Memory, coordination, and execution infrastructure for reliable AI agents.
Underpass AI helps agents recover context, coordinate specialist work, execute through governed tools, and leave auditable evidence behind.
We do not build foundation models. We build the operational substrate around them: navigable memory, event-driven coordination, governed execution, evidence, policy, and observability.
Problem
Agents are getting better at work, but weak infrastructure makes them hard to trust.
Modern agents can read code, call tools, run tests, and operate across complex workflows. But most agent systems still lose context between runs, repeat failed attempts, hide state inside frameworks, and produce weak audit trails.
- They lose relevant context between executions.
- They repeat failed attempts because past process memory is not navigable.
- They cannot always explain what evidence supported a decision.
- They execute tools without enough governance or inspection.
- They are difficult to observe, debug, and audit in production-like systems.
Platform model
Three infrastructure planes around the model.
Underpass separates memory, coordination, and execution so each part can be inspected, tested, and evolved independently.
Agents should not start every task from zero. They should recover scoped memory, understand previous attempts, coordinate the next step, execute through controlled runtimes, and write evidence back into the system.
- Domain event
- Specialist agents
- KMP by Underpass restores scoped memory
- MADE coordinates deliberation and work
- Runtime governs tool execution
- Evidence is recorded
- Memory improves for the next event
Components
Public infrastructure components.
KMP and Runtime are the main public technical assets. MADE is public and shares their contract-first discipline.
- Public memory plane
KMP by Underpass
Kernel Memory Protocol for temporal, multidimensional, auditable AI agent memory.
- Scoped memory with about and dimensions.
- Temporal traversal: goto, near, rewind, forward, trace, inspect.
- Evidence-backed deterministic retrieval.
- Explicit relations and provenance.
- Typed gRPC API plus MCP adapter over the same semantics.
- Kubernetes/Helm deployment path.
- Adapter-based persistence for graph, key-value, and event roles.
View KMP by Underpass on GitHubHow you run it
- Embedded One developer, one project.
- The kernel runs in-process inside the MCP stdio binary. No database, no service, no API key: memory lands in a per-project .kernel/ store, fsync-durable, and a local read-only viewer draws the graph, the timeline and the traces. Verified in Claude Code and Codex CLI.
cargo install kmp-mcp - Cluster Memory shared and audited across teams and services.
- The typed KernelMemoryService over gRPC, with graph, key-value and event persistence behind ports. Helm chart, TLS and mTLS on the boundaries, OpenTelemetry and structured logs.
- Public execution plane
Underpass Runtime
Governed execution plane for tool-driven AI agents.
- Isolated workspaces.
- Governed tool execution.
- Policy checks before execution.
- Telemetry and evidence trails.
- Adaptive tool recommendations.
- mTLS- and Kubernetes-oriented deployment.
- OpenAPI/gRPC-oriented runtime surface.
- Public coordination plane
MADE by Underpass
Multi-Agent Deliberation Engine: councils of specialist agents that produce a decision with a reconstructable why.
- Councils propose, critique, revise, validate, and score.
- Declarative YAML ceremonies: states, steps, roles, guarded transitions.
- Output contracts, with an explicit refusal when no proposal satisfies them.
- Optional LLM-as-judge ranks proposals by intrinsic quality.
- The debate is replayable as an OpenTelemetry trace.
- Deliberation metrics: judge discrimination, winner scores, token cost.
- Typed gRPC + AsyncAPI contracts, MCP adapter, Kubernetes/Helm.
View MADE by Underpass on GitHubHow you run it
- Embedded Run a ceremony from the terminal you already work in.
- The ceremony engine runs in-process over MCP stdio, calling the same use cases the deployable binary calls. Ceremony state persists in a local file, so a ceremony outlives the process. No service, no database, no credentials.
cargo install made-mcp - Cluster Deliberations that outlive a process and scale across replicas.
- The full gRPC contract, optional NATS messaging and Postgres persistence, provider-backed agents, the LLM judge, Prometheus metrics and OTLP traces. Helm chart with pinned images and hardened manifests.
Why now
Agentic systems are moving from demos to operational workflows.
LLMs are becoming capable enough to use tools, inspect code, run commands, and participate in real engineering workflows. That creates a new infrastructure problem: memory, governance, auditability, and observability need to become first-class parts of the system.
- Software engineering is one of the first domains where agentic workflows can produce measurable feedback.
- Tool-using agents need runtime isolation and policy checks.
- Long-running agents need memory that can be navigated, not just retrieved.
- Human operators need evidence trails, traces, and failure classification.
- GPU-backed local and distributed inference is a strategic technical direction.
Writing
From the lab notebook.
Longform notes on the ideas behind Underpass AI: navigable agent memory, auditable multi-agent deliberation, and the economics of running agents in production.
- AI infra
I let my GPU workers shut themselves down after five minutes
Running evidence-bound agent reviews on ephemeral AWS GPU workers while preserving traces, logs, metrics, and decision artifacts after shutdown.
- MADE
No queremos agentes que contesten. Queremos decisiones que se puedan auditar
A multi-agent design meeting as a declarative ceremony: proposals, peer critique, an LLM judge — and the whole decision replayable as an OpenTelemetry trace in Grafana.
- Operator
Operator: cuando responder no basta
Training Qwen 0.5B with LoRA to emit exact Kernel Memory Protocol actions from visible memory state, without turning it into a general reasoning model.
- KMP
Building Kernel Memory Protocol: navigable memory for AI agents
Why agent memory has to be temporal, multidimensional and auditable — and what a kernel-style protocol for it actually looks like in practice.
- Event pipeline
What an event-driven agent pipeline looks like when you trace it end-to-end
Walking a full agent pipeline as a stream of events: who emits what, where decisions get coordinated, and what the trail actually proves about behaviour.
- Economics
Why event-driven agents reduce scope, cost and decision dispersion
The architectural and economic case for event-driven coordination over chatty agent loops: smaller scope, cheaper retries, fewer dispersed decisions.
Open source
Built in the open.
Underpass AI is developed around open infrastructure principles. The public GitHub organization exposes the technical direction, the core repositories, and the foundation for future community collaboration.
KMP by Underpass
kmp · Rust · Apache-2.0 · formerly rehydration-kernel
View KMP by Underpass repo →MADE by Underpass
made · Rust · Apache-2.0 · formerly underpass-choreographer
View MADE repo →Founder
Founder-led technical infrastructure.
Underpass AI is created by Tirso García Ibáñez, a software architect with 15+ years of experience across software engineering, architecture, distributed systems, cloud-native platforms, and technical leadership.
The current focus is AI infrastructure for agentic systems: navigable memory, governed execution, Kubernetes-native runtime foundations, and production-oriented architecture patterns for LLM agents.
Contact
Talk to Underpass AI.
Interested in reliable agent infrastructure, technical validation, open-source collaboration, or early-stage product conversations?
- General contact@underpassai.com
- Founder tgarciai@underpassai.com
- GitHub github.com/underpass-ai