I use AI chatbots every day, and I teach my clients to use them too. Since ChatGPT launched in late 2022, people have been using chatbots to brainstorm, speed up research, draft content, summarize lengthy documents, analyze data, help with writing and debugging code, and translate text into other languages. Recently, the major chatbots have gained web search capabilities, so they can reach live information beyond their training data.
But getting good results from a chatbot takes a different way of thinking, especially if you are used to searching the web. I tell my clients to treat it like working with a human assistant: you get the best outcomes when you play to its strengths. These are the ten tricks I share most often.
Be specific and complete
Decades of search engines have trained us to type short, focused phrases built from keywords. Chatbots are the opposite. They thrive on specificity and detail. Prompting with “iCloud photos syncing” will not get you nearly as useful a response as “Tell me what might prevent iCloud from syncing photos between my Mac and iPhone.” Do not shy away from negative prompting either. Tell the chatbot what not to include. You can even specify the format you want, like a bullet list, a table, or a graph.
Treat every prompt as a conversation
We are used to standalone searches, where a failed search means starting over. Chatbots reward the opposite approach. Even a detailed prompt may not fully answer your question, or it may spark new ones. Ask follow-up questions, clarify what you want to find out or accomplish, give feedback, or redirect the conversation as needed. For the most conversational experience, try voice mode in the ChatGPT or Claude apps, where the chatbot talks back to you. I find it excellent for capturing ideas, refining my thinking, or just doing a brain dump.
Edit your last prompt
If a chatbot’s latest response is completely off the mark, try editing your prompt and resubmitting it rather than telling the chatbot it made a mistake. There is usually an edit link or pencil button that appears when you hover over your prompt.
Give it context
Most chatbots keep a library of your previous conversations, and you can search through them to find old ones easily. Because responses improve with more context, it helps to return to an earlier conversation when you want to explore that topic further. Similarly, if you are asking it to create something like work you have already done, share that previous work as an example.
Ask it to role-play
Another way to add context is to ask the chatbot to “act as” a particular kind of professional, such as an editor, coach, marketer, or software developer. You are asking it to respond from inside that role. It also helps to tell it who you are. Ask it to explain as if you were a high school student, someone with a basic understanding of the topic, or an expert in the field.
Know when to start over
Context is key, but chatbots have limited memory, and long conversations can overwhelm what is called the “context window.” If you notice the chatbot hallucinating, repeating itself, or wandering off topic, try saying, “Please summarize what we’ve discussed in a prompt I can use to continue working on this topic.” Then copy that prompt into a new chat and keep going.
Force web searches when it matters
Most chatbots make it clear when they are searching the web, which also tells you when they are not, and are therefore relying on potentially outdated training data. If you want the latest information, tweak your prompt to start with something like “Search for…”
Test its limits
A chatbot’s response is based only on what you put in the prompt, so it will not always go as deep as you would like on its own. Ask it to critique its own output, generate multiple options, or present the best argument for different perspectives. You can even ask it to be more cautious or more creative. It is perfectly fine to challenge a chatbot in ways that would be socially inappropriate with another person.
Save and reuse prompts that work
When you find prompts that work particularly well for recurring tasks, such as generating meeting summaries, analyzing data, or drafting specific types of content, save them so you do not have to start from scratch each time.
Do not believe everything it says
Chatbots are incredibly confident, and often truly astonishing in what they produce, but it is your job to verify important facts and details, just as you would with anything written by a human. The statistical models behind them can produce completely fabricated information. Web search makes this less likely, but even then, a chatbot can combine real facts in ways that simply are not accurate.
Bonus: try deep research
Many chatbots offer a deep research mode, in which the chatbot goes off for 5 or 10 minutes to gather information, work through it over multiple steps, and return with a much more comprehensive response. It is too slow for a back-and-forth conversation, but it is a great starting point when you are exploring a new topic that needs real depth.
If you would like to get comfortable with AI chatbots, or put AI to work in your business, I offer one-on-one AI training tailored to the way you actually work. Schedule a session with me and I will help you build real confidence with these tools.