Repository / Systems

System Landscape

Fourteen systems organized across four tiers: flagship platforms, infrastructure primitives, evaluation harnesses, and research programs. Each operates under the Deep Bound Research governance framework.

Retro workstation rendering the Deep Bound system landscape

Release Disclosure

Deep Bound Research currently publishes public-safe research papers, governance drafts, technical notes, system maps, and design-space artifacts.

Core systems including Ex1, Boundary, Plateau, ACE, Cerberus, and related harness infrastructure are not publicly released unless explicitly marked with: Live Repository, Public Release, Pilot Access, or Open Source Availability.

Public pages describe research direction, governance models, operational principles, and public-safe architectural abstractions. Internal implementation details remain private until reviewed for safety, stability, and release readiness.

FLAGSHIP SYSTEMSINFRASTRUCTURE SYSTEMSEVALUATION + GOVERNANCERESEARCH PROGRAMSEx1WORKSPACEACERETRIEVALStrongHoldDATA INGESTBoundarySIMULATIONCerberusDEFENSEPlateauDESIGN SPACEArtchivMEDIA ARCHIVEM-ClassEVALUATIONX-RouterROUTINGAMRAGENT RUNTIMETAGDELEGATIONGOCOPERATOR BRIDGETripSitterSUPERVISIONArrows indicate documented runtime dependencies between public system surfaces.
6
Flagship Systems
4
Infrastructure Systems
3
Evaluation + Governance
1
Research Programs
01

Flagship Systems

High-visibility systems on the public research surface. Each represents a distinct operational layer in governed AI infrastructure.

Controlled Research BetaPublic Site

Ex1

AI Workspace / Operator Runtime

A desktop-first AI-native workspace and operator runtime for governed agent execution, project-aware context, tool orchestration, and recoverable long-horizon work.

Why it matters

Ex1 gives operators a governed surface for running AI work across agents, tools, context, approvals, and recovery points instead of scattering work across disconnected chats, tabs, and ad-hoc automations.

Public site: everythingx1.com
Research PrototypeInternal Only

Boundary

Simulation Infrastructure

A simulation and environment framework for testing AI behavior, system interaction, workflow reliability, and controlled runtime scenarios.

Why it matters

Provides high-fidelity, repeatable environments for evaluating agents against safety and performance benchmarks before production exposure.

Internal R&DInternal Only

Cerberus

Defensive Security Harness

A defensive security research system for sandboxed analysis, runtime mitigation planning, blast-radius reasoning, and evidence-led security workflows.

Why it matters

Provides a constrained, evidence-led layer for studying agent failures and runtime risks without enabling offensive use.

Architecture DevelopmentInternal Only

StrongHold

Data Ingest & Archive Platform

A governed data ingest and archive platform for deduplicating, compressing, versioning, retrieving, and restoring large research and AI-system data streams.

Why it matters

Data ingest · Deduplicated archives · Adaptive compression · Retrieval · Restore planning.

Active Research SystemInternal Only

Plateau

Design-Space Intelligence Laboratory

A design-space intelligence system for generating, comparing, scoring, and archiving candidate architectures across products, infrastructure, machines, interfaces, and AI runtimes before engineering commitments are made.

Why it matters

Plateau gives the lab a governed way to explore what could be built before deciding what should be built. It turns ambiguous ideas into structured candidates that can be constrained, scored, promoted, archived, or rejected.

Planned ReleasePublic Site

Artchiv

Creative Media Archive Platform

The source of truth for creative media — a governed archive for versioned assets, provenance records, retrieval, and inspectable media lineage across studios, agencies, and production workflows.

Why it matters

Creative teams lose time and trust when final files, working masters, approvals, and delivery history live in disconnected drives, chats, and ad-hoc folders. Artchiv gives media a durable, queryable record.

Public site: artchiv.com
02

Infrastructure Systems

Underlying runtime primitives — routing layers, delegation protocols, context engines, and operator bridges that underpin flagship systems.

Open Source PreparationOpen Source Prep

ACE

Context Engine

A context and retrieval engine for ranking, routing, and surfacing relevant information across complex AI workspaces and agent workflows.

Why it matters

Treats context as infrastructure: ranks, prunes, and routes sources so agents see only what is strictly necessary for the current step.

Research DirectionNot Released

X-Router

Capability Routing Layer

A capability-routing pattern for sending work to the right model, tool, agent, or runtime based on task type, context, cost, and governance constraints.

Why it matters

AI systems should not treat every task as a default chat completion. Routing improves reliability, cost control, and operational clarity.

Internal R&DInternal Only

TAG

Typed Agent Gateway

A typed delegation protocol for routing work between agents, tools, models, and humans with scoped authority, structured handoffs, and replayable records.

Why it matters

Reliable agent systems need more than loose prompt chaining. TAG treats delegation as a governed protocol with explicit capability boundaries and evidence.

Architecture DevelopmentInternal Only

GOC

Governed Operator Bridge

A governed operator bridge for controlled interaction with CLI-first, browser-based, and legacy software surfaces.

Why it matters

Many useful tools do not expose clean APIs. GOC studies how agents can operate software through controlled, observable, and permissioned interaction layers.

03

Evaluation + Governance

Harnesses and oversight systems for measuring agent behavior, long-horizon reliability, and supervisory runtime control.

Internal R&DInternal Only

M-Class Harness

Governed Evaluation Harness

A governed evaluation harness for coding, reasoning, and multi-step AI workflows, focused on planning, revision, recovery, and evidence production.

Why it matters

As AI agents move beyond single responses, evaluation must measure sustained work: state tracking, tool use, failure recovery, handoff quality, and evidence trails.

Internal R&DInternal Only

Long-Horizon Harness

Agent Evaluation Harness

An evaluation framework for AI agents across extended tasks involving memory, context changes, interruptions, recovery points, and final artifact quality.

Why it matters

Short benchmarks miss the failure modes that appear in real work. Long-horizon evaluation tests whether agents can sustain coherent execution over time.

ConceptNot Released

TripSitter

Supervisory Safety Runtime

A supervisory runtime pattern for monitoring autonomous AI sessions, detecting uncertainty or drift, and escalating work back to a human operator when needed.

Why it matters

Autonomy requires active supervision. Systems need ways to notice when they are uncertain, off-policy, over-budget, or operating outside safe bounds.

04

Research Programs

Experimental research initiatives exploring controlled machine interaction and agent-software runtime boundaries.

Research PrototypeInternal Only

AMR

Agent-Machine Runtime Research

A research program for connecting AI agents to controlled machine interfaces while preserving observability, permission boundaries, and recoverable execution.

Why it matters

Many real workflows happen outside clean APIs. AMR studies how agents can interact with operational software without losing auditability or control.

Research Structure

Deep Bound Research uses a staged R&D structure. Early projects begin in a private founder workspace. Reusable research infrastructure is extracted into Deep-Bound-Research. Flagship systems live under House-Labs. Each system is governed by the Deep Bound Research maturity framework.

Concept → Research Direction
Exploration phase
Research Prototype → Internal R&D
Development phase
Public Draft → Operational
Release phase