The University of Baltimore

Community Action Playbooks

Move from insight to action

Community Action Playbooks

Each playbook turns a human-approved insight action into a starting sequence for accountable planning, piloting, validation, and follow-through. Suggested owners are invitations—not commitments—and all actions remain proposed until a responsible organization accepts them.

Where the ideas come from: Action Playbooks begin with recommendations generated from approved community evidence and are published only after human review.

Draft a Pilot Plan

Build a private AI-assisted draft from an approved recommendation. Human review is required before use.

Playbook 1 · Develop a short, plain-language draft guide covering common AI privacy cautions, including examples of information to avoid entering into AI tools, and clearly label it for review rather than as final policy guidance.
Implementation timeline · quick win

Develop a short, plain-language draft guide covering common AI privacy cautions, including examples of information to avoid entering into AI tools, and clearly label it for review rather than as final policy guidance.

planned
Suggested ownerLocal public-interest digital literacy, privacy, or small-organization support partners
Assigned ownerCAILI Admin (interim)
Target date2026-09-06
Evidence context5 approved stories across 5 represented ZIPs

Execution sequence

  1. Confirm the need

    Review the approved brief and confirm the objective with affected participants before committing resources.

  2. Name accountable partners

    Invite CAILI Admin (interim); record who accepts responsibility and who must be consulted.

  3. Run a bounded pilot

    Define the audience, scope, accessibility needs, privacy safeguards, schedule, and a stop-or-revise decision point.

  4. Validate and document

    Review the draft with small-organization participants, privacy practitioners, and affected community members before distribution; document disagreements and needed revisions.

Validation measures

MVP targets are planning assumptions, not observed outcomes.

Action-specific check: Review the draft with small-organization participants, privacy practitioners, and affected community members before distribution; document disagreements and needed revisions.

  • Number and diversity of voluntary participants engaged in validation, reported separately from this pilot sample.
    Target: 15 voluntary participants across at least 3 ZIPs and 3 role contexts · Current: 0 real validation participants recorded
  • Participant-reported usefulness, clarity, accessibility, and trustworthiness of privacy guidance, tool comparisons, workshops, and workforce materials.
    Target: At least 80% rate the materials useful, clear, accessible, and trustworthy · Current: Not yet collected
  • Number of questions asked and answered during pilot learning sessions, with privacy-preserving thematic summaries.
    Target: At least 20 questions logged; 90% answered live or within 5 business days · Current: 0 pilot-session questions recorded
  • Participant-reported confidence in identifying information not to enter into AI tools before and after reviewing guidance.
    Target: Improve correct privacy-safety responses by at least 20 percentage points · Current: Not yet collected
  • Participant-reported ability to compare AI tools using costs, limitations, and privacy considerations.
    Target: At least 75% correctly apply the cost, limitation, and privacy rubric · Current: Not yet collected
  • Catalog verification results: number of existing relevant resources and initiatives confirmed, added, or found unavailable.
    Target: At least 12 verified records; 100% reviewed within the 90-day cadence · Current: 7 current verified records
  • Feedback from administrative workers and worker-serving organizations on the relevance and uncertainty framing of workforce-preparation materials.
    Target: 8 administrative workers and 2 worker-serving organizations; at least 70% rate materials relevant · Current: 0 consultations recorded

Useful starting resources

Source: Draft Community Insight Brief: AI Learning, Privacy, and Workplace Preparation. Includes synthetic pilot evidence Do not generalize these findings to all Baltimore residents.

Draft a pilot from this recommendation

Playbook 2 · Map existing public AI-learning, privacy, workforce, and tool-comparison resources before concluding that services are absent.
Implementation timeline · quick win

Map existing public AI-learning, privacy, workforce, and tool-comparison resources before concluding that services are absent.

planned
Suggested ownerResource-catalog steward or community information partner
Assigned ownerCAILI Admin (interim)
Target date2026-09-06
Evidence context5 approved stories across 5 represented ZIPs

Execution sequence

  1. Confirm the need

    Review the approved brief and confirm the objective with affected participants before committing resources.

  2. Name accountable partners

    Invite CAILI Admin (interim); record who accepts responsibility and who must be consulted.

  3. Run a bounded pilot

    Define the audience, scope, accessibility needs, privacy safeguards, schedule, and a stop-or-revise decision point.

  4. Validate and document

    Ask libraries, workforce providers, small-organization support providers, and community participants to verify coverage and identify missing or inaccessible offerings.

Validation measures

MVP targets are planning assumptions, not observed outcomes.

Action-specific check: Ask libraries, workforce providers, small-organization support providers, and community participants to verify coverage and identify missing or inaccessible offerings.

  • Number and diversity of voluntary participants engaged in validation, reported separately from this pilot sample.
    Target: 15 voluntary participants across at least 3 ZIPs and 3 role contexts · Current: 0 real validation participants recorded
  • Participant-reported usefulness, clarity, accessibility, and trustworthiness of privacy guidance, tool comparisons, workshops, and workforce materials.
    Target: At least 80% rate the materials useful, clear, accessible, and trustworthy · Current: Not yet collected
  • Number of questions asked and answered during pilot learning sessions, with privacy-preserving thematic summaries.
    Target: At least 20 questions logged; 90% answered live or within 5 business days · Current: 0 pilot-session questions recorded
  • Participant-reported confidence in identifying information not to enter into AI tools before and after reviewing guidance.
    Target: Improve correct privacy-safety responses by at least 20 percentage points · Current: Not yet collected
  • Participant-reported ability to compare AI tools using costs, limitations, and privacy considerations.
    Target: At least 75% correctly apply the cost, limitation, and privacy rubric · Current: Not yet collected
  • Catalog verification results: number of existing relevant resources and initiatives confirmed, added, or found unavailable.
    Target: At least 12 verified records; 100% reviewed within the 90-day cadence · Current: 7 current verified records
  • Feedback from administrative workers and worker-serving organizations on the relevance and uncertainty framing of workforce-preparation materials.
    Target: 8 administrative workers and 2 worker-serving organizations; at least 70% rate materials relevant · Current: 0 consultations recorded

Useful starting resources

Source: Draft Community Insight Brief: AI Learning, Privacy, and Workplace Preparation. Includes synthetic pilot evidence Do not generalize these findings to all Baltimore residents.

Draft a pilot from this recommendation

Playbook 3 · Co-design and pilot a beginner AI learning session with live question-and-answer support, potentially in a trusted community setting such as a library.
Implementation timeline · medium term

Co-design and pilot a beginner AI learning session with live question-and-answer support, potentially in a trusted community setting such as a library.

planned
Suggested ownerLibrary system or community-based learning partner
Assigned ownerCAILI Admin (interim)
Target date2026-11-05
Evidence context5 approved stories across 5 represented ZIPs

Execution sequence

  1. Confirm the need

    Review the approved brief and confirm the objective with affected participants before committing resources.

  2. Name accountable partners

    Invite CAILI Admin (interim); record who accepts responsibility and who must be consulted.

  3. Run a bounded pilot

    Define the audience, scope, accessibility needs, privacy safeguards, schedule, and a stop-or-revise decision point.

  4. Validate and document

    Recruit a broader, voluntary group of affected participants to test format, accessibility, language, scheduling, and whether human support improves approachability.

Validation measures

MVP targets are planning assumptions, not observed outcomes.

Action-specific check: Recruit a broader, voluntary group of affected participants to test format, accessibility, language, scheduling, and whether human support improves approachability.

  • Number and diversity of voluntary participants engaged in validation, reported separately from this pilot sample.
    Target: 15 voluntary participants across at least 3 ZIPs and 3 role contexts · Current: 0 real validation participants recorded
  • Participant-reported usefulness, clarity, accessibility, and trustworthiness of privacy guidance, tool comparisons, workshops, and workforce materials.
    Target: At least 80% rate the materials useful, clear, accessible, and trustworthy · Current: Not yet collected
  • Number of questions asked and answered during pilot learning sessions, with privacy-preserving thematic summaries.
    Target: At least 20 questions logged; 90% answered live or within 5 business days · Current: 0 pilot-session questions recorded
  • Participant-reported confidence in identifying information not to enter into AI tools before and after reviewing guidance.
    Target: Improve correct privacy-safety responses by at least 20 percentage points · Current: Not yet collected
  • Participant-reported ability to compare AI tools using costs, limitations, and privacy considerations.
    Target: At least 75% correctly apply the cost, limitation, and privacy rubric · Current: Not yet collected
  • Catalog verification results: number of existing relevant resources and initiatives confirmed, added, or found unavailable.
    Target: At least 12 verified records; 100% reviewed within the 90-day cadence · Current: 7 current verified records
  • Feedback from administrative workers and worker-serving organizations on the relevance and uncertainty framing of workforce-preparation materials.
    Target: 8 administrative workers and 2 worker-serving organizations; at least 70% rate materials relevant · Current: 0 consultations recorded

Useful starting resources

Source: Draft Community Insight Brief: AI Learning, Privacy, and Workplace Preparation. Includes synthetic pilot evidence Do not generalize these findings to all Baltimore residents.

Draft a pilot from this recommendation

Playbook 4 · Test a neutral AI tool-comparison format that presents costs, limitations, privacy considerations, and uncertainty without endorsing a specific tool.
Implementation timeline · medium term

Test a neutral AI tool-comparison format that presents costs, limitations, privacy considerations, and uncertainty without endorsing a specific tool.

planned
Suggested ownerDigital literacy or consumer-information partner
Assigned ownerCAILI Admin (interim)
Target date2026-11-05
Evidence context5 approved stories across 5 represented ZIPs

Execution sequence

  1. Confirm the need

    Review the approved brief and confirm the objective with affected participants before committing resources.

  2. Name accountable partners

    Invite CAILI Admin (interim); record who accepts responsibility and who must be consulted.

  3. Run a bounded pilot

    Define the audience, scope, accessibility needs, privacy safeguards, schedule, and a stop-or-revise decision point.

  4. Validate and document

    Have participants with varied levels of AI familiarity assess whether the comparison is understandable, balanced, current, and useful for their decisions.

Validation measures

MVP targets are planning assumptions, not observed outcomes.

Action-specific check: Have participants with varied levels of AI familiarity assess whether the comparison is understandable, balanced, current, and useful for their decisions.

  • Number and diversity of voluntary participants engaged in validation, reported separately from this pilot sample.
    Target: 15 voluntary participants across at least 3 ZIPs and 3 role contexts · Current: 0 real validation participants recorded
  • Participant-reported usefulness, clarity, accessibility, and trustworthiness of privacy guidance, tool comparisons, workshops, and workforce materials.
    Target: At least 80% rate the materials useful, clear, accessible, and trustworthy · Current: Not yet collected
  • Number of questions asked and answered during pilot learning sessions, with privacy-preserving thematic summaries.
    Target: At least 20 questions logged; 90% answered live or within 5 business days · Current: 0 pilot-session questions recorded
  • Participant-reported confidence in identifying information not to enter into AI tools before and after reviewing guidance.
    Target: Improve correct privacy-safety responses by at least 20 percentage points · Current: Not yet collected
  • Participant-reported ability to compare AI tools using costs, limitations, and privacy considerations.
    Target: At least 75% correctly apply the cost, limitation, and privacy rubric · Current: Not yet collected
  • Catalog verification results: number of existing relevant resources and initiatives confirmed, added, or found unavailable.
    Target: At least 12 verified records; 100% reviewed within the 90-day cadence · Current: 7 current verified records
  • Feedback from administrative workers and worker-serving organizations on the relevance and uncertainty framing of workforce-preparation materials.
    Target: 8 administrative workers and 2 worker-serving organizations; at least 70% rate materials relevant · Current: 0 consultations recorded

Useful starting resources

Source: Draft Community Insight Brief: AI Learning, Privacy, and Workplace Preparation. Includes synthetic pilot evidence Do not generalize these findings to all Baltimore residents.

Draft a pilot from this recommendation

Playbook 5 · Co-design practical workforce-preparation materials focused on administrative work, possible task changes, transferable skills, and worker questions about AI use.
Implementation timeline · long term

Co-design practical workforce-preparation materials focused on administrative work, possible task changes, transferable skills, and worker questions about AI use.

planned
Suggested ownerWorkforce development partner, labor-serving organization, or employer-support intermediary
Assigned ownerCAILI Admin (interim)
Target date2027-02-03
Evidence context5 approved stories across 5 represented ZIPs

Execution sequence

  1. Confirm the need

    Review the approved brief and confirm the objective with affected participants before committing resources.

  2. Name accountable partners

    Invite CAILI Admin (interim); record who accepts responsibility and who must be consulted.

  3. Run a bounded pilot

    Define the audience, scope, accessibility needs, privacy safeguards, schedule, and a stop-or-revise decision point.

  4. Validate and document

    Validate content with administrative workers and relevant worker-serving groups; avoid presenting predicted job impacts as certain unless supported by separate evidence.

Validation measures

MVP targets are planning assumptions, not observed outcomes.

Action-specific check: Validate content with administrative workers and relevant worker-serving groups; avoid presenting predicted job impacts as certain unless supported by separate evidence.

  • Number and diversity of voluntary participants engaged in validation, reported separately from this pilot sample.
    Target: 15 voluntary participants across at least 3 ZIPs and 3 role contexts · Current: 0 real validation participants recorded
  • Participant-reported usefulness, clarity, accessibility, and trustworthiness of privacy guidance, tool comparisons, workshops, and workforce materials.
    Target: At least 80% rate the materials useful, clear, accessible, and trustworthy · Current: Not yet collected
  • Number of questions asked and answered during pilot learning sessions, with privacy-preserving thematic summaries.
    Target: At least 20 questions logged; 90% answered live or within 5 business days · Current: 0 pilot-session questions recorded
  • Participant-reported confidence in identifying information not to enter into AI tools before and after reviewing guidance.
    Target: Improve correct privacy-safety responses by at least 20 percentage points · Current: Not yet collected
  • Participant-reported ability to compare AI tools using costs, limitations, and privacy considerations.
    Target: At least 75% correctly apply the cost, limitation, and privacy rubric · Current: Not yet collected
  • Catalog verification results: number of existing relevant resources and initiatives confirmed, added, or found unavailable.
    Target: At least 12 verified records; 100% reviewed within the 90-day cadence · Current: 7 current verified records
  • Feedback from administrative workers and worker-serving organizations on the relevance and uncertainty framing of workforce-preparation materials.
    Target: 8 administrative workers and 2 worker-serving organizations; at least 70% rate materials relevant · Current: 0 consultations recorded

Useful starting resources

Source: Draft Community Insight Brief: AI Learning, Privacy, and Workplace Preparation. Includes synthetic pilot evidence Do not generalize these findings to all Baltimore residents.

Draft a pilot from this recommendation