Yes. The system prompt instructs the bot to prefer quoting from your campaign and brief posts over generating policy text. When a campaign post says "we support breaching the four lower Snake River dams based on the 2022 NOAA science," the bot uses that exact framing rather than paraphrasing. If a member asks about a position the site has not published, the bot says "we have not published a position on that yet" instead of inventing one. Position discipline is the credibility floor for environmental advocacy, and the bot respects it.

Yes. Each campaign typically has its own donation page or form variant for restricted giving. Those pages can be tagged in postmeta with the campaign slug, and the bot reads the mapping so a donor asking to give specifically to the salmon recovery campaign is sent to the correct form. Unrestricted general support has its own link. The bot will not silently route a restricted intent to an unrestricted form, because that defeats the donor's stated preference.

Yes, if chapters are stored as a custom taxonomy or post type. National environmental orgs often have state or regional chapters with their own campaigns, volunteer coordinators, and event listings. The bot reads the chapter assignment on each campaign and event post, and a visitor asking about Oregon-specific work gets Oregon chapter content first. Chapter contacts can also be surfaced for visitors looking to engage at the state level.

Yes. Display conditions can vary the bot's behavior by URL. On the press page or media room, the bot prefers to surface the communications director's email, the latest press release, and the named author of the brief in question. Reporters tend to ask precise attribution questions, and the bot routes them to the named contact rather than improvising a quote. The same applies to scientific inquiries that should go to a senior scientist by name.

Yes. If your site uses The Events Calendar, Event Espresso, or any post type with date and region fields, the bot reads upcoming events and quotes the next one by region. A visitor asking "is there a river cleanup near me in the next month" gets a list of dates, locations, and signup links. The bot can also filter by skill (e.g., kayak required, certified naturalist) when those fields are populated, so a volunteer is matched to a role they actually qualify for.

Yes, in the sense that it can quote the citation block from the brief post when asked for sources. The bot does not invent citations. If a member asks "what is your source for the 87 percent salmon mortality figure," the bot returns the citation as it appears in the brief, with author, journal, and year. If the citation is not in the brief content, the bot says so rather than fabricating a reference. Environmental orgs depend on citation integrity, and the bot respects it.

Yours. SleekAI is BYO API key. For a typical environmental nonprofit with steady traffic and a few campaign spikes per year, monthly model spend lands between $15 and $120. GPT-4o-mini and Claude Haiku handle most member and donor questions; larger models are worth it for nuanced press inquiries, multilingual briefs, or technical scientific exchanges. There is no per-message markup from SleekAI, just direct provider pricing.

It can listen and route, but it does not replace mental health support. The system prompt instructs the bot to acknowledge the difficulty of climate emotions without minimizing, and to point to resources like the Climate Psychology Alliance, the Good Grief Network, and 988 (the US Suicide and Crisis Lifeline) if the conversation crosses into acute distress. The bot does not attempt counseling. For most members, the right next step is a concrete action they can take through your campaigns, and the bot can offer that path as a constructive option.