JMP0X1B / IF

IF / Auditable prospecting autonomy

Every mission decision should survive replay.

IF helps space teams rank resource-prospecting targets and defend the mission decisions behind them with source-backed uncertainty, safety-gated orchestration, and replayable evidence.

Request a prospecting study Review current evidence

First focus: PGM-likelihood screening for near-Earth asteroid candidates. IF reports evidence and uncertainty; it does not claim mineral reserves.

01 / Problem

The hard part is joining evidence without losing its meaning.

Target and architecture decisions combine orbital observations, spectra, vehicle limits, timing, communications, resources, and human approval. Each input carries a source, unit, frame, epoch, quality, and uncertainty. When the chain lives across spreadsheets and specialist tools, a changed assumption can make the final ranking difficult to reproduce or defend.

Input / 01

Evidence fragments

Orbital and spectral observations have different coverage, provenance, and inference limits.

Decision / 02

Weights hide policy

Accessibility, composition likelihood, timing, and risk change with mission policy and assumptions.

Review / 03

Scores lose history

A result without rejected alternatives, exceptions, and replay is weak evidence for an engineering review.

02 / Method

Preserve the decision from observation to approved action.

IF treats provenance, uncertainty, mission state, candidate action, and safety approval as separate facts. A learned recommendation cannot bypass the hard-safety boundary, and a higher-fidelity adapter cannot claim capability it does not declare.

Source Capture observation, license, unit, frame, epoch, and quality.
Score Evaluate independent axes and expose sensitivity to policy.
Scenario Connect the target to assets, tasks, resources, timing, and failures.
Approve Route candidate commands through typed state and hard constraints.
Replay Return sources, assumptions, rejected actions, exceptions, and outcome.

Target-screening axes

Accessibility / Δv
Spectral / PGM likelihood
Data quality
Mission timing
Uncertainty

03 / Evidence

Working orchestration software, explicit product gaps.

The public IF Entity Orchestrator is a tested research demonstrator. Its evidence supports software architecture and orchestration behavior. It does not establish a target's composition, physical mission performance, or readiness for flight.

Implemented

Public architecture

  • Versioned mission and graph contracts
  • Typestate-gated orchestration
  • Orthogonal command/action plane
  • Deterministic simulation and replay
  • Local and Ray execution adapters
Tested

Software behavior

  • Baseline and graph-model planning
  • Separate hard-safety validation
  • Qualification and replay gates
  • Authenticated mission control
  • Dual-scale 3D entity replay
Planned

Product qualification

  • Reviewed orbital/spectral pipeline
  • PGM-likelihood scoring validation
  • Higher-fidelity dynamics adapters
  • Hardware-in-the-loop evidence
  • Payload and mission integration

04 / Engagement

Start with the decision that must be made.

JMP0X1B is validating fixed, mission-specific work with space robotics, spacecraft, payload, science, and program teams. Each engagement names its evidence, scoring policy, exclusions, uncertainty, acceptance review, and data rights before analysis begins.

Engagement / 01

Target screen

Compare a bounded candidate population and identify which targets warrant deeper analysis.

Ranked evidence / sensitivity / review
Engagement / 02

Mission evidence study

Connect target evidence to a defined architecture, entity graph, resource constraints, and failure cases.

Scenario / exceptions / reproducible report
Future path / 03

HIL qualification

Validate a bounded IF workflow against partner hardware and named acceptance evidence when the interfaces are ready.

Adapter / test plan / qualification evidence

05 / Status

Build evidence in the order the mission needs it.

Current limitations

IF has no flight heritage, customer mission deployment, production training dataset, qualified orbital or physical simulator, hardware-in-the-loop result, payload integration, or proven asteroid reserve. The public simulator qualifies orchestration behavior only. Every future capability must pass its own scientific, numerical, safety, security, and mission acceptance gate.

Team Founder-led software and AI
Company Formation and commercial validation in progress
Product Public research demonstrator
Seeking Prospecting-study design partners

06 / Contact

Bring the target decision you need to defend.

A useful first conversation covers the decision, timing, current evidence, acceptance review, and the reason an external study might fail.