feedoracle.io Contact
  1. Home
  2. Research Center
  3. Collaborate

Explore What We Can Discover Together.

Research collaboration with an independent technology and research company.

We work with people who bring a technically meaningful question, a way to test it, or a result they want checked. This page explains what we look for and how a collaboration would run.

How to proposeResearch opportunities
Four technology fieldsYour research questionShared experimentEvidenceStop
A question from outside meets the four fields in one shared experiment. Evidence and a clean stop are both valid outcomes.

Research opportunities

Four technology fields and the work between them. Each comes with the kind of question we find worth testing.

  • AI & Autonomous Systems

    Field page
    • How should pre-action authorization respond when an agent's stated intent changes during a task?
    • Can one execution receipt cover tool calls made by several cooperating agents?

  • Blockchain & Distributed Infrastructure

    Field page
    • What would a stateless verifier need in order to confirm replay protection for a receipt?
    • How can a payment record and an execution record be bound so that a third party can check both?

  • Data & Intelligence

    Field page
    • How should a data service report a gap or a stale input so that an agent reacts correctly?
    • Does provenance survive when values from several sources are combined into one?

  • Quantum & Scientific Computing

    Field page
    • When a circuit is compiled for real hardware, how can we show that it still computes the same thing?
    • Which experimental claims can be restated as theorems and checked in Lean 4?
    • Edge AI and IoT (part of AI & Autonomous Systems): how long can a device safely authorize actions offline before its policy must be refreshed?

  • Cross-disciplinary research

    Research Center
    • How should claim-verification systems be evaluated against simple baselines and independent test sets?
    • Can post-quantum signatures sit beside classical ones in existing verification tools?

Provisional research questions, not confirmed findings or open calls. A good proposal can start somewhere else entirely.

How collaboration works

From a question to a decision, in six steps. A negative result ends a step just as properly as a positive one.

  1. 01

    Research question

    You describe the problem and why it matters. One page is enough.

  2. 02

    Initial technical review

    We check the fit with our fields, the evidence that exists and what we could contribute. We may decline, with reasons.

  3. 03

    Scope & hypothesis

    Together we write down the question, a testable hypothesis and what would count as a negative result.

  4. 04

    Experiment plan

    Method, data, infrastructure, authorization and cost are agreed before anything runs.

  5. 05

    Evidence & evaluation

    Results are recorded with their evidence class, including null and negative results.

  6. 06

    Next decision

    Continue, change course, publish or stop. Each is a valid outcome.

Each step ends at a decision point. Either side can stop there. This is a proposed way of working, not a contract and not an automated process.

What partners can propose

We do not list partners we do not have. These are the kinds of collaboration we are open to.

Independent technical validation
Check one of our published results with your own tools and tell us what you find.
Research hypotheses
A testable idea in one of the five research directions, with what would refute it.
Joint experiments
A shared experiment with success and failure criteria agreed before it starts.
Infrastructure integrations
Connect a system you operate to ours for a defined and limited test.
Edge and IoT evaluations
A proposal to evaluate offline-capable governance on devices you operate, subject to technical review and separate authorization. We do not provide hardware.
Algorithmic research collaborations
Shared work on methods for verification, search or optimization.
Reproducibility studies
Repeat a published verification step and report the result, whatever it is.

Submit a proposal

There is no form on this page. Send your proposal through our contact page; a short, structured message is enough to start.

What a useful first proposal contains

  1. Research problemWhat is unknown or unsolved, and why it matters.
  2. Technical fieldOne or more of the four fields, or the research direction it belongs to.
  3. Hypothesis or intended outcomeWhat you expect to find, and what result would prove it wrong.
  4. Available evidenceData, prior results or publications that already exist.
  5. Required infrastructureHardware, data access, compute or services the work would need.
  6. Expected collaboration formatReview, joint experiment, integration test, study or something else.

A plain structure you can copy

1. Research problem:
2. Technical field:
3. Hypothesis or intended outcome:
4. Available evidence:
5. Required infrastructure:
6. Expected collaboration format:
Send through the contact page

A proposal creates no obligation for either side. It is not treated as confidential and implies no funding or acceptance. Share confidential details only after both sides have agreed how to handle them.

Research principles

How we work, whoever we work with.

  • Independent engineering

    We decide which problems we work on, and we build our own technology.

  • Evidence before claims

    Nothing is called demonstrated until there is evidence someone else can check.

  • Reproducibility

    Methods, data and checks are written down so that a result can be repeated.

  • Clear experimental boundaries

    Every result says whether it comes from simulation, hardware, one run or many.

  • Responsible authorization

    Experiments that spend money, touch systems or act for someone run only with explicit approval.

  • Failures stay on record

    Failed experiments, open experiments and demonstrated results are kept apart, and all three are reported.