervising a trading bot. Independent strategy sleeves can be tested and replaced individually.

Research and feedback

AI Stock Screening combines Qwen and Kronos with performance tracking. Across the family, broker integrations include Tiger, IBKR, Alpaca, Moomoo and OANDA. Risk checks, monitoring and reconciliation surround execution.

Explore the five systems and their architecture
A sunny trading research desk with market charts, laptop and a hand operating the mouse
Two roles for ClaudeMultibroker
AI tradesClaude decides

Market and account evidence informs the trade.

Bot trades · AI supervisesBot executes its strategy

Claude watches health, drift and exceptions.

Permission · risk · order validation
Broker adaptersTiger · IBKR · Alpaca · moomoo
Fills, reconciliation, performance and recovery
Broker coverage varies by implementation; the detail view maps each system.
03 · Agent platforms

OpenClaw,
NemoClaw & Hermes

Make the agent useful inside a complete system.

A model can understand a task. An application still needs context, tools, permissions, runtime services and a way to show people what happened.

What the work explores

These platform projects examine how agent runtimes connect with selected models, approved tools and operating controls.

What this brings to a client

Practical experience for choosing a platform, defining its boundaries and connecting it to the work your organisation needs done.

Explore the platform architecture
An engineer connecting a realistically proportioned Mac Studio on a sunlit workbench
What an agent needs around itSystem view
Person and taskIntent · constraints · current context
OpenClawNemoClawHermes
Models + context

Understand and prepare

Approved tools

Perform permitted work

Permissions · human authority · evidence · recovery
Platform engineering connects the model to practical execution and oversight.
04 · Applied AI · Mobile product

Swipe’n’Sweep

Your taste.
Your AI Picks.

A fun, fast swiping experience for photo and video clean-up. Personalised AI Picks learns from your Keep and Trash choices, while you decide what stays and goes.

AI that runs on the device

AI Picks uses learned preferences to find photos and videos worth another look. Apple Intelligence, Gemini Nano or Qwen support depends on the device and provider eligibility.

Recognition and focused review

Find document photos and read detected text. Compare exact copies and similar shots side by side, review quality issues, or begin with large files, albums, months and saved locations.

Fast swiping. Full control.

Swipe to Keep or Trash, Undo a choice and review before deletion. AI helps you move through your library; you decide what stays and goes.

Explore Swipe’n’Sweep and AI Picks
Swipe’n’Sweep’s approved product campaign showing an AI Picks screen on a phone by the seaAI helps.
You choose.
Apple IntelligenceGemini NanoQwen
AI Picks Document recognition Compare & review
Product campaign · illustrative app screen. Model availability varies by device.
Contact us

Which operating problem should we start with?

Tell us what is slowing the work down, what you want to change and which people or systems are involved. We can discuss a Blueprint, a pilot, an implementation or a defined enterprise technology mandate.

The problemWhat is happening today?The outcomeWhat should improve?The environmentWho and what is involved?
Singapore
Malaysia