The infrastructure layer for AI agents

ADAMAS is an adaptive, decentralized architecture for multi-agent systems. It provides the isolation, memory, orchestration, and accountability that agents need to operate in the real world.

Most agent frameworks give agents tools and hope for the best. ADAMAS enriches their context with domain knowledge, execution history, and policy constraints — then records what worked so the next agent is smarter.

From Greek adamas — unconquerable. The origin of diamond.

From tool access to informed judgment

Finding tools isn't enough. Agents need context about when to use them, how they relate to the task, and what happened last time.

ADAMAS doesn't just help agents find tools. It enriches their context with related code, requirements, policies, and execution history — then records what worked so the next run is smarter.

"Here are 50 tool schemas. Load them all."
"Based on your task, here are 3 relevant tools, their implementation patterns, constraints that apply, and what succeeded last time."

Cumulative intelligence

Every execution makes the system smarter. ADAMAS records what tools agents used, what succeeded, what failed — and feeds that back into future discovery and routing.

Semantic Discovery

Agents describe what they need; ADAMAS surfaces relevant capabilities via vector embeddings and graph traversal. Not just tool names — related code patterns, requirements they satisfy, and policies that constrain them.

Thinking-Guided Injection

ADAMAS observes agent reasoning in real-time and injects relevant context mid-execution. When an agent starts thinking about authentication, authentication-related atoms appear in context — without the agent asking.

Graph Traversal

Discovering a tool means discovering its neighborhood: the code that implements it, the requirements it satisfies, related patterns from similar tasks. Context flows through the knowledge graph, not from static schemas.

Progressive Disclosure

Agents don't receive a catalogue of every available capability upfront. They discover what's relevant as their understanding of the task deepens — matching the natural rhythm of reasoning.

The ExecutionAtom Flywheel
Discover Enrich Execute Record Learn Discover (smarter)
Atomic KG
Core knowledge representation. Code, data, models, workflows, and decisions organized into discoverable atoms across five AEONS domains.
Ops KG
System observability as a knowledge graph. Agent health, telemetry, distributed traces, and spec drift detection across the network.
Chat KG
Conversational memory as structured knowledge. Threads, context, user preferences, and interaction patterns accumulated over time.

Accountable autonomy

Autonomy without accountability is a liability. ADAMAS gives agents independence while maintaining the audit trail that real-world deployment requires.

Policy-Constrained Tools

Dynamic permissions scope tools per-task based on intent. A "summarize this document" task gets read capabilities only — no shell, no network, no file writes. Constraints live in the graph and are enforced at discovery time.

Forensic Audit Trail

Every tool call, policy decision, and approval request logs with cryptographic agent attribution. ExecutionAtoms record what tools were used, what the agent was thinking, what changed, and why policies allowed it.

Human Approval Workflows

High-risk actions pause for confirmation via real-time NATS messaging. Configurable per tenant, per action type, or triggered by causal patterns like "read email then execute code."

Compliance-Ready

The same ExecutionAtom that powers learning also serves auditors. Answer "what did agent X do on task Y?" with a single query. The audit trail and the intelligence layer are the same system.

Secure isolation

Prompt injection will succeed. Security isn't about perfect prevention — it's about limiting blast radius when attacks land.

"Assume every agent is already compromised. Then design the architecture so it doesn't matter."
01

Hypervisor Isolation

Each agent runs in a dedicated Firecracker microVM with hardware-enforced boundaries. Not containers — actual hypervisor isolation with sub-second cold start. A compromised agent cannot escape its VM.

02

Network Jail

Agents cannot reach the internet directly. All external communication routes through an API proxy that enforces domain allowlists and injects credentials on the agent's behalf.

03

Credential Isolation

Agents never see API keys. The DGP Gateway holds credentials in Vault and injects them at the network layer. A compromised agent cannot exfiltrate secrets it doesn't have.

04

Cryptographic Identity

Agents authenticate with Ed25519 keypairs, not bearer tokens. Identity is unforgeable, revocation is instant, and every action is cryptographically attributable to a specific agent instance.

05

Input Boundaries

Structured prompts differentiate trusted instructions from untrusted data. Content from emails, URLs, and webhooks is tagged at ingestion, with elevated scrutiny for actions following untrusted input.

What works

Permission restriction — scoping limits tools per-task

Deterministic policy — rules that can't be prompt-injected

Monitoring for forensics — understand incidents, don't pretend to block them

Blast radius limitation — assume breach, contain damage

What doesn't

Prompt-based defenses — bypassed trivially, can't be patched

AI guardrails — another model to attack; fails at scale

Perfect prevention — attack surface is infinite

Architecture

Beyond the three pillars, ADAMAS provides the execution substrate for multi-agent systems.

Durable Workflows

Long-running agent tasks survive process restarts and network failures. State checkpoints automatically via Restate, with exactly-once execution guarantees for multi-step workflows.

Multi-Model Routing

Default to fast, cost-effective models and escalate to larger models for complex reasoning. Automatic routing based on task complexity reduces cost significantly for routine operations.

Orchestration Patterns

Sequential pipelines, parallel fan-out, iterative refinement, hierarchical delegation, self-modifying workflows. Patterns combine automatically — describe goals, ADAMAS selects coordination strategy.

Cost Transparency

Token accounting, operation costs, and execution time tracked at every layer. Full provenance through ExecutionAtom audit trails. Every action is attributable and budgetable.

Built on open protocols

ADAMAS is powered by open specifications that can be adopted independently.

DGP
Domain Graph Protocol

An agent-first protocol for exposing business domains as unified, queryable graphs. Cost-transparent operations with semantic filtering and relationship traversal designed for LLM consumption.

dgp.dev →
TURN
Temporal Unified Resource Notation

A notation system for multi-agent orchestration — the equivalent of musical scores for agent systems. The orchestration patterns above are expressed and composed through TURN.

turn.dev →

Agents in the world

ADAMAS is the brain of AI agents that operate in real domains. These are the first two — each backed by its own knowledge graph, each getting smarter with every interaction.

Satya
सत्य — Truth, what is real
Business Intelligence

Competitive analysis, market trend synthesis, customer intelligence, and due diligence. Satya accumulates and connects knowledge across time — the cumulative intelligence model in action. Every research session makes the next one deeper.

satya.ai →
Nishchinta
निश्चिंत — Worry-free, at ease
User Support

Contextual support with long-term user memory, knowledge base integration, and intelligent escalation. Nishchinta tracks user history, preferences, and past interactions — creating ease through understanding, not scripts.

nishchinta.ai →

Build on ADAMAS

ADAMAS handles the hard parts of AI application infrastructure: intelligent context enrichment, security isolation, policy enforcement, and execution learning. Your application talks to ADAMAS via DGP; ADAMAS handles everything else.

We're currently in private beta. If you're building AI-native products and need infrastructure where agents get smarter over time — not just safer — we'd like to talk.