SYS systems engineering for the AI era—MODE distributed—STATUS ● live

/01 — thesis

Models generate
answers.
We operate the
control plane.

We built the control plane for applied intelligence: memory, coordination, verification, governance across a whole portfolio. We run our own company on it. And we publish the receipts.

ENGAGE →SEE THE WORK

root@nxtg:~/agent-meshCrooked mode: ON

MANIFESTO_v01.txt● live

AI writes code.
We write systems.

Generated code clears the demo. Then meets auth, audit logs, threat models, governance, pen tests, on-call, SLOs, cost ceilings, regulatory review. And stalls.

We operate the plane it runs on.

property-based tests·policy-as-code·attestation pipelines

/02 — manifesto

CORE THESIS

The model writes the code.
Someone has to write the system.

You don't ship a prototype.

You ship an obligation:

to your users, your board, your regulators, your on-call.

The model's output is the easy part now.

The operating plane it lands on is where companies live or die.

Twelve principles hold the plane up.
Every line of the system answers to them.

01 / Security and privacy are not featuresWe threat-model LLM agents, prompt-injection surfaces, and synthetic-identity flows from the first design, not the first patch — and we treat your data as yours: minimized, guarded, never the product. Safety is the ground floor, not the penthouse.
02 / Intent is the contractBefore anything runs, we translate purpose, outcomes, and constraints into machine-checkable acceptance criteria. Property-based tests, contract tests, and eval suites serve that contract — they don't replace it. Vague AI output never becomes a silent assumption.
03 / Governance is infrastructurePolicy-as-code, model registries, scoped authority, attestation pipelines. Auditable by default. Every exception visible.
04 / Everything consequential is traceableEvery consequential action and artifact is content-addressed, logged, and observable end to end. If it happened, you can prove what happened, why, and under whose authority.
05 / Names carry weightEvery name is descriptive, unique, consistent, and never ambiguous. One name means one thing, everywhere — because ambiguity is where systems rot.
06 / Nothing acts blind to time or contextEvery operation knows the current moment and the timestamp of everything it touches, plus its environment, its fidelity mode, and its state — or it says plainly that it doesn't. Temporal and situational awareness are built in, never bolted on.
07 / Intelligence compoundsEvery outcome, failure, override, and recovery that earns admission sharpens the whole system. It grows wiser as it runs — not larger, not noisier, not staler.
08 / Built for machines and humans bothEvery surface is operable by agents and people alike; every signal preserves its context in both directions. The agent is a first-class user.
09 / No one outranks evidenceEvery input is checked against evidence and the system's expectations — the model's, the agent's, the founder's, and yours. Verified reality outranks confidence, assumptions, defaults, and legacy. We exist to save you from yourself.
10 / Quality rises with consequenceReversible experiments move fast; consequential actions demand stronger evidence, higher fidelity, explicit authority, and a way back. Quality is not one bar for everything — it is the bar each action's impact demands.
11 / You authorize what acts in your nameNothing acts as you without explicit, inspectable, revocable authority. Inside that scope the system acts without asking again; outside it, it stops or escalates. What you see is what you authorize.
12 / Reality outranks statusA thing is done when evidence from the target environment confirms it — not when a task, a dashboard, or a deployment says so. We verify against reality, never against our own claims.

/03 — ecosystem

PRODUCTS

One control plane.
We build it, run our company on it, and publish the proof.

Forgedogfood

AI coding tools generate the first 80%. Forge governs the 20% that ships: file locking, knowledge capture, drift detection, and delivery governance across Claude Code, Codex CLI, and Gemini CLI.

forge.nxtg.ai →live · waitlist open

Dx³internal · dogfood

The intelligence platform we run the whole company on: PostgreSQL, pgvector, and Apache AGE unified into one living store for vectors, lineage, and retrieval that survives audit.

read the roadmap →platform

GEO Graderalpha

AI-visibility scoring: see how large language models read, cite, and rank your site. Live and scoring today.

Score your site →live at geo.nxtg.ai

Faultlinealpha

LLM output verification: check AI claims against real evidence. Install it: npm i @nxtg/faultline.

@nxtg/faultline →v0.9.0 · on npm

FamilyMindbeta

The consumer product that makes cold AI warm: one shared place for your family's calendar and lists — groceries, meal plans, wishlists. Less mental load, more time together.

familymind.nxtg.ai →live · for families

LIVEpublicly probeable: every entry links to its receipt

0x24ForgedogfoodGovernance-native AI development: drift detection, CRUCIBLE auditing, and security guards for Claude Code, Codex CLI, and Gemini CLI.forge.nxtg.ai →
0x11Dx³internal · dogfoodThe data-intelligence platform we run ourselves. PostgreSQL, pgvector, and Apache AGE unified. No public URL by design; the published research is the evidence.the research →internal platform; receipts public
0x32GEO GraderalphaAI-visibility scoring: see how large language models read, cite, and rank your site.geo.nxtg.ai →
0x33FaultlinealphaAI claim verification CLI: paste what an AI told you; it extracts every claim, verifies against live sources, and scores the risk.@nxtg/faultline on npm →
0x34AtlasalphaPortfolio intelligence for AI engineering teams. Scan repos, score health, find cross-project patterns.nxtg-atlas on PyPI →
0x4FArcadebetaThe NXTG game studio. Two-Clue Mini is live today.nxtg.ai/arcade →sign in to play
0x22skinny.cloudalphaLean AI ops. Audit your Claude Code context in 60 seconds.skinny.cloud →
0x2ASynAppsalphaVisual AI workflow builder.synapps.nxtg.ai →

LABin the lab: active R&D, not yet public

/04 — insights

INSIGHTS

How we think about
the part nobody demos.

$cat ~/insights/*.md | headnxtg.ai/insights →

READ ALL INSIGHTS →

/05 — receipts

PUBLIC EVIDENCE

Not a portfolio. A paper trail.

PRE-REGISTEREDRECEIPT — 01

Pre-registered research, public data.

Two predictions pre-registered on OSF before the outcomes were known (8mh2x, 2nker, embargoed), plus three openly-deposited Zenodo datasets. Read the write-up in the origin report.