AeionClaw Roadmap — Where the Platform Goes Next.
Public roadmap visibility matters more in AI than in any other platform category. Models change quarterly. Capabilities expand monthly. Compliance frameworks evolve weekly. Below is what AeionClaw has shipped, what's in active development, what's planned, and what's longer-term aspiration. Commitments aren't dates (dates are unreliable in AI) — they're priorities + investment direction.
Shipped + Stable (Q1-Q2 2026)
ZeroClaw conversational AI
native engine
47 module-specific skills
full catalog
57 AI providers
daily auto-refresh
8 auto-routing personas
RBAC-scoped
89 native MCP tools
REST/MCP/WebMCP/Forge surfaces
Event-driven Reactions
event-driven across all platform events
Smart Memory
semantic + 10 knowledge categories + RAG
Per-tenant cost caps
warn/soft-block/hard-block enforcement
9 commerce-AI agents
fraud, churn, cart-recovery, copy, pricing, inventory, upsell, loyalty, reviews
Skill Factory
Draft → Sandbox → Approved → Production
Skill kill-switch + deny list
instant per-skill disable + per-tenant deploy blocking
Immutable audit log
append-only — no update or delete path, for any role
SIEM export
stream audit + AI-action logs to Splunk / Datadog / Elastic / webhook
Streaming SSE responses
sub-second TTFT
WebMCP integration
browser-native tool execution
Forge MCP integration
agentic project management
Multi-tenant isolation
kernel-level per-tenant scoping
Honest no-op pattern
adapter failures are visible
In Active Development (Q3 2026)
Real-time voice conversations
sub-100ms TTS/STT via Cartesia + ElevenLabs streaming
Multi-modal vision agents
image + video understanding for support tickets (screenshot of error)
Cross-tenant analytics agent
for super-admins managing platform-wide observability
Voice-cloning consent workflow
formal consent + license tracking for AI voice cloning
EU AI Act conformity assessment automation
Governance module auto-generates required docs
Multi-agent collaboration
agents handing off context to each other (LangGraph-style but native)
Streaming tool calls
incremental result streaming for long-running tool invocations
AI-explainability dashboard
per-decision reasoning trace for regulated decisions
Federated knowledge
Aeion deployments can opt-in to shared (anonymized) knowledge base
Planned (Q4 2026)
Fine-tuning support
train custom models on tenant data (with explicit consent)
On-device inference
Aeion Bridge integration for local AI (Apple Silicon, NVIDIA Jetson)
Federated reasoning across tenants
opt-in cross-tenant insights with differential privacy
AI cost forecasting
predictive monthly AI spend per tenant + per skill
Skill recommendation engine
surface relevant custom-skill candidates from your usage patterns
Real-time embedding pipeline
sub-second indexing of new content
Multimodal output
generate image/video/3D from text within the same conversation
AI conformance testing
automated benchmark suite per skill, regression detection
Skill A/B versioning
gradual rollout of A/B'd skill versions with per-version rollback
Per-skill role restrictions
runtime enforcement (allowed-roles data model + SKILL.md seam ships today)
Long-Term Direction (2027+)
Multi-modal world models
Spatial + Cut + Claw composing 3D-aware agents
Cross-platform federation
Aeion deployments can opt-in to network-effect knowledge sharing
Autonomous Aegis-aware agents
agents that propose + execute backup-restore-time-travel actions safely
Fine-grained AI explainability
token-level decision attribution for regulated decisions
Quantum-resistant cryptography
for backup + audit log + customer-managed keys
Self-improving knowledge base
knowledge entries that update themselves from new conversations
AI fairness + bias monitoring
Aeionica-style fingerprinting for bias detection in AI decisions
AI sovereignty layer
for customers wanting to govern their AI policies across multi-cloud deployments
How the Roadmap Reflects Customer Voice
AeionClaw's roadmap isn't internal-team-curated — it's heavily customer-shaped.