Your AI outputs are only as good as your inputs. You've gotta write strong prompts to get solid results. But here's the catch: We're all busy, and we don't necessarily have time to become prompt engineering experts.

At Goldcast, we believe AI tools have the potential to help us optimize the way we use our time and let us focus on strategic, higher-level tasks. And we've figured out simple but effective ways to craft AI prompts that we're happy to share with you.

Today, we'll cover some insights from our Goldcast Insider event featuring two of our leaders: Co-founder Aashish Srinivas and Head of Marketing Kelly Cheng.

Read on to learn:

• What marketers should know about LLMs
• Goldcast's 5-step AI prompt framework
• 4 powerful AI prompt templates
• Use Goldcast Content Lab to discover your brand voice + more
• AI prompting frequently asked questions (FAQs)

We've handpicked the top moments for this blog, but there are lots of goodies we couldn't include. Watch the full session to see the full array of prompt templates we shared during the session, and let us know how they work for you!

WATCH NOW

This might be a refresher for you, but in case you're just getting acquainted with large language models (LLMs) like ChatGPT, let's talk about a few important points.

First, LLMs learn from the data they're trained on, and they make predictions based on probability. If you're paying for an LLM, you likely have the ability to change the "temperature" of the model, which gives it more creative license. However, the standard settings on a GPT typically don't allow it to think too far outside the box.

Because LLMs learn from the data they take in, it's critical to provide as much info as you can. The more context, the better. Studies have even shown that providing context, which is also called "chain-of-thought prompting," can improve accuracy for LLM models, even for simple math equations.

This means that instead of asking an LLM, "I have 5 apples, and I give 1 away. How many do I have left?", you would provide some example equations, as well as your thought process to figure them out, in your prompt. You might say, "I have 10 oranges, and I give 2 away. So the equation is 10-2=8. The answer is 8."

This prompt allows the LLM to understand exactly what it should do with the eventual questions you're going to ask it.

Finally, LLMs are limited in the amount of text they can consider at once. This will depend on the specific model you're using and whether you have a paid subscription, so be sure to find that info out ahead of time.

PS: If you need a rec for an LLM to start with, try ChatGPT or Claude. These are the most popular ones and we've had good results with them.

When we're writing AI prompts, here are the general rules we follow:

1. Be specific and concise. Use precise language and include any relevant details to reduce ambiguity. You don't want to send the LLM off on a wild goose chase.
2. Define clear objectives. What is the desired outcome? Also, give the prompt a persona (for example, including the language, "You are an event marketer" tells the LLM who it is supposed to be thinking as).
3. Provide step-by-step guidance for complex tasks. For blog posts, we've found it works well to tell the LLM to write the blog post first and then go back and edit the post to be a certain length. This results in better outcomes than starting with the limited word count upfront because LLMs can't pay attention to writing the blog post and getting all of the necessary information in a shortened word count at the same time.
4. Include context. Remember that chain-of-thought prompting? Methodically list out all the context that you would consider when performing the same task you're asking the LLM to do.
5. Test and iterate. You will very rarely get exactly what you're looking for on the first (or second, or third…) try. Keep going! Try not to get frustrated if the results aren't what you wanted. It takes time to get better at writing prompts.

Now that we know the components of a good prompt, let's take a look at some templates you can start your prompts with.

Many marketers are tasked with running event and webinar programs, and it only makes sense to put that content to work even after the webinar ends. Repurposing = working smarter, not harder.

Let's talk through some examples you can use to craft your own prompts to help you repurpose webinars and events.

Prompt: You are an expert content summarizer for B2B marketing teams. Please analyze the following webinar transcript and create a concise summary of the key takeaways.

Focus on actionable insights and recommendations that would be valuable for internal distribution to our marketing, sales, and product teams.

Write the summary. Then edit the summary to be around 300 words long and include 3-5 bullet points in the text that highlight the most important insights.

Let's be real. Not everyone who needs the info you cover in a webinar gets to attend — including the folks on your team. This prompt allows you to give everyone a quick summary of what happened at the event. Super helpful for your GTM folks who need to follow up with specific leads!

Here's how you would use Content Lab to generate the takeaways, using this prompt:

Super simple, and we got exactly what we asked for!

Prompt: As a B2B sales expert, create a series of 3 follow-up emails based on our recent webinar called 'WEBINAR TITLE'. Each email should be no longer than 150 words and incorporate key points from the event to engage and inform potential customers.

The emails should follow this structure:

Email 1: Highlight the main problem discussed and our unique solution

Email 2: Share a key insight or statistic from the webinar and its relevance to the prospect's business

Email 3: Offer additional resources and a clear CTA

Use a conversational tone that aligns with our brand voice*.

*Don't know your brand voice? Defining your brand tone is key so that all of your AI-produced content will be on brand!

Our recently launched Brand Voice feature within Content Lab allows you to input reference material like blog posts, LinkedIn posts, and other content assets and then rewards you with info on your unique brand voice! Learn more below, and get started with a free account.

TRY BRAND VOICE

The Goldcast marketing team is responsible for sales enablement. Creating a follow-up email series like the one above gives us a draft we then refine, resulting in engaging, on-brand emails that empower our sales team to nurture leads.

Prompt: You are a B2B marketing social media content creator. Based on this clip, create a LinkedIn post that includes:

A compelling hook (max 20 words)

A summary of the key point from the clip (max 50 words)

A call-to-action to watch the full webinar recording (max 20 words)

Use a professional yet conversational tone that aligns with our brand voice.

Many times, our webinars are jam packed with great bits of info and thought leadership. You can repurpose all of that into social media content to fuel your content strategy.

Again, Content Lab makes it a breeze to get those clips for social media platforms:

You can either have Content Lab create clips directly from the transcript, which works well if you know exactly what you want to highlight. Or you can have AI generate the clips for you. Content Lab will parse through the transcript, identify the key moments, and voila! You've got clips you can post.

Prompt: As a B2B tech content writer, create an outline for a keyword-driven blog post based on our recent webinar 'WEBINAR TITLE'. The blog post should be optimized for the keywords '[Keyword 1]' and '[Keyword 2], and encourage readers to watch the full webinar.

Please include:

A compelling title (including at least one keyword)

An introduction (100 words)

3-4 main sections (with subheadings)

A conclusion with a clear CTA

Suggested meta description (150 characters max)

Ensure the content naturally incorporates the keywords throughout.

Don't reinvent the wheel each time you write a blog. Use prompts like the one above to outline blogs or even write drafts you can use to start. Remember that we always need to assign the AI a role (here, it's a B2B tech content writer), and we want to be as specific as possible about the outputs we want.

We covered even more awesome examples of how you can use AI to repurpose content after your webinars, including email marketing sequences and longform e-book outlines. Don't forget to watch the full replay to see those in action!

WATCH ON DEMAND

And if you're intrigued about Content Lab and want to learn more about what it can do for you in terms of repurposing content and developing your brand voice, we invite you to join. It's always free!

TRY CONTENT LAB

Q: How do you train Goldcast Content Lab to understand your brand voice?

A: Content Lab's Brand Voice feature makes it easy to discover your unique tone! Simply take any content assets that embody the brand voice you want (for example, blog posts or brand guidelines) and feed that into the model. Then, Brand Voice will tell you the components of your own unique brand voice.

Check out this video to see how it works:

Q: How can marketers use AI for pre-event use cases?

A: Great question! We use AI to help plan our events. Before the event, we need an event title, an abstract, speaker bios, session descriptions, ideas for future topics, etc. AI helps us draft some of that up!

Obviously, we have to provide more context because we don't have an event transcript or much reference material to work with, but LLMs can still function as a handy assistant.

Q: I work in an agency environment. Can I train Brand Voice on multiple voices?

A: Another great q! Yes, you can absolutely have multiple brand profiles within Content Lab. You could have one for your own company brand, one for a thought leader within the company, one for an executive at a client's company, and so on.