What should small orgs and businesses be doing now to begin leveraging AI? Part 3

This article is part of a 3-part series being developed by Harrier. Learn more about Brian Birch here.

• Part 1: Audit your organization and customers, then create simple AI usage policies
• Part 2: Learn about generative search and start laying the foundations for digital authority
• Part 3: Learn and experiment with more than just writing prompts

Part 3: Learn and experiment with more than just writing prompts

It’s natural to begin by using ChatGPT, Gemini, and similar tools to write faster or act as a “Google-on-steroids,” but there is far more value to explore. Now is the time to challenge your small team to try new things and discover expanded use cases.

Give yourself and your team space to test and play with tools. Before you begin, set these rules, and follow through:

Rule #1:

Give your team explicit permission to spend 10–15% of their time each week exploring AI tools.

Rule #2:

As they explore, stay open to small investments in new tools and experiments. Manage this within your budget, and make trade-offs elsewhere if necessary.

Rule #3:

Bring the team together regularly to share what worked, what was easy, and what was hard. Establish a weekly or biweekly 20-minute cadence for quick updates and learnings.

Where to start with more advanced AI?

Vibe coding apps

Have a few of your more tech-savvy staff experiment with tools like Replit, Google AI Studio (free with Google Workspace), and Lovable. Encourage them to create simple apps and digital experiences, and offer these guidelines:

Tip #1:

Treat this phase as exploration. Use the tools to test ideas, visualize concepts differently, and learn how these platforms work.

Tip #2:

Use your LLM. Encourage the team to have ChatGPT, Gemini, Claude, etc., generate prompts or starter code for these tools based on the outcomes they want. This can save time, reduce errors, and conserve credits.

Tip #3:

Think of vibe-coded projects as pilots. It’s unlikely you’ll immediately build your next major product, but these tools are excellent for prototyping, visualizing ideas, gathering feedback, and shaping more robust project plans, content, or products.

Who thrives here? tinkerers, tech enthusiasts, and natural problem-solvers.

Here’s an example of a vibe-coded project created after significant experimentation:
Talk Nerdy 2 Me – Glossary Gamification Game for AI Terms

AI-enabled presentations

A substantial portion of small-organization work involves transforming information into easily consumable formats. AI tools excel here, and options continue to expand.

Apps like Gamma, Beautiful.ai, and Decktopus can rapidly generate on-brand presentations or microsites. Try converting something you’d normally build in PowerPoint or Google Slides into one of these tools.

• Begin with free or low-cost tiers while exploring.
• Create simple, brand-aligned templates first to ensure consistency.
• Be thoughtful about image generation. AI visuals can add value but may clash with your brand or alienate audiences less enthusiastic about AI-generated imagery.

Who thrives here? leaders, writers, editors, and salespeople—especially those who need to create polished content quickly without overloading design teams.

Deep dive research

There is a wealth of online information about your industry and audiences, and modern tools like Gemini, Perplexity, and ChatGPT’s deep research modes now allow you to:

• Pose complex research questions
• Generate structured research plans
• Edit those plans
• Execute them and receive synthesized findings

Additional tools like Elicit and Consensus offer more academic or evidence-driven research capabilities.

Deep research helps refine ideas, strengthen product proposals, uncover insights before committing resources, and spark new questions you may not have considered.

Best practices for AI-powered deep dive research

• Verify sources. Ensure citations are valid and representative. Ask the tool to revise its research if sources seem weak or overly repetitive. Challenge it to find primary or more authoritative sources, and provide ones you know are quality if they are missing.
• Avoid prompt bias. Don’t design prompts that steer the research toward a predetermined conclusion. Aim for objectivity, let the research tool do the work, than apply your analysis.
• Validate with humans. Use experts, colleagues, and your own review to confirm accuracy before relying on statistics or claims in the research.

Pairing deep research with AI-enabled presentations and vibe coding tools can significantly accelerate innovation and strategy development for small organizations. It's all about putting it all back together.

Harrier Takeaways on Experimenting With More AI Tools

There is no greater teacher than trying things—and getting feedback from real people.

Always keep humans in the loop. Never publish or launch anything without thorough review and vetting.

Using AI tools alone is valuable, using them together is extremely powerful