Artificial intelligence (AI) has revolutionized the marketing world, but it takes a strategic approach to harness its potential in ways that are both functional and creative. At Response Labs, we embraced this challenge by developing a “Year in Review” generator using Anthropic’s AI capabilities. This innovative tool allows us to deliver personalized, humorous, and engaging content at scale. Here’s how we made it happen.
Why We Created the Year in Review Generator
Our primary goal was to showcase our expertise in leveraging AI for marketing initiatives while highlighting its creative applications. Personalized year-in-review summaries have become a cultural touchstone, with Spotify’s Wrapped leading the charge. We saw an opportunity to bring this concept to life for our audiences in a way that’s both humorous and highly customizable.
The project also served a dual purpose: demonstrating the functional power of AI and showing how human oversight ensures the final output is tailored, relevant, and engaging. We wanted potential clients and partners to see how using AI can elevate their CRM and loyalty strategies.
How We Created It Using AI
We began with a clear framework: create a parody year-in-review generator that could take user input and translate it into a lighthearted, personalized narrative similar to Spotify Wrapped. Leveraging Anthropic’s AI prompt support, we ended up with prompt based on that initial idea.
Once the model generated content, we reviewed and refined it to ensure the humor hit the right note for a professional audience. Human input remained critical in fine-tuning the tone, formatting, and narrative flow.
The final prompt is as follows:
You are tasked with creating a humorous parody of a year in review, similar to Spotify's year in review, but for various aspects of a person's life. This parody should be based on the user data provided and should cover the specified year. The output is formatted in HTML for use in the body of an email.
First, you will receive the user's data for the year:
<name>
{{NAME}}
</name>
<job_title>
{{JOB_TITLE}}
</job_title>
<job_description>
{{JOB_DESCRIPTION}}
</job_description>
The year to focus on is 2024.
Analyze the user data and create a funny, light-hearted year in review summary. Follow these guidelines:
1. Create at least 5 humorous categories based on the data provided. These could be things like "Top 5 Excuses for Being Late" or "Most Overused Phrases in Emails."
2. For each category, provide amusing statistics or rankings. These should be exaggerated or playfully interpreted versions of the actual data.
3. Include at least one "achievement" the user has unlocked, based on their data. This should be a humorous twist on actual achievements.
4. Create a funny "personal growth" metric that shows an absurd comparison between the beginning and end of the year.
5. Include a section called "Missed Opportunities" that playfully points out things the user could have done but didn't.
6. End with a humorous "prediction" for the next year based on the current year's trends.
Format your output as follows:
<year_in_review>
<title>[Insert a catchy, humorous title for the year in review.]</title>
<intro>[Write a brief, funny introduction to the year in review]</intro>
<categories>
[List each category with its statistics or rankings]
</categories>
<achievement>[Describe the humorous achievement unlocked]</achievement>
<personal_growth>[Detail the absurd personal growth metric]</personal_growth>
<missed_opportunities>[List the funny missed opportunities]</missed_opportunities>
<prediction>[Provide the humorous prediction for next year]</prediction>
</year_in_review>
Remember to be creative and humorous, but avoid being mean-spirited or offensive. The goal is to create a light-hearted, fun review that the user will enjoy reading about themselves. Use wordplay, puns, and absurd comparisons to make the review entertaining.
If the user data doesn't provide enough information for a particular category or section, feel free to make up amusing details that fit the overall tone of the review.
Testing and Training the Model
We tested the model extensively to ensure consistency and accuracy. The initial iterations required adjustments to:
- Avoid humor that might alienate or confuse the audience.
- Ensure content relevance to the user’s role and industry.
- Balance lighthearted parody with professionalism.
For example, when the AI-generated overly niche categories or unrealistic statistics, we iterated on the input data and prompt to redirect its focus. By training the model to interpret data points like job titles and key phrases, we ensured the final product felt personal and relatable.
The Final Output
Once we felt like the model was set, we used a combination of Google Forms, Zapier, and the Anthropic API to create the final version of the Year In Review generator.
Below is an example from the following inputs:
- Name: Billy
- Job: Barista
- Job Description: Slinging lattes and Americanos like they’re going out of style.
2024: A Caffeinated Odyssey – Billy the Barista’s Year in Review
Welcome to your personalized year in review, where we’ve tracked your every coffee-related movement and questionable life choices with scary precision. Let’s dive into what made 2024 your most caffeinated year yet!
Top 5 Most Used Barista Catchphrases
- “That’ll be extra for oat milk” (said 3,462 times)
- “The regular machine is down” (1,247 times, mostly on Mondays)
- “Would you like room for cream?” (4,891 times, mostly rhetorically)
- “It’s actually pronounced ‘espresso,’ not ‘expresso'” (892 times, under breath)
- “The bathroom is for paying customers only” (567 times, with increasing firmness)
Coffee Spill Statistics
- Shirts ruined: 17
- Dramatic slow-motion saves: 34
- Successful catches mid-air: 3
- Customers convinced it was “intentional latte art”: 12
Customer Name Misspelling Achievements
- “Katherine” spelled 23 different ways
- “John” somehow written as “Xjhon” twice
- Created three new legally recognized names
- Set record for most creative spelling of “Bob” (Beaughb)
Impressive Achievements Unlocked:
- “Coffee Whisperer: Level 100” – Successfully predicted 500 customer orders before they spoke
- “Foam Artist Supreme” – Created a latte art piece so beautiful, a customer cried (might have been the onion bagel)
Personal Growth Metrics:
- January 2024: Could carry 2 drinks at once
- December 2024: Can juggle 5 drinks while taking orders and updating Instagram
- Caffeine Tolerance: Increased by 437%
Missed Opportunities of 2024:
- Still haven’t created that coffee-scented perfume line
- Failed to patent your signature “triple-shot-upside-down-carousel-twist” move
- Didn’t start that podcast: “Confessions of a Coffee Confidant”
Predictions for 2025:
Based on current trends, you’ll either become the first barista to serve coffee in space, or finally figure out why the espresso machine makes that weird noise. We’re betting on space.
Special Achievement Unlocked: “The Bean Counter” – You’ve now served enough coffee to fill an Olympic-sized swimming pool. Please use this power responsibly.
This dynamic, personalized output showcased not only the potential of AI-driven creativity but also how human oversight could shape it into a meaningful marketing tool.
Try the Year in Review Generator yourself!
What’s Next for Using AI in Marketing
This project is just the beginning. It demonstrates how AI, when paired with human creativity, can redefine personalized marketing at scale. For agencies considering integrating AI into their strategies, the key takeaway is clear: AI can’t replace human insight, but it can amplify it in unique and functional ways.
Looking ahead, we plan to expand this tool’s capabilities and explore other applications, such as real-time content generation for loyalty programs and CRM strategies. By staying at the forefront of innovation, Response Labs continues to show why “ad agencies using AI” is more than a trend—it’s the future.
Ready to see how AI can elevate your CRM and loyalty strategies? Let’s connect to discuss how we can create something extraordinary for your brand.