You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
$5,000 left to reach the funding goal.
Platformology proposes a limited public evidence-review component, 12 neutral constructed examples and a local reproduction guide. It would help reviewers distinguish task instructions, source evidence, software checks and the claims those checks support. The underlying workbench is new and intended first to improve the company’s margins, then increase client capacity without adding employees. The grant request is restricted to the public component and neutral materials, usable without purchasing company services; it does not fund general client delivery or expose the full workbench. No new revenue, measured efficiency improvement, external adoption or validated safety result is claimed.
What are this project’s goals? How will you achieve them?
Platformology owns an existing local software workbench. I propose extracting a bounded evidence-review component, with neutral examples and documentation. It would preserve the original task and source references, distinguish software-check results from supported claims, and make unsupported success statements, incorrect citations and unsupported absence claims easier to inspect. The proposed assessment would use 12 constructed cases with stated criteria and report failures as well as successes. These would be educational, descriptive examples, not a representative benchmark or a controlled study. Only selected original code and neutral material would be considered for public release after owner and dependency-license review; private customer records, credentials and the full company repository are excluded.
How will you know it worked?
Before implementation, specify reference evidence relationships for 12 constructed examples across four documented failure categories: unsupported completion, incorrect references, treating missing evidence as proof of absence, and incomplete inventories. Show the same neutral source and candidate output in an ordinary presentation and an evidence-linked view. Record what the component exposes, misses or leaves ambiguous. Michael both implements and reviews the cases, so the report is descriptive, not independent, blinded or a causal study of human decisions. Publish actual failures and limits alongside any useful result.
How will this funding be used?
We request $5,000 for 100 estimated hours of Michael’s implementation and assessment work at a proposed $50/hour over approximately five weeks after funding: 20 hours ($1,000) defining neutral cases and criteria; 50 hours ($2,500) implementing the component and reproduction path; 30 hours ($1,500) running descriptive checks and documenting results, failures and limitations. This is planning compensation, not an existing salary or incurred expense. No new equipment, travel, hires, overhead or cloud services are included. The minimum and maximum request are both $5,000 for this full bounded scope. Related applications are alternatives; no same work would be charged twice.
Who is on your team? What’s your track record on similar projects?
Michael R. Drew would be the accountable implementer using AI coding tools. Platformology owns the existing local workbench. Internal development reviews identified problems motivating this proposal; the workbench is not production-qualified, and no external users or independent validation are claimed. Michael’s established professional background concerns publishing and author services. Four business team members are not being represented as software researchers. Public professional background: https://www.linkedin.com/in/michaelrdrew/ .
What are the most likely causes and outcomes if this project fails?
The neutral cases may not generalize; evidence mappings may be incomplete; the extra view may add complexity without making a claim easier to review; checks may miss the intended failure. Existing evaluation/logging tools may already cover the useful functionality. The report would retain negative findings and unresolved cases, with no promised improvement in model reliability, catastrophic-risk reduction or customer outcomes.
How much money have you raised in the last 12 months, and from where?
This new public component has received $0 in external grants or investment to date. The underlying workbench has generated no new revenue yet. Related applications remain unawarded: FUTO $1,000; Foresight $5,000; BlueDot $5,000; Lightcone $5,000. These are overlapping funding alternatives, not money received or additive budgets. Any support would be disclosed and reconciled before accepting funds or charging costs. This statement does not describe total company historical financing or claim that internal development had no costs.
Application preparation
AI tools assisted preparation of this proposal. Michael is responsible for its accuracy. Manifund recommends writing in the applicant’s own words and says its homepage filters out projects flagged as AI-written; this proposal does not claim human-only authorship.