Development Agent FAQs

Blackbaud is introducing a series of artificial intelligence agents that will act as staff support or “virtual team members” by autonomously executing complex workflows for nonprofits.

The first Agent for Good in the series, Development Agent, helps organizations expand donor engagement by building personalized relationships at scale, crafting outreach strategies, and managing cultivation plans with minimal human intervention. Unlike assistants that simply suggest actions, the Development Agent takes initiative, freeing up human fundraisers to focus on high-impact, personal outreach strategies. This innovation marks a major leap forward in fundraising automation, empowering nonprofits to raise more with greater efficiency and precision.

No, the agent is treated as a user with its own permissions and roles, not just a fundraiser.

Yes, Canadian organizations are supported.

It follows a phased approach:

• Start by assigning 10 donors to the agent.

• Review communications and get comfortable.

• Gradually ramp up the number of assigned constituents.

See Onboarding progress.

It helps users build confidence and avoid overwhelming situations like sending 1,000 emails prematurely.

See Onboarding progress.

You should monitor the Needs Attention section for escalated issues. The agent flags questions it shouldn’t answer so that you can step in.

See Escalated actions.

Actions are recorded when the agent completes an activity. In most cases, they appear shortly after, but display timing can vary slightly due to system processing and page refresh behavior.

Yes. Use Brand voice to teach the agent how your organization writes and Knowledge to give it information it can reference in communications. See What's the difference between brand voice and Knowledge? for details on what to add to each.

They serve different purposes and work best together:

• Brand voice controls how the agent writes. Use it to define your preferred tone, length, formality, and urgency, add words to avoid, and provide voice samples that show the agent your style.

• Knowledge controls what the agent knows. Use it to give the agent information it can draw on when interacting with constituents—such as your mission, upcoming events, active campaigns, and organizational history.

For example, brand voice might tell the agent to write in a warm, conversational tone and avoid jargon. Knowledge might tell the agent about your annual gala or a scholarship program a constituent helped fund.

Yes. If a brand voice entry contains factual information about your organization rather than an example of your writing style, move it to Knowledge. Select the menu next to the entry and choose Move to knowledge.

Although brand voice allows for a 30,000 character entry, a knowledge entries have a 10,000 character limit to ensure the agent can use factual information more precisely.

In Knowledge, add information the agent can reference in a conversation with a constituent. Examples include:

• Upcoming or recurring events

• Active or past campaigns

• Programs or services you offer

• Impact stories or outcomes you're proud of

• Historical facts about your organization

For time-sensitive entries like events or campaigns, use the Valid dates fields so the agent only references the content while it's current.

Note: Mission statements are sourced only from your Blackbaud Verified Network profile. Manual mission statement Knowledge entries are no longer used. To update your mission statement, see Manage Your Nonprofit Profile.

There are two ways to keep content current:

• Valid dates — Set a start and end date when adding or editing an entry. The agent will only use the entry while it's within the active date range. This works well for events, seasonal campaigns, or time-limited programs.

• Content review — Assign a reviewer and a review frequency (every 30, 60, 90, 180, or 365 days). The assigned reviewer receives a reminder to check the entry on the scheduled cadence.

Both fields are optional. If an entry is always valid and doesn't need a scheduled review, you can leave them blank.

Emails and SMS messages are dynamically generated based on the donor’s record. The content is personalized and not pulled from historical communications unless configured otherwise.

Yes. Use Brand voice to customize how the agent writes, and Knowledge to give it information it can reference in communications. See What's the difference between brand voice and Knowledge?

Yes, by default it uses the primary name and salutation, but you can adjust the preferred name and salutations.

Communications are generated approximately seven days before their scheduled send date to allow time for review. Thank You communications follow a shorter timeline and are generated closer to their scheduled delivery.

By default, each constituent receives an email or SMS roughly once every six weeks. You can configure the cadence in Engagement options.

The agent sends messages within defined delivery windows to ensure messages reach constituents at appropriate times based on regulatory guidance and engagement best practices:

• Email messages send between 10 AM and 6 PM Eastern Time (ET).

• SMS messages send between 11 AM and 7 PM ET.

For SMS, the earliest delivery time aligns with 8 AM Pacific Time (PT), which meets TCPA guidelines for the continental United States.

Communications now generate about seven days before their scheduled send date so they can incorporate recent Knowledge and Brand voice updates while still providing time for review before delivery.

For constituents assigned on or after July 29, 2026, an updated scheduling experience generates communications one at a time instead of creating a full year's communications plan upfront.

Constituents assigned before July 29, 2026 continue using their existing scheduled engagements and generated content. Once those communications have been sent, future communications follow the updated scheduling experience.

The agent doesn’t follow a fixed script or send asks after a set number of messages. Instead, it evaluates engagement signals, such as timing, past interactions, and responses, to determine when a solicitation is most appropriate. This helps ensure that asks feel timely, relevant, and respectful of each constituent’s relationship with your organization.

If more than 5 messages go back and forth in a conversation, it's escalated for human review. This ensures that complex situations or sensitive topics get appropriate attention from your human team.

Open the constituent’s record and select Actions.

Find an action created by the agent, then select more in the Action summary to expand the details.

You’ll see:

• Interaction Summary – a quick overview of the conversation

• Key Info – important details shared by the constituent

• Notes – context and any required next steps

If the interaction was escalated, the Notes section includes an Action Required message that tells you what to do next.

If the agent detects language that may indicate emotional distress or self‑harm, it automatically sends a brief, supportive message encouraging the constituent to seek help. It's important to note that the message does not assess intent, provide counseling, or continue the conversation.

Because the agent doesn't know a constituent’s location, it directs them to the global resource, Find a Helpline, which helps people locate free, confidential crisis support services.

For example, on SMS, a constituent may receive:

"It sounds like you're going through a really hard time. You don't have to go through this alone. Find free, confidential support at https://findahelpline.com."

These crisis‑support messages are sent automatically and are not monitored by Blackbaud.

Yes, you can change your donation form at any time.

To change the form, open the agent (On the left, select Portfolio, Development Agent), then on the right, select Engagement Types. Under Solicitation, enter the new donation form URL in the Donation form field and select Save.

Escalations can occur when a constituent asks complex questions or when the agent lacks enough information to respond confidently.

• If a constituent asks numerous or complex questions, you may be asked to review the interaction to determine whether human intervention is necessary.

• If the agent lacks sufficient information to respond to a question, you may be asked to assist by providing an appropriate answer.

When these situations arise, use your judgment to assess the need for human support and provide clear, accurate assistance to ensure the constituent’s questions are fully addressed.

See Escalated actions.

You can either take over the conversation or provide guidance for the agent to respond appropriately.

If an escalation isn't resolved, the conversation remains in an escalated state. Additional constituent messages are still associated with the escalation until it is addressed.

The agent tracks when an escalated interaction is addressed and uses that information to determine when normal conversation processing can resume.

Yes. While a conversation remains in an escalated state, the agent can continue sending automatic replies if it receives additional messages. After the escalation is handled, escalation-related automatic replies stop.

Yes, the goal is for the agent to adapt to important dates and giving patterns specific to each organization.

Reactivity refers to the system’s ability to adjust communication plans based on constituent actions or changes, rather than sticking to a pre-set schedule.

The agent uses select constituent and organizational information from your organization's Raiser's Edge NXT constituent records and, if available, Blackbaud Verified Network profile. While it uses information such as giving milestones and engagement insights to inform communications, it does not currently provide detailed donor history or donation record information during constituent conversations.

The agent currently uses information from your organization's Raiser's Edge NXT constituent records, such as:

Basic constituent details

• Name

• Primary salutation

• Birth date

• Religion

Contact information and consent

• Email address

• SMS phone number

• Email consent status

• SMS consent status

Giving and engagement insights

• First gift

• Latest gift

• Giving milestones

• Relationships

• Education

• Active constituent codes

• Donor persona

• Planned Giving Likelihood

• Sustainer Giving Likelihood

Note: While the agent uses this information to guide interactions, it doesn't currently provide direct access to detailed constituent records or donation history, such as specific gift amounts, fund designations, or complete donation records during constituent conversations.

If you have a Blackbaud Verified Network profile, the agent may also use:

• Mission statement

A Blackbaud Verified Network profile is optional. If your organization doesn't yet have a profile, you may receive a notification prompting you to add one.

All data remains secure and private.

The Development Agent uses a carefully selected set of constituent and organizational data to personalize communications and support engagement activities. Rather than using all available data in your database, Blackbaud reviews and selects data points that are appropriate for constituent outreach and engagement.

This approach helps ensure communications remain relevant, consistent, and aligned with responsible AI practices. It also recognizes that organizations may use certain fields for internal purposes, notes, or processes that are not intended to be used in constituent-facing communications.

As the Development Agent evolves, Blackbaud continues to evaluate additional data points and capabilities to ensure they provide value while maintaining appropriate safeguards.

The agent primarily uses data from your organization's Raiser's Edge NXT constituent records.

It can also access organization profile data from the Blackbaud Verified Network (YourCause), if your organization has a profile there.

When the agent detects a new donation, it adjusts upcoming engagements to prioritize a timely thank‑you.

• If an introduction has already occurred, the agent schedules a thank‑you engagement for two days later.

• To avoid over‑contacting the donor, the agent ensures there are at least 7 days between engagements, pushing other engagements as needed.

This happens automatically and helps ensure donors are acknowledged promptly while maintaining appropriate contact spacing.

When you provide thumbs up or thumbs down feedback on an engagement, the agent uses that feedback to update the current message review flow.

Thumbs up

A thumbs up confirms that the generated content met your expectations. The agent records the positive signal but no changes to generation behavior are triggered.

Thumbs down

A thumbs down flags that the content did not meet your expectations. You can add written feedback and regenerate the message.

• Thumbs down with regenerate: The feedback applies immediately to the regenerated output only.

• Apply this feedback to future engagements selected: The same feedback is also included in the process that improves future generated content for your organization.

• Apply this feedback to future engagements cleared (default): The feedback is used only for the current regeneration.

How feedback builds rules

When you select Apply this feedback to future engagements, written feedback is aggregated by the agent into a rules document specific to your organization. These rules capture your organization's preferences and act as guardrails that the agent applies when generating future messages.

How the feedback loop works during generation

Each time the agent generates a message, a grading step runs automatically:

1. The agent generates a message using your organization's rules.

2. A separate grading agent evaluates the output against those rules.

3. If the message passes, it proceeds through the normal workflow.

4. If the message fails, the grading agent identifies which rules were violated and provides a critique. The agent uses that critique to revise and retry the message.

5. If the message still does not pass after multiple retry attempts, it is escalated. You will see it in the Needs Attention section so you can review and handle it manually.

This process means that feedback you provide over time directly shapes how the agent writes for your organization, without requiring any manual configuration changes.

From the main Development Agent page, under Related Links, select Scheduled communications to review engagement details.

If significant changes occur (such as a new gift or a deceased record), the message may be regenerated and sent back for review before sending.

No, the review window is fixed at 7 days. Future enhancements may allow snoozing or accelerating the send. Thank You communications have a shorter, 48-hour review window.

Messages are generated about seven days before the scheduled send date, providing a seven-day review cycle before delivery. This timing allows communications to incorporate recent Knowledge and Brand Voice updates while still giving users time to review generated content.

See Why are communications generated closer to send dates?

The agent automatically adjusts to the time zone set in your browser. If your browser reflects your current time zone, the times shown will match your local day and time.

No. As of July 2026, Development Agent now schedules and generates communications one at a time rather than creating a full year's communications plan when a constituent is assigned. This allows communications to be generated closer to their scheduled send dates while still providing time for review before delivery.