delentia-os v1.0.2a0 (Public Alpha) launched in April 2026. This article is a technical summary of what shipped, what is currently being completed, and what the planned trajectory looks like through 2027 — with explicit scope boundaries for the open-source tier.
If you want the raw roadmap document, it lives at ROADMAP.md in the repository.
What Shipped: v1.0.2a0 Public Alpha (April 2026)
The April release was the first public availability of the RCT Platform SDK. What "public alpha" means in this context: the core algorithms are stable and tested; the API surface is not yet considered stable for production dependencies without pinning the exact version.
Core SDK modules (stable behavior, alpha API):
| Module | Path | What it provides |
|---|---|---|
| FDIA Scorer | core/fdia/ | Constitutional scoring: F = D^I × A, pure-function implementation, no external calls |
| SignedAI Registry | signedai/ | HexaCore 7-model registry, TIER_S/4/6/8 consensus, Ed25519 attestation |
| Delta Engine | core/delta_engine/ | State compression, warm recall, 74% compression rate on typical agent context |
| JITNA Protocol | rct_control_plane/jitna_protocol.py | RFC-001 implementation, 6-field intent schema, multi-agent routing |
| Regional Language Adapter | adapters/ | 8 ASEAN language pairs |
| Control Plane DSL | rct_control_plane/ | 15 modules, rct CLI entry point |
Reference microservices (5): intent-loop, analysearch, vector-search, crystallizer, gateway-api. These are reference implementations — they show how to compose the SDK modules into a service, not production-ready deployments.
Quality benchmarks at release:
- 1,272 tests passed, 0 skipped, 0 warnings, mypy clean, ruff clean, 94%+ coverage
- Bandit security scan: 0 HIGH severity findings
- CI/CD: GitHub Actions with ci.yml + security-scan.yml on every push
- Whitepaper: 450+ pages, bilingual EN/TH, CITATION.cff for academic attribution
Currently In Progress: v1.0.3a0 Playground (May 2026)
The v1.0.3a0 milestone is a usability release with one goal: zero-friction first experience. A new user should be able to run FDIA, SignedAI, and Delta Engine without cloning the repository or installing anything locally.
Already completed:
notebooks/rct_playground.ipynb— runnable Colab notebook with live FDIA, SignedAI, and Delta Engine demonstrationsbenchmark/run_benchmark.py— unified benchmark runner CLI for reproducible performance testing- Binder / Colab / Codespaces quick-launch badges in README
- Hypothesis property-based tests — mathematical correctness guarantees for FDIA, Delta Engine, SignedAI
Remaining for v1.0.3a0:
docs/benchmark/hallucination-methodology.md— the 100-prompt public dataset and methodology description (enables external reproduction)- GitHub Discussions activation in the GitHub UI — Q&A, RFC Discussion, Show & Tell categories
- API stability guarantees documentation for
core/fdia,signedai/core,core/delta_engine
v1.0.0 Stable — PyPI Release (Q3 2026)
The v1.0.0 milestone marks the transition from "public alpha with pinned version required" to "production-ready dependency with stability guarantees."
What stability means in practice:
pip install delentia-osworks from PyPI without cloning or building locally- Semantic versioning guarantee: no breaking changes without a major version bump
- Pre-built wheels for Python 3.10, 3.11, 3.12
- Complete type stubs (
py.typedmarker + full__init__.pyi) - GitHub Release with signed artifacts for supply-chain verification
- Full API reference documentation in
docs/api/ - Community-verified external reproduction of the hallucination benchmark
What does not change at v1.0.0: the underlying algorithms. FDIA, SignedAI, Delta Engine, and JITNA behaviors are already stable in the alpha — v1.0.0 is about the API contract, packaging, and documentation, not about changing how the algorithms work.
v1.1.0 — Observability and Integrations (Q4 2026)
This is the release that moves delentia-os from "SDK you use in your code" to "infrastructure you run alongside your systems."
Observability stack:
- Prometheus
/metricsendpoint, live inrct serve - Grafana dashboard template (
docs/assets/grafana-dashboard.json) — pre-built panels for FDIA score distribution, SignedAI consensus latency, Delta Engine compression ratios, JITNA routing decisions docker-compose.monitoring.yml— Prometheus + Grafana local stack, one command to run the full observability setup locally- OpenTelemetry trace exporter for distributed tracing in multi-service deployments
Third-party integration adapters:
- n8n integration adapter — bring RCT Platform capabilities into n8n automation workflows
- Home Assistant integration adapter — local AI reasoning for home automation
- Obsidian plugin — knowledge graph to JITNA intent tagging (the platform becomes your knowledge OS, not just a code library)
Protocol update:
- JITNA Protocol v2.1 draft — bidirectional agent negotiation. The current RFC-001 is one-directional (request → route → execute). v2.1 adds bidirectional negotiation: agents can propose alternative interpretations of their intent and negotiate with the routing layer before committing to an execution path. This is the foundation for multi-agent collaboration rather than just multi-agent parallel execution.
v1.2.0 — ASEAN Expansion (2027)
The 2027 release target addresses the gap between the platform's regional language capabilities (8 ASEAN pairs in v1.0.2a0) and full ASEAN regulatory alignment.
Language expansion:
- VN (Vietnamese), ID (Indonesian), MY (Malay) adapter additions, bringing the total from 8 to 11 ASEAN language pairs
- These are not translations — each adapter includes region-specific intent vocabulary and JITNA routing adjustments for regional communication styles
Regulatory compliance modules:
- PDPA (Thailand) compliance module with audit evidence export — structured export of constitutional enforcement decisions for PDPA compliance reporting
- PIPL (China) adapter for E3 model slot in the HexaCore roster
- ASEAN AI Governance checklist alignment
Community infrastructure:
- RCT Platform certification program (community-driven) — a framework for organizations to certify that their deployment meets specific constitutional compliance standards
- JITNA Protocol RFC-002 — cross-platform agent identity standard. This addresses the problem that arises when multiple organizations deploy
delentia-osinstances and need their agents to communicate across organizational boundaries without sharing their underlying knowledge stores
What We Are Not Building (Open Source Tier)
This section sets explicit expectations:
| Not in scope for open source | Where it exists | |---|---| | Full production microservice stack (62 services) | Enterprise tier — delentia.com | | Genome Creator Profile API | Enterprise proprietary | | Hosted inference engine | Hardware/cost constraints outside OSS scope | | Hosted API / SaaS endpoint | Runs at delentia.com — enterprise licensing | | Full Vault-1068 production deployment | Enterprise — requires organizational knowledge governance |
The open-source tier is a developer SDK — the algorithmic building blocks. The full production system (62 microservices, genome creator profiles, hosted inference, Vault-1068 as a managed service) is the enterprise platform at delentia.com.
This distinction is not about withholding capability — it is about honesty regarding what an SDK can reasonably provide vs. what requires operational infrastructure, organizational governance, and SLA commitments.
How to Shape the Roadmap
- GitHub Discussions — once enabled after v1.0.3a0, this is the primary channel for Q&A, RFC Discussion, and Show & Tell. The three categories are structured to capture different kinds of community input.
- GitHub Issues — bugs, specific missing features, and documentation gaps
- Issue upvotes — high-engagement issues move up priority. If a feature matters to you, finding the existing issue and reacting to it is more useful than creating a duplicate.
- Enterprise timeline requests — if your organization needs a specific feature by a specific date, contact founder@delentia.com. Enterprise commitments can accelerate specific items.
One explicit note on timeline realism: v1.0.3a0 is a May 2026 target with most items already shipped. v1.0.0 stable is a Q3 2026 target — the PyPI packaging and API stability documentation are well-understood work items. v1.1.0 (Q4 2026) and v1.2.0 (2027) are planned with less certainty — community input and enterprise commitments will influence prioritization within those milestones.
Where to Start Today
If you want to use delentia-os now:
git clone https://github.com/delentia-labs/delentia-os.git
cd delentia-os
pip install -e ".[dev]"
pytest # 1,272 tests should pass
Or open the Colab notebook directly for a no-install exploration of FDIA, SignedAI, and Delta Engine:
Related articles: 7 Genome System Architecture · FDIA Equation Explained · SignedAI Multi-LLM Consensus · Delta Engine 74% Compression
What enterprise teams should retain from this briefing
delentia-os v1.0.2a0 is live. Here is what was shipped, what is in progress, and what is coming in v1.0.3a0, v1.0.0 stable, v1.1.0 Observability, and v1.2.0 ASEAN Expansion — and what we are explicitly not building in the open-source tier.
Move from knowledge into platform evaluation
Each research article should connect to a solution page, an authority page, and a conversion path so discovery turns into real evaluation.
Previous Post
ASEAN Enterprise AI Deployment Guide: Governance, Compliance, and Regional Scale
Deploying enterprise AI across ASEAN means operating under 10 different data protection regimes, 8 primary languages, 6 distinct cloud sovereignty requirements, and one shared expectation: that your AI system is explainable, auditable, and compliant at every step. This guide maps the ASEAN AI governance landscape onto practical deployment architecture.
Next Post
Regional Language Adapter: Thai NLP in Enterprise AI
Thai is not a simplified version of a language — it has no spaces between words, stacked tone markers, 44 consonants with positional meaning, and PDPA-sensitive personal data patterns embedded in everyday syntax. The RCT Regional Language Adapter solves this at the enterprise layer without compromising governance boundaries.
Ittirit Saengow
Primary authorIttirit Saengow (อิทธิฤทธิ์ แซ่โง้ว) is the founder, sole developer, and primary author of Delentia Labs — a constitutional AI operating system platform built independently from architecture through publication. He conceived and developed the FDIA equation (F = (D^I) × A), the JITNA protocol specification (RFC-001), the 10-layer architecture, the 7-Genome system, and the RCT-7 process framework. Public-facing proof uses public sdk verification lane at 1,791 tests, while the broader runtime footprint is disclosed separately as an enterprise runtime snapshot.