Native perception and communication
Full vision, audio, realtime conversation, software tools, and the model freedom to choose the right intelligence for each part of the job.
The audacious roadmap · 2026
AGI is intelligent orchestration.
ADK-Rust is building the Rust-native platform for high-performance agent systems that coordinate specialists, operate securely, and adapt across cloud, edge, enterprise, spatial, and physical environments.
Our thesis is practical: the future will be built from capabilities working together—models, tools, memory, agents, people, and machines—on top of a runtime developers can inspect and control. This roadmap is how we intend to get there.
Production runtimeSuper AgentsEnterpriseEmbodied AISelf-improvementSpatial OS
The long-term system architecture
Many capabilities. One understandable runtime.
The framework grows by adding composable capabilities around a stable execution contract, so each product compiles only the system it needs.
01
Intent and environment
A person, another agent, software event, sensor, or physical environment gives the system something to understand.
People
AI agents
Applications
Business events
Sensors
3-D worlds
observe · understand · choose the right capability
02
Intelligent orchestration
A coordinator assembles specialists for the job, chooses an execution path, preserves context, and adapts as new information arrives.
See
2-D · 3-D · world models
Hear & speak
voice · audio · realtime
Reason
models · memory · RAG
Use software
tools · MCP · coding
Coordinate
agents · graphs · routing
stream every meaningful decision through a typed runtime
03
ADK-Rust runtime
Agents share one async contract for invocation context, tool use, state, events, cancellation, artifacts, and protocols.
Agent + Runner
Tools + state
A2A · ACP · AWP · MCP
Sessions + artifacts
04
Transparent control plane
Humans and operators can inspect, constrain, evaluate, approve, and improve the work.
Policy
Telemetry
Evaluation
Approvals
deploy the same agent system where the work needs to happen
05
Adaptive deployment
The installed surface declares its available compute, tools, interfaces, permissions, and hardware. The agent works within those real capabilities.
Cloud
Enterprise
Edge
Device
Spatial
Physical
The engineering belief
Why intelligent orchestration?
No single model owns every skill, every piece of context, or every system where useful work happens. Production agents need specialist capabilities: one may reason over policy, another may see a 3-D scene, another may write code, and another may operate a financial workflow.
Orchestration turns those pieces into a system. It decides which capability should act, carries context between them, runs independent work concurrently, records progress, and brings a person into the loop when judgment or authority is required.
Rust gives that system a strong runtime foundation: explicit types, predictable performance, safe concurrency, small deployable binaries, and control over where execution happens. ADK-Rust adds the agent contract, workflows, protocols, state, evaluation, and operating controls around it.
The end-of-2026 target
One modular platform for the complete agent system.
We want developers to choose only the capabilities their product needs while keeping a clear path to a much larger system. These six outcomes define the platform we are working toward.
Autonomy you can shape
Long-running work, schedules, memory, tool use, recovery, and human checkpoints assembled for the level of autonomy your product can responsibly support.
Enterprise security on every surface
Identity, authorization, secrets, isolation, approvals, audit evidence, and deployment policy designed into the runtime boundary.
A transparent control plane
Typed events expose what an agent saw, chose, called, produced, and changed so teams can operate the system with evidence.
Physical and spatial capability
A path from browser and business automation to devices, 3-D worlds, robotics, and interfaces that understand the environment around them.
A mature developer platform
Scaffolding, Studio, examples, evaluation, benchmarks, deployment workflows, and documentation that shorten the distance from experiment to product.
The 2026 delivery path
Each milestone proves a harder form of orchestration.
The roadmap moves from software agents with business value to systems that can improve, understand space, and act through physical interfaces. Status labels separate what developers can build on today from active experiments and longer-horizon platform work.
Q1 · FOUNDATION
AVAILABLE AND EXPANDING
Prove the runtime through Super Agents and enterprise deployment.
The first milestone turns framework primitives into complete products. Developers should be able to study an agent that creates business value, then deploy the same architectural patterns with production controls.
Super Agents
Open, vertically integrated agents that combine coordination, tools, memory, persistence, deployment, and a measurable business outcome.
Multi-agent coordinationTool use and business systemsMemory and durable stateProduction deployment patterns
Enterprise platform
An open platform for agents that generate revenue, reduce cost, and remain understandable to the organisation operating them.
Deploy, scale, observeSecure configurationRelease and rollbackEnterprise integrations
Where this stands: The framework foundation for both tracks is available. The work continues through stronger Super Agent case studies, managed operations, and enterprise product delivery.
Q2 · EXPERIMENTAL FRONTIERS
ACTIVE FRONTIER
Move agents beyond browser tabs and business APIs.
This phase asks whether one composable runtime can coordinate software intelligence, spatial understanding, and physical action without losing the controls developers rely on.
Autonomous robotics
Build the first autonomous robot or robotic software system powered by ADK-Rust, with hardware access treated as a governed tool boundary.
Sensor inputAction planningHardware toolsSafe interruption
3-D worlds
Connect agents to Unreal Engine, similar tools, and world models so they can understand and automate spatial environments.
World state3-D perceptionSimulationSpatial actions
Mega Agent
Create one public showcase that brings the framework’s strongest native capabilities together in a system developers can inspect and run.
MultimodalMulti-agentLong-runningProduction operated
Where this stands: Robotics and Spatial OS remain experimental targets. The Mega Agent is the integration proof: it should expose the seams, trade-offs, and operating evidence for developers to inspect.
Q3 · SELF-IMPROVEMENT
NEXT SYSTEM CAPABILITY
Let agents improve through measured, reviewable change.
Self-improvement begins with a closed engineering loop. The system evaluates real outcomes, finds a weakness, proposes a change, tests it against baselines, and promotes it only when the evidence is stronger.
01Observe
02Evaluate
03Diagnose
04Propose
05Test
06Promote
Foundation already in place
Evaluation, prompt optimization, reflection, benchmarks, regression baselines, and typed execution events provide the raw material for improvement.
Trajectory evaluationLLM judgesPrompt optimizationRegression detection
The target
Autonomous improvement loops that can refine prompts, tools, routing, and execution policy while retaining review, rollback, and a record of why change occurred.
Bounded changesIsolated trialsHuman approvalReversible promotion
Where this stands: The standard is useful improvement developers can reproduce. Every promoted change should carry its inputs, evaluation method, result, and rollback path.
Q4 · SPATIAL OS
AUDACIOUS NORTH STAR
Consolidate agents across cloud, edge, device, and space.
Spatial OS is the proposed secure deployment platform for Super Agents. Its optional spatial interface explores how people may work with agent systems when conversation, screens, physical space, and hardware become one environment.
One deployment model
Describe an agent system once, then place its capabilities across servers, PCs, phones, homes, cars, TVs, edge hardware, or robots.
CloudEdgeDevicesPhysical systems
Capability awareness
Each installation tells the agent what models, tools, sensors, displays, permissions, and compute it actually has before work begins.
DiscoverConstrainRouteAdapt
Optional spatial interface
Explore 3-D and holographic interfaces where they genuinely improve control, understanding, or physical-world interaction.
Spatial UIWorld contextHuman controlEmbodied feedback
Where this stands: Spatial OS remains a research and platform target. Publishing it on the roadmap makes the direction open to technical challenge and contribution.
The proof is a working product
Super Agents turn framework capability into business outcomes.
A roadmap is credible when developers can inspect what it produces. Each Super Agent combines multiple ADK-Rust capabilities around a complete vertical job, including the interfaces, approvals, persistence, and deployment choices surrounding the agent itself.
zSpreadsheet
A production spreadsheet agent that creates models, works with formulas and charts, and turns a business request into an inspectable artifact.
Read the case study
Amos AI Accountant
An accounting agent that works across ERP workflows through voice and chat while keeping financial actions inside business controls.
Read the case study
JobHunter
A multi-agent job search system that researches, prepares applications, preserves state, and waits for human approval at consequential steps.
Read the case study
ADK Gateway
A channel and capability gateway that connects agents to people, tools, memory, websites, and independently deployed agent systems.
Read the case study
What success looks like
A serious framework for the next generation of agent systems.
By December 31, 2026, we want ADK-Rust to be recognised by developers for five things: performance, orchestration, deployment, production readiness, and an open path toward spatial and embodied AI.
Everywhere work happens
Run securely in a server, PC, phone, home, car, TV, edge device, or robot—with capabilities matched to the installation.
Interfaces beyond the screen
Explore voice, vision, spatial, and holographic interfaces as practical ways to understand and control an agent system.
Software and physical action
Use the same governed tool model for APIs, coding environments, browsers, simulations, and hardware interfaces.
Open to inspection
Keep execution, evidence, policy, evaluation, and deployment choices visible to the developers and organisations responsible for them.
Build the future in the open
Challenge the roadmap. Build one of its missing pieces. Show us a better route.
The most useful contribution may be a runtime improvement, a reproducible evaluation, a Super Agent, a deployment pattern, a robotics experiment, or a design that changes the plan. The vision is ambitious because it is meant to attract ambitious engineering.