What we can build — an AI feasibility plan for agentic instruments and agentic systems.
What this document argues and what it refuses to argue. Cost physics, not pricing.
The givens: team, Mindgap · ATP · Agent Rung, Innovation Centre siblings, Vietnamese market, BigSmall partnership.
Twelve structural properties. Three tiers of client engagement. Cross-venture synergy with cybersec and IoT.
What the venture can build that client self-build and freelancers cannot — a property-by-property comparison.
Pillar 1 catalogue with three readiness zones. Pillar 2 reference-architecture mobilisation — one-to-two-week scoping.
Two standing, one per-engagement. Six expansion roles — hired by workload triggers, not calendar.
Both pillars run a venture backend. Pillar-1 shared multi-tenant, pilot-measured. Pillar-2 ~$30/deliverable. QS quote anchors the salaries.
Seven AI-operating risks. Mitigations are engineering-operational — KB cadence, validator, deployment posture.
The entire plan in one page — feasibility claim in its shortest form.
Reference instances, Vietnamese regulatory context, Anthropic platform, and the claims original to this drafting.