Cousin's AI Circulation

Tag: responsible AI

  • AI Is Becoming a Workflow Layer for Small Orgs

    Astute Intelligence Insights

    Do More of What Matters.
    The Big Picture Edition Vol. 2, No. 12 2026-08-21
    AI Is Becoming a Workflow Layer
    By Warren Wiggins — Astute Intelligence

    The Trend: AI Is Becoming a Workflow Layer

    This week’s announcements point in the same direction. Momentive introduced agents for nonprofit and association work. Wix described a multi-agent system for small businesses. Anthropic described connected workflows that can act across common business tools with approval gates. These are different products, but the pattern is similar: AI is moving from generating an answer to taking a series of steps. Momentive announcement Wix announcement Anthropic announcement

    That shift changes the question leaders need to ask. “Which model is smartest?” is still relevant, but it is not enough. We also need to ask: What can the system access? What can it change? Who checks the result? What happens when the service is down or the source is wrong?

    What It Means for Mission-Driven Organizations

    For a nonprofit, connected AI could help with routine work such as drafting, routing, translation, or internal search. It could also put trust at risk if it mishandles a donor record, misstates a program detail, or makes a service decision without context. Candid’s review of AI equity research found that only about 15% of surveyed nonprofits had an organizational policy for responsible AI use, and that smaller and midsize organizations often lack the time and process support to turn learning into governance. Candid on AI equity

    That is the equity gap in practical terms. Larger organizations can pay for training, privacy review, security controls, and staff time. Smaller organizations may be eager to experiment but have fewer people to check the work. NetHope’s governance research also describes a missing middle between broad principles and the operational tools nonprofits need. NetHope governance report

    The answer is not to wait for perfect tools. It is to fund and protect the boring work: a short policy, a current data map, a named reviewer, a simple incident log, and a manual fallback. A pilot should make those habits stronger, not hide them behind an impressive demo.

    Strategic Question of the Week

    If an AI system took over one repeatable task next month, what review step would protect the people your organization serves—and who would have the authority to stop the system?

    Weekend Read

    Read NetHope’s “AI Governance and the Nonprofit Sector: Mapping the Missing Middle.” It reviews 53 governance instruments and asks what sector-specific structures are missing between high-level principles and day-to-day nonprofit practice. Read the NetHope report

    If your team wants hands-on help thinking this through, you can book time with Warren. Book time with Warren
    Curated by Warren Wiggins | Produced with Perplexity Computer

  • How to Build an Approval-Gated AI Workflow for Small Teams

    Astute Intelligence Insights

    Do More of What Matters.
    Tool Time Edition Vol. 2, No. 11 2026-08-19
    Build an Approval-Gated AI Workflow
    By Warren Wiggins — Astute Intelligence

    The Technique: Build an Approval-Gated AI Workflow

    The most useful “tool” for a small team may be a clear workflow, not a new subscription. An approval-gated workflow lets AI gather, sort, or draft, but keeps a person in charge of an external message, a sensitive decision, or an irreversible action.

    You can build that pattern inside several tool families. Google for Nonprofits is pairing its Workspace tools with nonprofit-specific AI training and coaching. Google.org update Microsoft’s nonprofit work has emphasized Copilot skills, change management, and responsible AI governance. GeekWire coverage of Microsoft Elevate for Changemakers Anthropic’s small-business package emphasizes connected workflows with approval before sending, posting, or paying. Anthropic announcement

    The trade-off is fit. Google may be the easiest starting point for a team already living in Workspace. Microsoft may fit an organization built around Microsoft 365. Anthropic’s connected workflow approach may be attractive when a business needs actions across several apps, but more connections mean more permissions to review. None of these choices removes the need for a clear owner and a fallback process.

    Who It’s For

    Nonprofits: Use it for an internal grant-summary draft or a first-pass volunteer inbox sort. Keep beneficiary, donor, and case data out until your policy and vendor review are ready.

    Small businesses: Use it for a low-risk intake or follow-up draft. Keep contracts, payments, and customer promises behind human approval.

    Schools: Use it for an internal meeting summary or a family-facing draft that a staff member checks. Do not treat a fluent draft as proof that the underlying record is correct.

    How To Get Started

    1. Pick one task that repeats every week.
    2. Write the trigger and the final output in plain language.
    3. List the data the task needs. Mark sensitive data as off-limits for the first test.
    4. Choose the tool your team already understands: Google Workspace, Microsoft 365, or another approved option.
    5. Give the AI a narrow job: sort, summarize, or draft—not “run the department.”
    6. Add a human approval step before anything external or irreversible.
    7. Test 10–20 examples and record errors, missing context, and editing time.
    8. Decide whether to keep, change, or stop the pilot.

    Warren’s Take

    The risk is not only a wrong answer. It is a wrong answer that looks finished and moves through a connected system. NetHope’s governance review points to a gap in practical, sector-specific guidance, including data protection in low-infrastructure settings. NetHope governance report That is why I would rather see a small workflow with a visible reviewer than a big “AI transformation” plan with no stop button.

    The equity question is simple: can a five-person organization run this safely, or does it require staff time and technical help that only a larger organization can spare? If only the well-resourced teams can maintain the guardrails, the gap gets wider.

    Read the full breakdown and past editions on the site. Visit Astute Intelligence Insights
    Curated by Warren Wiggins | Produced with Perplexity Computer