Designed around existing infrastructure
The MVP direction is to register already-provisioned Linux VPS rather than automatically buying or provisioning cloud machines.
About ANDIP
Make distributed AI infrastructure easier to operate.
ANDIP is a pre-commercial infrastructure project exploring how engineering teams can coordinate agent workloads across the servers they already operate.
Project vision
Manage execution capacity—not a collection of disconnected sessions.
As agents take on coding, research, browser, data and operational tasks, their compute and policy needs become a distributed systems problem. ANDIP’s long-term direction is to make that infrastructure more coherent for smaller engineering teams.
Users should be able to express goals, budgets and constraints while the platform coordinates placement, isolated execution, monitoring, recovery and result aggregation. That is a long-term vision—not a claim about currently released product functionality.
Development & technical validation
Project identity
- Founder
- Lucian Nguyen
- Company
- Technology and Media Services, LLC
- Location
- 3208 Owens Ave, Cody, WY 82414, USA
- Stage
- Pre-commercial; architecture and validation stage
- Beta target
- Q1 2027
Technical direction
Build the missing substrate around reusable foundations.
The project plans to reuse agent and browser capabilities where appropriate while engineering the distributed queue, scheduler, worker control, isolation, budget and recovery systems that a multi-server platform requires.
Workload-aware by design
Coding, browser, research and data workloads differ in resources, permissions and risk; adapter boundaries are intended to preserve those differences.
Explicit boundaries and approvals
Agents should receive only task-scoped permissions. High-risk external actions need approval and durable audit evidence.
Measure before making scale claims
The 100-agent objective requires repeatable load, recovery and resource tests—not a count of open processes or simulated requests.
Current status: ANDIP is pre-commercial and undergoing technical validation. Community beta is planned for Q1 2027, with timing and availability subject to engineering progress.