One short call.
Then a clear next step.

Engagements range from US$20K for a focused two-week diagnostic to US$200K for a six-month build-and-handover program. Most production workflow engagements land between US$60K and US$120K. We send a fixed-scope SOW after the scoping workshop — no time-and-materials surprises.

Six to ten weeks from contract signature for a typical mid-market engagement. The two-week scoping workshop converts the brief into a working plan, then build sprints take four to eight weeks. A full hand-over runbook is delivered before the engagement closes.

You pick the region. Default is the country of the contracting entity. We run in HK, SG, JP, and KR (from Q3 2026); other markets default to Singapore unless you elect otherwise. Full residency table on the Trust page.

No. Client data is never used to train shared models, and the same term is held with every model provider in the loop. Where fine-tuning is part of the engagement, the resulting weights are owned by the client and isolated to their environment.

Provider-agnostic. We deploy a model-router into every engagement so you can switch between Anthropic, OpenAI, Google, AWS Bedrock, Azure OpenAI, and open-weight models in your own VPC without rewriting the workflow. The selection rubric is documented in the engagement runbook.

Yes. A short call costs you 30 minutes and the conversation is useful either way. If a deterministic system, a process redesign, or hiring a person is the better answer, we say so. We turn down work where AI does not fit — it protects both sides.

Yes. Engagements run in English by default and in Japanese, Korean, Mandarin, Cantonese, or Bahasa where the team and the documents require it. Each office has a local working language; the delivery standards are the same in every market.