Project summary

3Dogs Nexus is a live, commercially operating platform where high-stakes questions get run through a panel of independent AI models that challenge each other before answering. The system's stated confidence comes from how much genuine disagreement survived that challenge — not from any single model asserting how sure it is.

What are this project's goals? How will you achieve them?

We want to fund a bounded evaluation project: build a labeled corpus of high-stakes questions with known or adjudicated outcomes, run them through the existing multi-model deliberation panel versus a single-model baseline, measure calibration (Brier score, expected calibration error), and publish the methodology and results openly.

How will this funding be used?

Compute costs for running the evaluation corpus through both the multi-model panel and single-model baselines, plus corpus-labeling and adjudication time — roughly 50 to 100 labeled high-stakes questions across 2-3 domains.

Who is on your team? What's your track record on similar projects?

Alan Finney, founder. 35 years of professional experience, including 5 years in IBM enterprise software sales. Self-funded and built 3Dogs Nexus from scratch. The platform is in production today at app.3dogs.ai, not a prototype. A cross-domain validation case — a blind head-to-head against a leading single-model AI assistant — already exists as a public write-up.

What are the most likely causes and outcomes if this project fails?

The corpus proves inconclusive due to too few labeled cases or adjudicator disagreement, or the results show no calibration advantage over single-model baselines. Either way we publish the honest result, since a null result is still useful information for the field.

How much money have you raised in the last 12 months, and from where?

Self-funded and bootstrapped. No outside capital raised to date. We are currently pursuing several non-dilutive grant and credit programs in parallel, none of which have resulted in funds received yet.