Co-Designing Organizational Intelligence
DC CAP Enterprise AI Initiative
April 6 – June 10, 2026
DC CAP SCHOLARS
Over the next 60 days, you are building this together.
You log what you observe. You surface what breaks. You teach what you learn.
You will write prompts that produce nothing useful. You will get outputs that miss the mark entirely. That is the process. Every iteration teaches something. The goal is fluency, and fluency requires practice.
Bad prompts and confusing outputs are not failures—they're data. Every output that misses teaches you something about how Claude thinks. This is how growth happens. We measure you on improvement, not perfection.
This pilot supports you in doing more of what only humans can do—judgment, relationships, mission alignment—and less of what pulls you away from that. Claude drafts. You own what goes out.
The world is moving. Our scholars need us to move with it.
Our scholars will graduate into workplaces where AI fluency is expected. The people who serve them should model it first.
DC CAP's strategic priorities require increased capacity across grant writing, program evaluation, partnership analysis, and data-informed decision-making. AI amplifies every one of those.
We are early to this. A $600K KPMG grant, 24 enterprise seats, a governance framework, and 24 custom skills already built. Most nonprofits are still talking about whether to start. We are starting.
What This Makes Possible
With Fluency
- Draft grant narratives in 30 minutes
- Build reusable project templates
- Audit outputs against org standards
- Scale one person's expertise across the team
Where Most Organizations Start
- "Can you just check if AI can write this?"
- Ad hoc requests with no framework
- One-off experiments that don't compound
- Trial and error without shared learning
How Claude Works
The 4Ds of AI Fluency
Claude's output is only as good as your input. The 4Ds are how you give better input and evaluate what comes back.
Delegation + Description
Delegation
Drafting follow-up emails to scholars is repetitive and structured. Let Claude start the template.
Description
Discernment + Diligence
Discernment
- Is the factual content accurate?
- Does the tone match the audience?
- Would you sign your name to this?
Diligence
- Classify before you upload
- Review before you send
- Disclose when relevant
Three Ways to Work with Claude
Automation
You direct. Claude executes.
Example: "Draft this email."
Primary Ds: Delegation + Diligence
Augmentation
You collaborate. Claude contributes.
Example: "Let's develop this grant narrative together."
Primary Ds: Description + Discernment
Agency
You architect. Claude operates.
Example: "Build a Project that any staff member can use."
Primary Ds: All four
Which Mode Are You In?
What's Already Built
The Data Classification Pyramid
The 60 Days
- Kickoff presentation (Day 1)
- Orientation sessions (Apr 8–9)
- Governance acknowledgment (Apr 10–12)
- First supervised Claude session (Apr 20–22)
- Baseline capture (Apr 22)
- Weekly activity cards begin
- Rotating team-led sessions
- Apply skills to real work tasks
- Peer feedback and cross-audits
- Train-the-Trainer candidates identified
- Governance framework review
- Capstone presentations (5 min each)
- 4D self-assessment
- Evaluation and board briefing prep
- Q1 rollout planning
What Happens Next
Three prerequisites before your first Claude session: (1) Review the Start Here Guide and AI Governance Framework. (2) Take the Pre-Launch Assessment. (3) Complete Claude 101 and AI Fluency for Nonprofits.
DC CAP SCHOLARS