In January 2023, leaders at Salesforce looked around and saw a world that was changing. Defying the initial predictions of its detractors, crescendoing ChatGPT usage had already pushed the platform past 100 million active monthly users only two months after its launch, and LLMs in general were showing every sign of being poised to take the world by storm. But at the same time, nonprofits were contending with what has by now become a familiar reality: rising demand, shrinking resources, and limited capacity to experiment with emerging technology.

Seeking to address those widening funding gaps and deficits of technical support, Salesforce launched the Salesforce Accelerator — Agents for Impact: a cohort-based program designed to help nonprofits responsibly adopt AI through a combination of unrestricted funding, technical coaching, pro-bono support, and access to Agentforce, Salesforce’s enterprise agentic AI solution. By leveraging its organizational superpowers, Salesforce set out to help nonprofits experiment with, implement, and scale the AI solutions that hold the potential to help them better deliver on their missions and meet their current demand.

“Historically in the social sector, the organizations that are on the front lines are left behind in technological revolutions.”

— Amy Guterman, Senior Director, AI for Impact at Salesforce

The Problem: Nonprofits are under increased pressure to adopt transformative new technologies at a moment when they are also contending with persistent resource constraints. While AI tools hold the potential to rapidly reshape how organizations operate, many frontline organizations lack the technical expertise, implementation support, flexible funding, and organizational capacity to safely experiment with them — despite the irony that they often serve the very communities most likely to be affected by the emergence of new tech.

Those challenges are compounded by a philanthropic landscape that has historically tended to underfund operational infrastructure and devalue early-stage experimentation, leaving many nonprofits without the resources needed to responsibly test and implement new tools.

The Solution: Salesforce designed the Accelerator program as a comprehensive support system that uses three primary levers to support nonprofits: unrestricted grant funding (typically between $200,000 to $400,000); access to Salesforce’s proprietary technology, including Agentforce; and hands-on technical guidance from Salesforce’s own employees. Through an 18 month, cohort-based model, participating nonprofits receive strategic coaching, implementation support, and training in the form of a dedicated six-month curriculum on issues like governance, data strategy, and agentic AI best practices that they can convert into mission-aligned deployment.

That pro bono support has become one of the program’s defining features. Rooted in Salesforce’s “1-1-1” philanthropic model — which commits 1% of the company’s equity, product, and employee time to supporting nonprofits and schools — participating organizations are paired with volunteer technical architects, project managers, and solutions engineers from Salesforce who work alongside them as ad hoc consultants throughout the implementation process. According to Salesforce, these volunteers function not only as technical advisors but as strategic thought partners, helping nonprofits build the confidence, governance structures, and strategy needed to use AI tools effectively and responsibly.

Why It’s Different: 

“A check is great. The technology is great. But unless you have the pro bono volunteers — the technical experts helping you best use the funding or best use the technology — you either don’t use it, or you burn through all your funding hiring consultants before you’ve even built the solution.”

— Amy Guterman, Senior Director, AI for Impact at Salesforce

Rather than simply providing funding or software licenses, the Accelerator was intentionally designed around the understanding that many nonprofits lack not only the internal capacity to independently navigate rapidly evolving AI systems, but the space to experiment, navigate challenges, and determine ROI prior to fundraising. This “risk-tolerant” support, in particular, is what sets the Accelerator apart: Rather than requiring organizations to arrive with fully articulated use cases, the program was designed as a space for learning and experimentation — a proving ground that is critically needed given the rapid pace of technological change.

And in the spirit of that learning, the Accelerator also has broader ambitions to become an ecosystem that acts as a central repository for those collected insights around responsible AI implementation, allowing organizations to share in each others’ learnings, avoid duplicating each others’ mistakes, and build on one another’s successes over time.

Impact: 

  • $16 million deployed through the Accelerator to date
  • 100% of participating nonprofits reported increased AI capacity
  • 94% of organizations predicted that the Accelerator would have a meaningful impact on mission delivery
  • The Accelerator has expanded internationally, including new India- and UK-based cohorts
  • 94% of Salesforce volunteers reported improved AI and agentic-AI skills through participation, a powerful secondary outcome

Some of the clearest evidence of the Accelerator’s impact has been reported anecdotally by the program’s early participant organizations:

College Possible: College Possible — an organization focused on expanding college access by helping first-generation students finish post-secondary education — reported that the Accelerator had helped to usher in a 400% increase in its coach-to-student efficiency and significantly reduce costs per student, even amid budget constraints that initially threatened to derail operations.

Good360: Another participant organization, Good360 — which distributes in-kind donations from major retailers and corporations to disaster-affected communities — reported using AI tools developed through the Accelerator to save their disaster recovery team over 1000 hours annually, connecting donated goods with communities 3x faster. 

“Our premise is that even if the solutions the cohorts are developing aren’t successful, at least they’re building the capacity and the skills and that thought process to their other work, so that the skills around AI capacity are durable to other projects in the future.”

— Amy Guterman, Senior Director of AI for Impact at Salesforce

Key Enablers:

  1. Salesforce’s 1-1-1 model: Baked into Salesforce’s core value system, the 1-1-1 model grants employees 56 hours of heavily-encouraged, paid volunteer time off annually, which is what allows the company to mobilize teams of technical experts to work directly with participating nonprofits. That pro bono support has proved especially valuable in helping organizations navigate the “fuzzy front end” of AI adoption: defining viable use cases, building governance frameworks, pressure-testing strategy, and developing confidence around responsible implementation before investing significant resources into full deployment.
  2. Risk-tolerant capital: While traditional philanthropy is often hesitant to fund technological experimentation before the outcomes are fully proven, the Accelerator was specifically designed to absorb some of that early uncertainty and give nonprofits the room to experiment responsibly before needing to demonstrate clear ROI.
  3. Embedded technical expertise: According to Salesforce, the real secret sauce of the Accelerator is the integrated recognition that it’s not just the funding and technology itself that nonprofits lack, but sustained support around enablement. By pairing participating organizations with Salesforce volunteers who can offer personalized guidance, nonprofits are better equipped to clarify strategy, pressure-test ideas, develop governance frameworks, and build confidence around responsible AI deployment before investing significant resources into full-scale deployment.   

Future Plans:
Beyond acting as a nonprofit support program, Salesforce’s systems-level ambition is for the Accelerator to help shift how the philanthropic sector approaches the adoption of emergent technologies. Over time, the company hopes that the model will encourage more risk-tolerant investment in AI experimentation while simultaneously creating stronger mechanisms for nonprofits to share lessons they’ve learned, their implementation strategies, and any evidence they’ve seen of impact. According to Salesforce, the ultimate goal is to reduce duplicative efforts across sectors and help organizations build on existing successes rather than repeatedly funding similar early-stage experiments in isolation.

Lessons for Other Leaders:

  1. Pair funding with implementation support. Providing capital or technology alone is often insufficient for organizations navigating complex technological change. Embedding technical guidance, coaching, and governance support is the lever that drives long-term sustainability and helps organizations dealing with persistent capacity constraints take their impact to the next level.
  2. Treat operational technology as mission-critical. As AI tools and data systems become increasingly embedded in organizational workflows, investments in data systems and implementation support should be treated as core mission support.
  3. Leverage your area of corporate expertise. Rather than creating a generic grant program, Salesforce designed the Accelerator around its expertise in AI and technical product infrastructure, and also leveraged its deeply rooted culture of employee volunteerism — all assets where it can provide unique value. Relying on institutional superpowers and values embedded deeply in the DNA of the company has allowed for more hands-on guidance and proficiency than capital alone could ever provide.
  4. Create mechanisms for organizations to share lessons learned and observed successes. Cohort models, peer exchange, and open sharing of implementation lessons can help organizations avoid duplicative experimentation and accelerate adoption of proven practices.
  5. Build systems that can adapt to a rapidly changing world. AI evolution shows no sign of slowing down, and Salesforce’s program has had to adapt in kind. What began as “AI for Impact” quickly evolved into “Agents for Impact” as the technology landscape shifted towards prioritizing agentic AI, and continued evolution will almost surely be necessary down the line. The ability to remain nimble and continually adapt support structures is key to longevity and sustained impact, particularly when it comes to something as volatile and fast-evolving as AI.