BEACON

BEACON AI complements the BFD portal by simplifying resource matching and inquiry responses. It enables navigators to update the business’s portal dashboard with actionable next steps (e.g., task assignments like applying for grants) using information from the AI tool.

1. Beacon AI- Dual Tracks

Through user research conducted across various EOED agencies, we identified two critical pain points faced by Navigators.

The first challenge is the lack of comprehensive knowledge about the eligibility requirements for EOED’s wide range of resources and grants. While Navigators are often familiar with the offerings of their home agency, they may not have the same level of expertise for resources across other agencies. As a result, accessing this information becomes a manual and time-consuming process.

The second pain point arises when businesses inquire about EOED resources. In these situations, Navigators must either possess an extensive understanding of the available programs or spend significant time searching for the necessary information across disparate systems and documents.

2. Filtering based on Business Resource Intake Form

When businesses access the Business Front Door (BFD) portal, they complete a Resource Intake Form that captures key details about their operations and needs. BEACON AI integrates the responses from these forms directly into its system. Navigators can input this information into the tool using intuitive drop-down menus that cover criteria such as the business’s life cycle stage, size, category, and specific needs. By selecting the relevant filters and clicking “Filter Results,” Navigators can instantly generate a tailored list of grants and programs that align with the business’s unique requirements. This streamlined process saves time and ensures more precise recommendations.

3. Removing irrelevant grants/programs based on Navigator Knowledge

Once the list of relevant grants and programs is generated, it is displayed in a clear and organized tabular format. The table provides key details, including the agency, grant name, and the type of grant (e.g., informational or program). Navigators can review this list to ensure it aligns with the business’s specific needs. If a Navigator determines that a particular grant or program is not relevant, they can easily remove it by clicking the “Remove” option in the Actions column. This feature allows Navigators to refine the recommendations further, ensuring the final list is tailored and accurate to the business.

4. Pre-populated Prompt

After finalizing the list of grants and programs, Navigators can click “Summarize Resources,” which transitions them to the chatbot. Here, they are presented with a pre-populated prompt that reads:

“Based on the filters selected, I found these resources: [XYZ]. Could you please summarize these resources and their key benefits, and highlight all eligibility requirements or deadlines (if any) for each resource? Maintain all of the formatting requirements of the original system prompt.”

This feature reduces repetitive typing, minimizes human error, and ensures consistency in how requests are made to BEACON AI.

5. Generate relevant information regarding grants and programs

The tool provides detailed information about grants and programs by generating a summary, eligibility criteria, key benefits, and deadlines for each resource based on the input prompt. It also offers a conversational interface, allowing Navigators to request additional details or clarification about any resource by simply interacting with BEACON AI. This dynamic capability ensures that Navigators can quickly access comprehensive information. BEACON AI is trained on the Request for Proposals (RFPs) and one-pagers for each grant and program, which are submitted by program managers across EOED agencies, ensuring the information it provides is accurate and reliable.

6. Email Generation and Sources

After a response is generated, Navigators can view the sources of the information. Additionally, the tool offers an email generation feature, enabling Navigators to easily share the information with businesses, if needed. This functionality not only enhances the tool’s usability but also aligns with insights from our user research, which revealed that email is the primary mode of communication between Navigators and businesses. By streamlining email creation, this feature saves time and further supports Navigators in providing timely and professional responses.

BEACON AI Tool Architecture

Metrics for Success

The following metrics will evaluate BEACON AI’s success:

• Reduction in Manual Effort: Track the time saved per inquiry compared to the current manual process.
• Improved Accuracy: Measure the percentage of accurate referrals and responses.
• Faster Turnaround: Record the average response time for business inquiries.
• Navigator Satisfaction: Collect qualitative feedback from Navigators to assess usability and effectiveness.

Future Work

Going forward, AI for Impact would like to continue work with EOED to test the current version with navigators and businesses. User testing sessions would help refine BEACON AI and assess its impact as we move into the next phase of development.

As Navigators become accustomed to using the Business Front Door (BFD) portal, further testing of BEACON AI will be conducted to ensure seamless integration between the two platforms. This process will involve gathering feedback from Navigators on their experiences using both systems together, identifying pain points, and areas for improvement. This iterative feedback loop will guide the refinement of BEACON AI, ensuring it continues to meet the practical needs of its users. Based on the feedback collected, enhancements to BEACON AI will be prioritized to better align its functionality with the workflows and challenges identified during real-world use. These changes will ensure that the tool evolves to support Navigators more effectively, increasing both efficiency and user satisfaction.

BEACON AI has the potential to evolve into a public-facing resource for businesses. Transforming it from an internal tool to a platform accessible to business owners would require significant enhancements. These include achieving high accuracy rates to minimize reliance on human oversight, implementing self-service dashboards with real-time eligibility checks, and incorporating multi-language support to cater to the diverse business community. Such advancements would greatly expand BEACON AI’s functionality and its impact on the business ecosystem.

The tool’s future development will focus on adding valuable features such as automated tracking of grant deadlines with reminders and monitoring task completion rates within the BFD portal. These capabilities would provide additional utility to users and streamline grant and program management further.

One current limitation of BEACON AI is its reliance on data stored in an Excel sheet in an S3 bucket for filtering grants and programs. While this approach is effective, it is not without drawbacks. Changes to the Excel file—such as adding or removing columns—can disrupt the Resource Finder, as the data logic is hard-coded into the backend. To address this, future updates may incorporate AWS Agents to dynamically update the information, making the system more robust and adaptable to changes.

Additionally, we aim to enhance the chatbot functionality of BEACON AI by enabling seamless crawling of Mass.gov websites. At present, restrictions on web-crawling Mass.gov content limit the tool’s ability to access and utilize this valuable information. In the future, finding a viable workaround to these restrictions would significantly improve the chatbot’s ability to generate comprehensive and accurate responses.

The user interface of BEACON AI is another area identified for improvement. Plans include updating the checkbox and dropdown menus so options filter in real-time as selections are made, eliminating the need for a dedicated filter button. Additionally, adjustments to the font and overall design will ensure the interface is more accessible and user-friendly for all Navigators.

Currently, BEACON AI operates under the assumption of Navigator oversight. Navigators are expected to review and verify all information generated by the tool before sharing it with businesses. This human oversight acts as a critical safeguard, ensuring the accuracy and reliability of the tool’s outputs. While future developments aim to enhance BEACON AI’s automation and precision, the Navigator’s role in maintaining quality control will remain essential

PROJECT TEAM

• Suvir Ghai
• Khushi Patel