Playground

Try the work before you take the call.

Six things you can use right now. Estimate how your support bill steps down, watch AI read a document with a person approving every field, get a four-week plan for your problem, look inside our ship-management ERP, walk through a 38-million-record migration programme, and watch an AI agent work through governed MCP tools in healthcare, logistics, manufacturing or shipping.

Everything runs in your browser. Sample documents and illustrative figures only. Nothing you type is sent anywhere unless you choose to book a call.
01Support-cost calculator 02AI document extraction 03Four-week plan generator 04Ship-ERP preview 05Data migration programme 06Agent-ready systems (MCP)
01 · Support-cost calculator

What your support bill looks like when the work shrinks.

Enter today's numbers. We show the fixed monthly fee stepping down at quarterly checkpoints as incidents fall, the way our commercials are written.

What this showsWe price against a measured baseline, never per head or per ticket.

Your current baseline

Illustrative only. Uses the step-down from our representative engagement (fee −15% at month 4, −30% at month 7, −40% at month 9; incidents −55% by month 9). Your real numbers come from a twelve-month baseline measured before any promise.

—Today, per year (all-in)
—Run-rate after month 9, per year
—First-year saving
Monthly cost, indexedIncident volumeQuarterly checkpoint

02 · AI document extraction

The model reads it. A person approves it. Every step is logged.

Pick a sample document and run extraction. Low-confidence fields are flagged for review, totals are cross-checked, and nothing posts until someone approves it.

What this showsAI inside the workflow, with governance designed in, not a chatbot on the side.
Ready.
Extracted fields will appear here.
audit log · this session

Sample documents with pre-computed results, so nothing leaves your browser. In a discovery session we run the same flow on your own documents, inside your environment.

03 · Four-week plan generator

Your problem, week by week, with the demo dates on it.

Answer four questions. You get the plan we'd bring to a first call: what happens each week, when you see a demo, and what we'd need from you.

What this shows"Four weeks" is a method, not a slogan.
Systems it needs to talk to
Your plan will appear here.
04 · Ship-management ERP

Software you can see running before you sign anything.

A preview of the core modules with sample fleet data. Ask for a sandbox and we'll give you your own read-only tenant to click through.

What this showsAn ERP accelerator that already works, so implementation is configuration, not construction.
SeaOS · Sample fleet

Get a sandbox

  • Your own read-only tenant with a sample fleet
  • All six core modules, end to end
  • A 30-minute walkthrough with the engineers who built it

Sample vessels and data are fictional.

05 · Data migration programme

Seven systems, 38 million records, one weekend. No surprises.

A fleet operator consolidating the systems it inherited through two acquisitions onto one platform. Walk through how we plan, rehearse and cut over a migration this size, and how we prove nothing was lost.

What this showsMigration engineered like a product: golden records, dependency-ordered waves, timed rehearsals and business-level reconciliation.

Representative programme with fictional data. Volumes and timings reflect a mid-size fleet operator; yours are measured in discovery.

06 · Agent-ready systems (MCP)

Let AI agents use your systems, with your rules attached.

We put a Model Context Protocol (MCP) server in front of your core system. Any MCP-capable agent, such as Claude, Copilot or your own, can then look things up and propose actions. Writes wait for a named person to approve, and sensitive data never reaches the agent.

What this showsOne governed doorway for AI into any operation: scoped tools, approvals, data minimization and a full audit trail.
Industry:
Pick a question to watch the agent work.
mcp audit log · this session

Scripted replay against fictional data. In a session we connect a live agent to a sandbox MCP server over your test environment.