Opportunity through technology
Make AI useful.
Make access
possible.
Stronger businesses. More independence. Greater opportunity.
BrdgFwd helps small and midsize businesses, nonprofits and underserved individuals turn AI into useful work. A small cohort of students and interns is already building an AI harness. We are developing hands-on training, business support and shared computing access—with success measured by time saved, work completed accurately and costs reduced.
An initiative of Watson Foundation Incorporated.
Learn by building. Turn AI into useful work.
Purpose-led. Greater independence. Stronger communities.
Practical by design. Measured by what people can do.
Why this work matters / National evidence
AI opportunity.
Unequal starting points.
Black-owned businesses are building, adopting and planning for AI. BrdgFwd aims to help small and midsize businesses turn that ambition into dependable, affordable work—with practical support, shared computing and measurable outcomes.
A business AI adoption gap.
Black-owned employer firms reported AI use at 35%, compared with 48% of white-owned firms—a 13-percentage-point gap.
Intent that deserves support.
31% of Black-owned employer firms were not using AI but planned to within 12 months, versus 13% of white-owned firms.
Unequal access to financing.
Among applicants for loans, credit lines or merchant cash advances, 32% of Black-owned firms received full approval, versus 57% of white-owned firms.
The risk we are working to prevent
Access to computing can determine what people get to build.
Compute means the processing power available through a computer, a shared GPU workstation or cloud services. Running local models and building repeatable AI workflows can require more resources than occasional chatbot use.
Our concern is that unequal access to equipment, paid services and technical support could compound existing economic barriers as AI becomes part of everyday work. That is a risk we aim to reduce, not a measured forecast.
What funding makes possible
Shared resources. Practical results.
We seek support for suitable computers, shared GPU capacity, cloud credits and guided project time. Participants will use resources matched to a real task: prepare job materials, analyze business information or improve a community service.
We will record access barriers at intake and follow up at 30 and 90 days: suitable device and connectivity, available cloud/GPU access, out-of-pocket cost, tasks blocked by cost or capacity, and tasks completed independently. We will report results alongside time saved and cost per successful participant.
See our proposed outcome measures →What these numbers do—and do not—show
Source: Federal Reserve 2026 Firms in Focus race and ethnicity chartbook, based on the 2025 Small Business Credit Survey. AI questions were optional. AI-use samples: 508 Black-owned and 3,925 white-owned employer firms; financing samples: 282 and 1,550. These national comparisons are not BrdgFwd results or causal estimates.
AI adoption does not measure GPU ownership or cloud capacity. Among AI users, reported full integration was 8% for Black-owned firms and 6% for white-owned firms; lower overall adoption does not mean every adopter is behind. See page 43.
The survey uses a weighted nonprobability sample. Read the methodology. Financing constraints may make technology investment harder; that is our rationale to investigate, not proof that financing caused the AI-use gap. We will measure actual device, compute, cost and support barriers with participating businesses. Demographic reporting will be voluntary and protect identities. Sources reviewed September 7, 2026.
One connected approach
Access takes more
than a connection.
Technology is useful when it helps someone finish real work. Our early build cohort is developing an AI harness. We are preparing practical learning and business support that combine capable tools, guided implementation and accountable review.
AI skills for everyday opportunity.
Guided workshops on understanding information, preparing job materials, organizing work and using AI responsibly. Practice checking outputs, protecting personal information and knowing when a human expert is needed.
AI support for small and midsize businesses.
Help businesses and nonprofits with limited technical capacity tackle routine work, such as organizing documents, drafting customer communications and preparing operational reports. Start with one workflow that its users can review, maintain and measure.
Tools people can actually use.
Suitable computers, shared AI workstations and sponsored cloud resources to support the learning and clinic programs. Evaluate equipment for compatibility, usefulness and ongoing cost before putting it into service.
Different communities. Concrete barriers.
Opportunity should
be within reach.
We aim to reach adults and community organizations underserved by existing technology support, including communities facing economic or structural barriers. We will shape a manageable first cohort with community partners and the people the program is intended to serve.
Initial service area, participant selection and delivery partners will be confirmed before enrollment. We welcome prospective partners who understand the needs of their communities.
Support the next phase
Your support has
a practical purpose.
Build on the work students and interns have started. Support the people, computing resources and business assistance needed to expand responsibly, with a defined scope, budget and outcome measures.
- +Sponsor learning
Support instruction, participant access, learning materials, coordination and evaluation.
- +Offer technology
Contribute suitable computers, complete GPU workstations, cloud resources or software access. Advanced computing requests will be tied to defined program needs.
- +Keep the program working
Help cover connectivity, hosting, maintenance, shipping and responsible administration.
- +Become a delivery partner
Bring participant referrals, space, instructors, technical expertise or community insight.
How we measure success
Useful work.
Measurable business value.
We aim to help small and midsize businesses and nonprofits get reliable work done with less time and expense. We will measure results on real workflows, including the effort people spend reviewing and correcting AI output.
Proposed targets—not achieved results. Our student-and-intern build cohort is underway. Business outcome measures apply to the next service phase. We will agree the task, baseline, quality standard and reporting period with each participating organization before evaluation.
20% less hands-on time.
Our proposed target is a median reduction of at least 20% in staff time per selected workflow, while maintaining quality. Examples include preparing an operational report, organizing business documents or drafting customer communications.
Lower cost per completed task.
Our goal is to make dependable AI-assisted work affordable. We will compare cost per correctly completed task with the starting workflow and report whether the savings justify the ongoing expense.
Quality that holds up in review.
A faster workflow must still meet the business’s agreed accuracy, completeness and privacy requirements. Our goal is to maintain or improve its acceptance rate as users take on more work.
Do businesses keep using it?
Check continued use at 30 and 90 days, tasks completed, time recovered and support needed. Report adoption out of all enrolled businesses, including dropouts and nonresponses. Track revenue or customer-service changes only when data exists; do not assume AI caused them.
Can participants do more independently?
For students and interns, track working contributions, tests passed and the ability to explain and maintain their work. For the expanded learning track, our proposed target remains 80% independently completing three useful tasks, checking an AI error and protecting private information. Establish a baseline and report actual results.
Show the return on support.
Publish an aggregate scorecard within 30 days of the 90-day follow-up: businesses and learners served, target versus actual results, quality, cost per completed task, continued use and barriers still unresolved. Explain missing data and protect business and participant information.
These are evaluation plans, not promises of profit, employment or income. Small cohorts and before-and-after comparisons have limits. We will report those limits and what we change as we learn.
Our leadership
Enterprise leadership.
Community purpose.
Mario Watson
Founder & President · Watson Foundation Incorporated
Mario Watson brings enterprise product leadership, commercial growth experience and nonprofit governance to BrdgFwd. He is Vice President of Product Management, Retail Media at Albertsons Companies, with a career built around turning technology strategy into platforms, teams and businesses that scale.
During 18 years at Target, Mario advanced through analytics, digital commerce, software and product leadership. He helped build and scale Target’s billion-dollar retail media business, leading global product teams across advertising platforms, self-service tools, measurement and monetization. His work connects product vision with the operating discipline needed to deliver it: clear priorities, accountable teams and measurable results.
As a director on the Normandale Community College Foundation board, he brings that same perspective to educational access and opportunity. Through Watson Foundation, which he founded in 2017, Mario is now developing BrdgFwd Community AI Lab—bringing practical AI learning, organizational support and computing access to communities with fewer resources.
Why this work matters to Mario. After helping large organizations turn technology into measurable growth, Mario wants smaller businesses and communities with fewer resources to have practical ways to benefit from AI. BrdgFwd brings that purpose into hands-on building, learning and business support.
Success means showing what participants can do, how much time businesses and nonprofits recover, what each completed task costs and whether those gains last. Early activity and proposed targets will be reported separately from measured results.
Explore Mario’s professional background ↗
Career and board affiliations describe Mario’s experience; they do not imply organizational sponsorship.
Our nonprofit foundation
A registered charity.
A clear public purpose.
BrdgFwd Community AI Lab is an initiative of Watson Foundation Incorporated, a Minnesota nonprofit corporation recognized by the IRS as a 501(c)(3) public charity.
Incorporated January 13, 2017 · IRS exemption recognized March 2017 · Public records checked September 7, 2026
Questions from supporters
A thoughtful start.
Is the pilot already running?
Yes. A small cohort of students and interns is building an AI harness—a system for organizing, testing and reviewing AI-assisted work. We are developing the next phase of practical support for small and midsize businesses, nonprofits and underserved individuals. Expanded enrollment, delivery partnerships and schedule will be announced when confirmed.
Can we offer equipment instead of funding?
Yes—equipment offers are part of the proposed sponsorship process. We will assess whether the system is complete, supported and useful, and whether shipping, power and maintenance are affordable. Please do not send equipment before acceptance is confirmed.
How will AI be used?
Our current cohort is building and testing AI-assisted workflows. Business support will focus on useful, routine work with human review. Learning covers errors, privacy and appropriate limits; sensitive or consequential decisions remain with qualified people.
How will sponsors know what their support achieved?
Sponsors will receive the agreed scope, budget and measures before committing funds. Our scorecard will report time saved, cost per correctly completed task, output quality, continued use and learner progress—with targets and actual results clearly separated. See how we measure success.
BrdgFwd Community AI Lab
Test the technology.
Make the evidence useful.
BrdgFwd Community AI Lab is the public identity for our AI evaluation work. We plan to connect transparent testing with practical decisions about which models, tools and computing resources can serve communities well.
Clear attribution
Our benchmark entries will identify the model, the AI harness and the BrdgFwd Community AI Lab team. A model's abilities and the surrounding software must remain distinguishable.
Evidence before claims
Published results should link to the benchmark version, tested configuration, run evidence, costs and limitations. Self-run results will be labeled separately from results accepted by a leaderboard.
Community value
Technical scores help guide evaluation. They do not establish better learning, employment or community outcomes; those require separate program measurement.
Benchmark results and leaderboard links will appear here after verification and publication. No ranking or accepted submission is claimed on this page.
Build the opportunity with us
Bring your resources.
Your expertise. Your perspective.
We welcome conversations with funders, equipment donors and community partners who want to make technology more useful and accessible.
Contact [email protected] to discuss program sponsorship, equipment or collaboration. Please do not send confidential records or account credentials.