The opportunity

Three programs Oakland families already know

The foundation already puts books, meals, and play spaces into the hands of Oakland families. Each of those moments ends with a family holding something they can use. Family AI Coach asks what would happen if a small, guided activity travelled with it: five to ten minutes in which a caregiver and child read a short passage, ask a question about it, check an answer against the words on the page, and make something of their own. The proposal keeps the foundation in charge of the mission, the content standards, and the family experience. Salesforce and Agentforce would provide the proposed operational infrastructure, and 1440 is a potential messaging connector.

“Could every book, family event, and play experience become an opportunity to build confidence with reading and AI?”

The central question for foundation reviewers

Who this is for, and what would change

This is written for a caregiver in Oakland with limited time, possibly limited data, and a child somewhere between struggling and confident with reading. Today a book event ends with a book, and what happens after it is left to chance. Twelve weeks in, the proposal's aim is that a family has a habit they can name: read a bit, ask a question, find the proof, say plainly when the story does not tell us. That habit costs nothing to repeat and does not depend on the app that taught it. Whether it forms at all is exactly what a pilot would test.

EAT

Help caregivers find verified food resources and explore approved food or garden stories. Children practice checking explanations against evidence.

LEARN

Pair distributed books with educator-reviewed discussion activities. Children explain their interpretations, spot unsupported claims, and create alternate endings.

Existing program

Connects to book distribution and the foundation's literacy work with Oakland schools and reading partners.

Family activity

After a book event, a caregiver receives one discussion card. The child explains an idea, points to the words supporting it, catches a deliberate mistake in a practice answer, and writes a new ending that still fits the story.

Foundation would provide

Educator-reviewed discussion cards tied to the titles being distributed, in English and Spanish, with a printed equivalent for every digital card.

Learning skill

Explaining an interpretation, citing evidence, and noticing an unsupported claim.

eatlearnplay.org/learn

PLAY

Turn stories into cooperative games and adapt activities for different abilities. Connect caregivers with verified play opportunities.

Context for existing programs

meals served: 35M+meals servedSource
elementary students reached by investment in high-impact tutoring: 10,000+elementary students reached by investment in high-impact tutoringSource
rebuilt schoolyards: 25rebuilt schoolyardsSource

Foundation-reported figures reviewed September 6, 2026.

The tutoring figure describes investment in high-impact tutoring reaching that group of elementary students. It is not a count of completed tutoring placements. These figures are context for existing programs, not outcomes of this proposal.

Why this is worth exploring

This is worth exploring, not yet worth claiming. Children are already encountering AI systems that answer confidently and are sometimes wrong, usually without an adult nearby to model what to do about it. The skill of asking for evidence, noticing an unsupported claim, and saying "the story doesn't tell us" is a reading skill first and an AI skill second. A family activity can practise both at once using material the foundation already distributes. What we cannot say in advance is whether it works: this proposal makes no claim that AI improves literacy outcomes, increases participation, or reduces cost. Those are questions for a pilot, tested against equivalent printed activities that receive the same facilitator support.