We spend a lot of time reading about AI, but most coverage focuses on new model launches, press releases from big companies, and industry drama, not the practical, day-to-day use cases and lessons we’re all searching for.
At NU, we’ve been running Practical AI training sessions to fill that gap. We’ve done over 30 sessions, for a variety of groups: PE and VC firms, COO and CHRO networks, boards, executive teams, talent teams, law firms, and entire companies. These interactive sessions focus on how you can use AI in your day-to-day work. We cover things like:
- Prompting best practices
- In-depth demos of ChatGPT’s capabilities with real-world examples
- How to build custom GPTs, agents, and tools
- Using connectors and assistants to add context and take action
- Demos of other AI-based tools
- Strategies for adopting AI organization-wide
We learn as much from our participants as they learn from us. We wanted to share some of our observations and lessons from being on the ground with so many participants:
1. Demand for AI education is surging.
Interest in these sessions has been overwhelming. Many clients request additional workshops and ask follow-up questions about implementing what they learned during the session.
2. Early winners and trends are emerging, but most organizations are still very early in the adoption cycle.
ChatGPT, Perplexity, and AI note takers are all gaining significant adoption.
Common uses for AI generally include search, research, idea generation, problem-solving, and content creation/editing/summarizing.
But most organizations are still experimenting to find deeper value, learning how to use the new technology, and slowly becoming comfortable from a compliance and security perspective.
3. Chat is the go-to interface.
We’re not seeing a lot of custom tools, connections, and agents. Mostly because of lack of education, lack of systematic, organizational adoption, and lack of approval from compliance teams. Everyone can jump into a chat, but the hurdles are higher for building even simple tools and connections. We expect adoption to grow a lot as the value of those products becomes clear.
4. Skill levels vary widely within every group.
We do sessions for all sizes of audiences. One relatively consistent aspect of the audiences is that they range dramatically in AI knowledge and use. Almost every audience has people who are barely using AI and others who use it every day, often in advanced ways.
5. Few organizations have a systematic AI adoption plan.
If you want your team to use AI more, then create a culture around it, implement triggers to use it, train and support your team, add the incentives, and assign an AI czar to ensure it all happens. A systematic, centralized approach can make it move much faster.
6. Balance security with flexibility for safe but necessary AI adoption.
Some organizations are testing everything because they have no rules. More are testing almost nothing because they’re scared of the risks. To move forward carefully, make sure you balance security with flexibility. Assign someone to own the legal/compliance aspects and someone to own the technical review and product settings.
7. There are so so so many new products out there, mostly from new companies.
New companies can launch products more easily than established companies because new companies have fewer customers and less to lose. In AI, that advantage is even more valuable because AI is a new technology, non-deterministic, and even scary or confusing, so established companies are adding simple, low value features more than transforming their products or businesses, while startups are trying everything.
AI development and adoption are still early. We all have a lot to learn, but the value is already there if you take the time to find it. And it will only increase from here.
Resources for Startup Operators 