The session will explore the AI landscape from chat models to video & image generation, showcase practical examples of how nonprofits can leverage AI, and more
- Gain a comprehensive understanding of how AI can transform the grant-writing process and other nonprofit operations.
- Learn practical strategies and tools for incorporating AI into their organization's grant-writing efforts.
- Witness firsthand the power of AI through a live demonstration of its applications in identifying funding opportunities and generating proposals.
- Feel inspired to explore the broader potential of AI and generative AI to revolutionize the way their nonprofit works.
In a recent session, Mahesh showcased innovative approaches for companies looking to develop AI agents aimed at volunteer management. By leveraging existing resources such as company websites and internal documents, these agents can effectively address volunteer inquiries, integrate with internal systems, and streamline tasks through low-code or no-code platforms. He stressed the importance of utilizing high-context data at the outset to boost accuracy and the value of iteratively refining the agent over time.
The discussion also delved into best practices for crafting effective prompts,focusing on defining objectives, providing clear instructions, setting output formats, and establishing guardrails. Additionally, Mahesh highlighted how utilizing AI efficiently and responsibly can contribute to reducing environmental impact. He also covered in-house AI solutions like Copilot to ensure data privacy and explored the exciting possibilities of merging AI with blockchain technology.
Q: How can we build an AI agent for volunteer management at Goodera?
A (Mahesh):
You can create an AI agent for volunteer management with minimal technical knowledge. The process involves:
- Click the “Create” button to start building your agent.
- Define the agent’s name, tasks, and goals (e.g., “Answer volunteer questions on upcoming events”).
- Add knowledge sources such as company websites, uploaded documents, Salesforce, SharePoint, or Azure SQL.
- Configure triggers, responses, and actions, e.g., running Power Automate flows or updating Google Sheets.
- Deploy the agent across multiple channels: your website, Teams, Facebook, or Skype.
- Track analytics on agent usage, failures, and user interactions to refine its performance over time.
Q: What data is needed to start building a basic AI agent?
A:
Start with minimal but highly contextual data, such as your company website. A small dataset of 100–200 files is usually enough to handle roughly 1,000 queries per day. More data improves performance, but irrelevant or low-context data can reduce accuracy.
Key points:
- Focus on quality over quantity.
- Gradually expand data based on user queries and feedback.
- Drag-and-drop uploads make adding files easy.
Q: Can AI agents work without historical data?
A:
Yes. Agents can work with only the existing content on your website or documents. Historical data helps improve contextual understanding, but it is not mandatory. Start small and incrementally add more contextual information as users interact with the agent.
Q: How do we write effective prompts for AI agents?
A:
Effective prompts should include four essential sections:
- Goal/Title – Clearly define what the agent is trying to achieve.
- Instructions – Step-by-step guidance for performing the task.
- Output format – Specify the structure or style of the expected output.
- Guardrails – Rules to prevent the agent from performing undesired actions.
Additional tips:
- Include examples of good and bad outputs to refine results.
- Use AI tools like ChatGPT to generate optimized prompts if you’re unsure.
Q: How should companies address the environmental impact of AI?
A:
AI consumes significant computational resources, which can increase environmental impact. Companies can mitigate this by:
- Using AI efficiently, minimizing unnecessary computations.
- Supporting or volunteering for energy-efficient AI research, such as photonic chips.
- Offsetting environmental impact through sustainability initiatives.
- Encouraging remote work to reduce commuting-related emissions.
Q: How do large corporations ensure data privacy and compliance with AI?
A:
Large companies often have strict rules for internal data. In-house AI agents like Copilot address this by:
- Keeping data within the company, without sharing it externally.
- Respecting existing authentication and authorization processes.
- Allowing secure adoption of AI while maintaining compliance and privacy.
Q: What impact does combining AI with other technologies like blockchain have?
A:
Combining AI with blockchain enhances trust and discoverability:
- Allows secure cross-company data sharing.
- Uses protocols like Model Context Protocol and projects like Nanda for verified information transfer.
- Contextual and trusted data can then be used for more intelligent decision-making.
- The combination ensures that only authorized users access data and prevents impersonation.
Q: Which AI tool is best: Claude, ChatGPT, or Gemini?
A:
It depends on the task:
- Gemini: Best for multimodal tasks (e.g., video transcription and content generation).
- ChatGPT: General-purpose, good for reporting, text generation, and widely supported.
- Claude: Ideal for coding and technical tasks.
There’s no universal winner. The choice depends on the specific use case and available tools within your organization.
Q: Does Goodera plan to offer AI certifications?
A:
Yes, Goodera is considering certifications focused on:
- AI concepts like neural networks and transformers.
- Using AI agents effectively within CSR and corporate programs.
- Building AI agents for volunteer management and other use cases.
Feedback from participants will help finalize the content and structure of the certification.
Q: How many queries can a small dataset solve?
A:
A dataset of 100–200 files can address roughly 1,000 queries per day. The key is to start small with contextual data, monitor usage, and expand the dataset based on actual user questions to improve accuracy.
