4.5/5
Average on G2.
Governed AI
How do we use AI withoutlosing control of our donor data?
By putting the AI inside the platform your team already governs. Ignite’s agents work from your live record within an explicit scope, research and draft rather than act on their own, and hand writes to the approval gates your team configures. Agent actions are recorded. Prompts and outputs are not used to train foundation models by default. Your people make the decisions, and the agents do the preparation.
Governed AI, in short
- Governed AI is not a chatbot pointed at your donor file. Ignite’s agents run inside the platform, read the same record your team maintains, and show the sources behind their work.
- Each agent has an explicit capability scope. Anything outside that scope fails closed, so an agent cannot reach data or actions it was never given.
- Agents research and draft. Writes follow the approval gates your team configures, and agent actions land in the audit trail: what was proposed, who approved it, and what changed.
- The deterministic parts stay deterministic. Duplicate matching, prospect scoring, and lapse-risk segments run as engines and rules, not model guesses, and the segment rules are visible in the platform. AI is reserved for briefs, research, and judgment work.
- Your donor data is yours. Prompts and outputs are not used to train foundation models by default, models run on Azure-hosted endpoints, and data is encrypted at rest.
The AI policy is written. The tools are not governed.
Most nonprofits now have an AI policy. Fewer have a way to enforce it once a staff member pastes a donor list into a public chatbot to write an appeal.
Donor data in a browser tab
The fastest way to draft a thank-you is to paste the donor’s history into a general chatbot. Nobody meant to send constituent data to a third party, but that is what happened.
Nobody can say what the AI did
A record changed. A note appeared. Was it the intern, the sync, or the AI tool the vendor turned on last quarter? The audit question has no answer because there is no audit trail.
The board asks about risk
A trustee wants to know whether donor data trains someone’s model and who approves what the AI sends. The honest answer today is a shrug and a vendor’s marketing page.
The evidence
69%
of participating nonprofits report having policies, procedures, or guidelines for generative AI, up from 42% in the 2024 study.
Source: M+R Benchmarks 2026
What it means for your team
The sector has moved past whether to use AI and on to how to govern it. A policy on paper does not scope an agent, gate a write, or produce an audit trail. Governance has to live where the work happens: inside the platform that holds the donor record, with controls your team sets and can show a board or an auditor.
Follow one agent action from request to record.
Inside Ignite, this is the Propose, Review, Record workflow.
Governed AI means you can trace any agent action back to what was asked, what was proposed, who approved it, and what changed.
A person asks, within scope
A fundraiser asks an agent for research, a brief, or an answer. The agent works only from the live record and only within the capability scope it was given. Anything outside that scope fails closed.
- Does the work
- Ignite Intelligence
- Your team
- decides which agents are on and what each may touch
The agent researches and drafts
Prospect Research assembles the summary, Major Gift prepares the brief, Retention flags the risk, Constituent Support answers the question. Each shows the sources behind its work and separates evidence from inference.
- Does the work
- The Ignite agents
- Your team
- reads the work and the sources before anything moves
A write waits for the gate you set
When an agent proposes a change to the record, the approval gates your team configures decide whether it waits for a person. Gates apply at platform, tenant, or agent level, so a new agent can start conservative.
- Does the work
- Ignite Intelligence
- Your team
- approves or rejects the proposed write
The action lands in the record and the trail
Approved work lands on the constituent record like any other change, labeled AI-assisted, and the action is recorded: what was proposed, who approved it, and what changed. The trail is there when the auditor asks.
- Does the work
- Ignite CRM
- Your team
- reviews the trail, answers the auditor
The data stays yours
Prompts and outputs are not used to train foundation models by default. Models run on Azure-hosted endpoints, data is encrypted at rest, and access follows platform roles. Your security reviewer can walk the data flow with us.
- Does the work
- Trust & Security
- Your team
- runs the security review with the documents on the trust page
The Ignite modules underneath.
Every module works on the same constituent record, so this outcome is a workflow across the platform rather than an add-on.
Ignite Intelligence
Where the agents live: which are on, what each may touch, and the approval gates your team configures.
Explore Ignite IntelligenceIgnite CRM
The record every agent reads and every approved write lands on, with the trail of what changed.
Explore Ignite CRMIgnite Marketing
Where drafted appeals become sends. Agents draft; your team reviews and sends every email.
Explore Ignite MarketingConstituent Support agent
Answers plain-language questions from live data with the sources shown, so the answer can be checked.
Meet the agentProspect Research agent
Researches from the live record with sources shown, and finished research lands back on the record.
Meet the agentProven foundation
Built from the ground up on a modern, AI-first architecture.
Ignite is a new platform, built by the team nonprofits have trusted with their constituent and financial data for over 15 years.
20M+
Constituent profiles managed across the StratusLIVE platform family.
$1B+
In annual giving processed across the StratusLIVE platform family.
Aloha United Way
$3.5M+
Raised for Maui wildfire relief
Under 1 hour — emergency donation page live.
Read the Aloha United Way storyOn the StratusLIVE platform family“One source of the truth about our contacts, donors, and partners that is easily kept up to date.”
Robert W.
Director of Fundraising Operations & Digital Solutions
Verified review · G2
“A powerful tool that empowers our team members to not only enter new data but also see a historical perspective of our donors and corporate partners.”
Maria M.
Director of Investor Management
Verified review · G2
Governed AI
Bring this exact question to a 30-minute working session.
See one live workflow in 30 minutes, with real data and no slide deck.
Questions about governing AI in a nonprofit
The questions buyers ask about this outcome, answered without a slide deck.
Related outcomes
Can an Ignite agent change our donor data without anyone knowing?
No. Agent actions are recorded: what was proposed, who approved it, and what changed. Agents research and draft, and writes follow the approval gates your team configures at platform, tenant, or agent level. Each agent also has an explicit capability scope, and anything outside it fails closed, so an agent cannot reach data or actions it was never given.
Does our donor data train an AI model?
Prompts and outputs are not used to train foundation models by default. Your donor data is yours. Models run on Azure-hosted endpoints, data is encrypted at rest, and your security reviewer can trace the data flow with us before you commit. The trust page carries the current provider, encryption, and compliance details for a formal review.
How is this different from our staff using ChatGPT with donor data?
A general chatbot has no idea who your donors are until someone pastes them in, and then the data has left your control. Ignite’s agents work inside the platform, read the record your team governs, stay within their scope, and show their sources. Nothing needs to be exported, and every action is recorded where your administrators can see it.
Is the AI making decisions about donors?
No. Your staff decide, and the agents prepare. Prospect scoring, duplicate matching, and lapse-risk segments are deterministic engines and rules, not model guesses, and the segment criteria are visible in the platform. The agents use those facts to research, draft, and recommend, with evidence separated from inference, and a fundraiser decides the ask, the send, and the next step.
Does StratusLIVE have a SOC 2 Type II report?
Not yet, and we will not imply otherwise. SOC 2 Type II is a target StratusLIVE is working toward, no report exists today to share, and we will publish the scope and coverage details on the trust page when it does. Encryption at rest, role-based access, and the agent audit trail are available for review today.
Start the security review on the trust pageWho decides which agents our staff can use?
Your administrators. Agent subscriptions and permission levels are managed in the Intelligence Hub, so you choose which agents are on and what each may touch. Approval gates are configured by your team rather than fixed by us, which means a new agent can start conservative and earn broader permissions as your team trusts it.
Bring your AI policy to a working session.
We show one agent action end to end, from request to audit trail, in 30 minutes with no slide deck.