From scattered experiments to capability
Most animal protection organizations already have some AI use underway: one staff member has a ChatGPT habit, another has tried Claude for writing, and others remain skeptical. Shared capability begins when the organization connects those experiments.
Capability means the organization can repeatedly notice useful AI opportunities, test them safely, learn from them, and scale what works. Moving from individual use to that shared practice requires organizational decisions about expectations, access, learning, and ownership.
AI tools will continue improving and changing quickly. Chat assistants, reasoning models, deep research, coding agents, and agentic coworkers have arrived in quick succession, so a static plan ages quickly. Build a team with two qualities: fluidity, the ability to move with change without needing a new strategy every week, and agency, the habit of acting on change before everything feels settled.
We presented this framework at AVA Summit 2026.
Where organizations are now
Before working the pillars, locate yourself honestly. An accurate starting point is more useful than an ambitious score.
| Stage | Description | What it often looks like |
|---|---|---|
| 0. Avoiding | AI is mostly unused, discouraged, or treated as too risky or confusing. | No shared tools, no policy, no clear ownership. |
| 1. Individual helpers | Some staff use ChatGPT, Claude, Gemini, or similar tools on their own. | Useful one-off work, but inconsistent practices. |
| 2. Shared practices | The team has basic norms, examples, training, or shared prompts. | More people know what is allowed and useful. |
| 3. Workflow pilots | AI is used inside repeatable workflows with human review. | Drafting, triage, research, reporting, or intake pilots. |
| 4. Strategic redesign | Roles, processes, planning, and resources are changing around AI. | AI is part of management, strategy, hiring, and operating rhythms. |
Most organizations we work with are at stage 1 or 2. Their next constraint usually sits in one of the six areas below.
The six pillars
Six areas need attention together. An organization that races ahead on tools while ignoring guardrails gets anxious staff and quiet mistakes. One that writes policy before anyone experiments gets a beautiful document nobody uses.
1. Narrative
People will form a story about AI whether leaders speak or stay silent. Silence leaves room for fears about layoffs, automation mandates, and hidden use. Leaders can instead frame AI around capacity, experimentation, human judgment, and impact for animals.
State that position early and ground it in current examples. Our guide to common staff concerns about AI covers the conversations this pillar usually requires.
2. Tools
The tool landscape includes several useful categories:
- General assistants: Claude, ChatGPT, and Gemini help with research, analysis, writing, and everyday questions.
- Agentic coworkers: Claude Cowork and Claude Code, along with ChatGPT Work and Codex, can work across files, tools, and multi-step tasks. Access varies by plan and platform.
- Automation platforms: n8n, Make, and Zapier connect tools and run repeatable processes.
- AI-native apps for specific jobs: Granola supports meeting notes, while Gamma creates presentations, documents, and lightweight websites. Many other specialized tools serve particular roles and workflows.
Begin with the work: what are we trying to improve, how sensitive is the data, how often does the task repeat, who will own the workflow after the first test, and what review standard applies before output is used? The answers narrow the tool choice and make responsible use easier. The tool overviews in this library are written for exactly this decision.
3. Training
Curiosity on its own reaches only a few people. Staff need tools, time, examples, training, and permission.
Useful training material is widely available online, and an AI assistant can help turn it into a curriculum for one person, a team, or the whole organization. Ask it to organize the material around real roles and tasks, then have someone knowledgeable review the plan before the team relies on it.
One model we have seen work well is an AI champion sitting down with individuals or teams, answering their questions, watching how they currently work, and showing them how to approach specific tasks differently with AI. This gives people practical help at the moment they need it and helps the champion spot patterns that broader training should address.
Training can also include protected experimentation time, demos, show-and-tells, shared prompts, or a small working group. Programs like Amplify for Animals offer outside support, and the broader movement AI ecosystem includes fellowships, hackathons, and learning communities.
4. Guardrails
Policy should make the safe path obvious. A useful guardrail structure has three zones: encouraged (internal, low-risk work), review first (sensitive data or anything external-facing), and off-limits for now (high-stakes decisions, regulated data, automations that act without approval).
Two principles anchor everything else: staff remain accountable for anything they publish, send, or decide, and sensitive data is protected by default. See examples of AI policies from movement organizations for real approaches, then use our policy worksheet to draft your own document section by section.
5. Workflows
This is where AI moves from individual help to organizational capacity. The progression could look something like: task (one prompt, done by hand) → workflow (the same pattern, saved and repeated) → system (AI plus process plus review, running on real work).
Write the process down before automating it. The team can then fix unreliable steps, decide where AI helps, and place review checkpoints before errors repeat at scale.
6. Rhythm
AI becomes organizational capability when it changes how the organization learns and manages work. Put it into the normal operating rhythm:
- One-on-ones: What did you try?
- Team meetings: What example should we share?
- Monthly reviews: What should become policy or a workflow?
- Quarterly planning: Which pillar do we focus on next?
A useful leadership standard
A strategy conversation should end with six concrete things: one priority area, one owner, one small experiment, one guardrail, one review date, and one next conversation with your team.
Where to start
Have your leadership team score the organization from 1 to 5 on each pillar independently, then compare. Discuss the largest gaps, choose one pillar, and give it a 90-day focus.
