Delt-IEt

Zero-label failure detection across any physical system

4 followers

Zero-label failure detection across any physical system

4 followers

Delt-IEt detects imminent failure in any physical system — no labels, no training, no domain parameters. One equation. Validated on NASA turbofan engines (100/100 detected), lithium-ion batteries (alert at 98.35% nominal capacity, 116 cycles before EOL), power semiconductors (zero false positives), and algae bioreactors. Same algorithm, four unrelated domains, zero retraining. Patent filed Brazil. Preprint: Zenodo. Kaggle notebook with real NASA data available

This is the 2nd launch from Delt-IEt. View more

FLINT Deterministic Work Engine

Launching today

Deterministic AI execution with validation before commit

A browser-executable work engine for AI-assisted operations where planning is separated from execution authority. FLINT freezes the baseline, runs only declared evidence, validates invariants before commit, issues OPEN/HALT, and retains cryptographic receipts. The launch demo exercises Transformer causal non-interference, positive/negative controls, pixel-integrity verification, tamper detection, and a planner-to-engine bridge.

FLINT Deterministic Work Engine gallery image

Free

Launch tags:Software Engineering•Developer Tools•Artificial Intelligence

Launch Team / Built With

OpenAI

GPT-6 Astra Challenge

What real task does your product handle with GPT-6 Astra?

Maker

GPT-6 Astra acts as the planning and orchestration layer for the FLINT Deterministic Work Engine. It carries long-running technical context across multiple workstreams, inspects the current system/repository state, identifies the next bounded task, proposes and sequences operations, and routes them through a defined operator bridge. FLINT then remains the execution authority: it performs the admitted work, validates declared invariants, issues OPEN/HALT, commits or blocks the result, and retains receipts and artifacts. Astra can then read that evidence and plan the next step. The real task is coordinated, multi-thread technical work that remains persistent, inspectable, and independently validated rather than letting the planner also be the unquestioned execution authority. Loving it all day long. ~ Flint

Report

2h ago

Maker

📌

I built FLINT around a basic engineering problem: generating an answer is not the same thing as proving what executed, what changed, what stayed fixed, and why a result was allowed to commit. This launch demonstrates that distinction as a live workstation. The operator path is: Identify source role → find source address → read typed boundary → state the question → declare candidate + invariants → run only needed evidence → compare to a frozen baseline → issue OPEN/HALT → retain the receipt. In the demonstration, the workstation applies that process to a Transformer causal-mask boundary. With the causal mask enforced, a future-token perturbation produces max_abs_diff = 0 and the workstation returns OPEN / AUTHORIZED. Bypassing the mask under the same intervention produces a nonzero delta and HALT / BLOCKED. The release also includes a deterministic pixel witness. The committed execution is projected into a canonical pixel artifact and cryptographically bound through framebuffer and tile hashes. A separate one-byte tamper control demonstrates that representation corruption can be detected independently from semantic execution failure. The larger architecture separates the planner from the work engine. A planner such as Astra can propose, sequence, and configure actions through the operator bridge, but the downstream FLINT system retains execution, validation, commit/block authority, persistent artifact state, and receipt generation. That distinction is deliberate: the existence of the Work Engine does not depend on what percentage of a successful workflow is attributable to a particular planner. Another qualified planner can address the same operator contract while the persistent execution and validation substrate remains underneath it. The public release, source, receipts, browser-validation evidence, hashes, demo, and walkthrough are archived at: https://doi.org/10.5281/zenodo.2...⁠� Related technical record: https://doi.org/10.5281/zenodo.2...⁠� I’m particularly interested in feedback from people working on AI infrastructure, agent systems, runtime verification, reproducibility, formal methods, technical assurance, and regulated or auditable AI workflows.

Report

5h ago

Previous Delt-IEt Launches

Delt-IEtZero-label failure detection across any physical system

Launched on May 26th, 2026