You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
$600,000 left to reach the funding goal.
Corsen Academy proposes a free, multilingual learning lab that helps adults and educators check AI answers, trace claims to sources and make better judgments under uncertainty. We request USD 600,000 for the complete two-year programme, including a dedicated team and independent evaluation.
The existing French Academy catalogue lists 124 courses and 1,855 lessons, and the site exposes 24 locales. These figures describe published material. The new learning lab and its measured outcomes are the work we are proposing to fund.
The lab would combine claim-to-source exercises, comparison of conflicting evidence, confidence calibration and educator feedback. Learners would practice identifying unsupported conclusions and deciding what additional evidence is needed. Code, synthetic practice datasets, assessment rubrics and grant-funded materials would be released openly and remain free to use.
We would start with French and English, then add Portuguese and Spanish after review. Keyboard and screen-reader testing and low-bandwidth, text-first delivery are included. Learners would not need to purchase a paid AI subscription.
Months 1-3: recruit the team, define the study and review initial exercises with adults and educators. Months 4-9: release the French/English lab and run initial cohorts. Months 10-18: improve the tool from observed errors, add two reviewed languages and extend recruitment. Months 19-24: retention assessment, independent analysis, open release and an educator implementation guide.
The two-year target is 1,200 consenting adult learners, including 200 educators. These are recruitment goals, not existing users or committed partners. We would compare baseline and final performance on held-out tasks and test retention after eight weeks. We would report recruitment, attrition, uncertainty and limits of subgroup comparisons.
Success means that learners can trace claims, recognize unsupported conclusions and update their confidence on new tasks. Course visits and completion alone would not demonstrate that. We would publish the findings even if the intervention fails to improve outcomes. Private learner data and information about minors would not be published.
We request USD 600,000 for 24 months. All amounts below are prospective estimates, not current salaries or supplier quotations. Personnel allowances include employer costs.
Learning-design lead: USD 160,000 for two years. Research/software engineer: USD 160,000. Editor and cohort facilitator: USD 120,000. Independent learning-outcome evaluation: USD 40,000. Translation and accessibility review: USD 40,000. Hosting, study tools and delivery infrastructure: USD 20,000. Allocated founder oversight: USD 30,000. Explicit contingency: USD 30,000. Total: USD 600,000.
The full programme request remains USD 600,000. The minimum funding threshold is USD 500 so that useful initial work can begin before the full amount is raised. At that minimum we would produce one openly licensed claim-to-source exercise with its answer key and review notes, using existing Academy infrastructure. Larger partial awards would fund proportionate, agreed phases. The three dedicated roles and the full two-year programme would start only with sufficient funding; a partial award does not commit us to the full programme at a reduced budget.
Mordehai Attia is founder and president of CORSEN AI, a French SASU. Yoram Halberstam is co-founder and CTO. The team maintains Academy, Apache-2.0 WebMCP Explorer and MIT Corsen Context, including a published WordPress plugin. The proposed educational staff and evaluator would add expertise needed for the new programme.
Public work: https://academy.corsen.ai/fr/formations/ ; https://github.com/CorsenAI/webmcp-explorer ; https://github.com/CorsenAI/corsen-context
Professional profiles: https://www.linkedin.com/in/mordehai-attia/ ; https://www.linkedin.com/in/yhalberstam/
Our existing work establishes software and educational-content delivery. It does not establish a completed independent trial of this new learning lab. No outside educator, institution or participant is represented as already committed.
We may fail to recruit a sufficiently diverse group, teach skills that do not transfer beyond the exercises, or underestimate the effort required for reviewed translations and accessibility. Early signals include high dropout, weak performance on held-out tasks and learner feedback that the workflow is too difficult to use.
We would examine those signals after the first cohorts, revise the tool and teaching approach, and discuss material scope changes with funders before scaling. A documented negative result and reusable artifacts would still be released. We would not claim educational impact from content volume alone.
This new learning-lab programme has received no confirmed cash grant as of 13 November 2026. CORSEN AI is bootstrapped and maintains existing public resources. Company-wide financial records can be provided privately during grant diligence; no company cash balance is claimed here.
We are applying to several funders and request the full USD 600,000 from Manifund. Other applications remain pending, not secured financing. If several offers arrive, we will review their conditions and agree the allocation of costs and deliverables before accepting funds. No pending application reduces the amount requested here.
CORSEN AI is a for-profit company. The proposed grant would be restricted to the public-benefit programme described here, with free access and openly licensed outputs. We do not claim charitable registration for the company. The intended recipient and any required grant conditions would be confirmed with Manifund.