This post may contain affiliate links. See our affiliate disclosure for more.
Do more with less.
It’s a mantra every marketer is well aware of. Less budget. Less time. Less buy-in. Fewer team members.
Equally cringe-worthy is the phrase founders and CEOs love, “We’d like to 10X this!”
As a marketer, this type of rhetoric is like attempting to squeeze blood from a stone. And I’m the stone!
But that each one modified on November, thirtieth 2022 when OpenAI introduced us all to ChatGPT. Suddenly, every marketer had an intern without charge. For those who could articulate your need rigorously enough, you may slash your production time to a fraction of what it was before.
Flash forward to today, marketers are well-versed in doing more, with less.
However the query is: can we actually get A.I. to put in writing exactly like us? Can it sound like a human and supply as much value (or more) than we are able to as real human beings?
I Did Do More With Less
I used to be in a do less with more situation about 60 days ago. We had just downsized our content team to the bone. I’m talking like 80% of the human resources—cut in a matter of weeks. How was I imagined to proceed publishing anywhere near the identical cadence we were before?
I used to be desperate.
Like lots of you, I had spent hours teasing out whether ChatGPT could write for me. It did just a few things well, for instance:
- Ideating titles
- Outlining articles
- Creating bulleted lists
- Populating article metadata
- Writing featured snippet optimizations
AI was writing filler stuff for me. I used to be saving some time, possibly getting 10% of my time back. But I didn’t want 10% more productivity, I needed to 10X my productivity.
So as to add to its lackluster performance, what AI was writing, wasn’t great. AI just wasn’t grasping the rhythm of my previous content team. I needed more examples, smoother transitions, and practical advice—just higher writing.
Then I discovered Claude, from Anthropic (not a sponsor). Immediately, I used to be floored by its writing ability. Clearly, Claude was a greater author out of the gate. Not only was it a greater author, but I discovered that its writing read more “human” than ChatGPT did after I would plug it into plagiarism or AI-written content checkers.
But Claude had one game-changing advantage over OpenAI’s Chat GPT model—I could upload training documents to Claude.
I had a theory. What if I uploaded a series of documents that trained an AI to my company’s mission, values, and editorial standards? Even higher, what if I uploaded my top-performing articles, with training on how to put in writing like us?
Do you see where I’m going with this?
Treat AI Like an Intern
LLMs (Language Learning Models) like Claude or ChatGPT are mainly interns who cost $20/mo (should you’re on the paid plans…free otherwise) and don’t need breaks. They’ll research, write, summarize, and even design stuff now.
While it could sound dehumanizing to think about an AI as an intern, it’s actually a game-changer. I’ve had only great experiences with interns I’ve mentored in my profession, and the important thing to a thriving intern is direction. Interns are hungry for work, but they know next to nothing. Sound familiar?
AI is like an intern. It so desperately desires to be busy and prove itself. Nevertheless it needs direction. Your direction.
Just like all recent hire, who HR showers with company documents and training during their first week, LLMs profit from the identical sort of “onboarding.” That is where the training documents are available in. Many organizations have names for these sorts of documents, standard operating procedures (SOP).
Treating an LLM AI, like Claude or ChatGPT, as in the event that they were a recent hire, by loading up some rigorously crafted trainings drastically improves their effectiveness.
Writing SOPs for LLMs (OMG, am I right?)
Whether you utilize Claude or GPTs (more on GPTs in the following section) for this purpose, each will use PDFs to learn the subject material you wish them to put in writing about. So what are the essential documents required to coach these LLMs?
You can go nuts with this, and I even have, but through trial and error I’ve found which you can reach some extent of diminishing return. Sometimes an excessive amount of training can overload the AI, and offer you something that reeks of “AI-generated content.” So start with the essentials:
- About: Let the AI understand your organization. Who’re your competitors? Who’s your goal avatar? What are the values that drive your organization? Do you might have a founding story value sharing? What’s your core mission? Aim for 1-3 pages.
- Editorial Standards: Teach the AI what makes your content sound such as you. What are your editorial values? What does every article need to perform? Are there specific dos and don’ts it should pay attention to? Do you utilize AP, or Chicago, or a midfield style guide? What’s the goal reading level of your audience? This training ought to be between 3-7 pages.
- Subject Matter Guidance: Give the AI ample resources to know deeply what makes your organization different out of your competitors. This may appear like a glossary of terms or sections that you just continuously cover. What are the industry pain points, and the way do you solve them? What are conventional narratives that you just stand in opposition to? These documents can vary in size, starting from 1-2 pages to 10-20. It relies on your industry, compliance, and the way much you might have to say.
- Article Anatomy and/or Examples: Instruct the AI on the flow and structure of your article, literally teach it what a successful introduction looks like, in addition to body sections and a conclusion. You should want to add just a few accomplished articles on this PDF for the AI to understand what the ultimate versions appear like. Aim for 1-2 pages of instruction, and 3-5 article examples.
Can’t You Do This With GPTs Now?
Yes!
Well, form of.
For individuals who don’t know, GPTs are a product feature inside OpenAI’s paid model where users can essentially train an AI, using documents and other directions, to play a particular part. There are a whole bunch of GPTs with a whole bunch more being published daily. They vary from skilled uses like code debuggers to innocent bedtime storytellers.
As mentioned before, Open AI is a special framework than Anthropic. That has benefits and downsides. I’ve spent just a few days twiddling with GPTs, and my consensus is that there’s a wealthy future here, but I haven’t abandoned Claude. I’ve spent more often than not attempting to construct a author for my very own site, Farmivore.
It’s been marginally successful.
The advantage of GPTs is that that is rather more streamlined after you get the product up and running (which may very well be done in a matter of 10 minutes). It’s very layman in that you just don’t need robust SOPs to start out, however the experience isn’t super dialed in yet. It seems to me that Claude is definitely reading my trainings, whereas GPTs skim what I’ve given it.
For complex and subjective difficult tasks, like writing, I’ve found that Claude does just a few things rather well (like 80% quality). GPTs appear to do every part but with less gusto (50-60% quality).
Nonetheless, I’m bullish on GPTs. But I’ll be thrilled when Claude arrives at the same offering.
Prompt Sequences
Contrary to popular belief, there’s no magical super prompt that can create an ideal article with one single prompt. The one exception I’ve found is for very short, objective content. For instance, aggregate news, covering stories that reputable sources have already broken. It doesn’t take a number of creativity to repeat the identical story using different words.
I find myself experimenting with recent prompts every week, but I’ve found probably the most success with the next formula:
- The primary prompt: Set the stage and ask for an outline.
- Next few prompts: Ask the AI to put in writing the document in sections, somewhat than the whole thing directly. That is tedious as a user, however the wait is value it.
- Final prompt: Once the article is accomplished ask the AI to now act like an editor, review your training documents, and see what the finished draft needs.
- Ad-hoc prompts: Then use the identical chat to expand on one-off topics as you edit the document in a separate Google Doc. For instance, if a piece is lacking detail, ask the AI to rewrite that section, giving it more direction.
Probably the most technical prompt is the primary one. Your first prompt should consist of three parts:
- First, set the stage by giving the AI a job, like “Imagine you might be an editor for Forbes with 20 years of experience…”
- Second, make a bulleted list of the training documents, instructing the AI what specifically you wish it to learn from each.
- Third, clearly prompt the AI to create an outline first.
All the time be testing! Prompting is equal parts art and science. Occasional deviations out of your standard procedure will allow you to get the very best prompt sequence possible.
Lessons Learned
For the reason that inception of this concept months ago, I’ve been prompting multiple articles daily for my site, Farmivore. Listed below are some learnings that can prevent hours.
- Super-prompts don’t exist. There’s no magical “one-prompt-fits-all” prompt that gets the job done. Slightly, develop a sequence of prompts that construct on one another.
- Be specific about what you wish AIs to do together with your uploaded information. Don’t just say, “read this PDF.” As an alternative, tell it exactly what you wish it to learn and do with the knowledge from the document.
- I find that some times of day, like afternoons, AIs are only dumber. They’re experiencing high user metrics, and I feel that affects their computing power.
- Get used to coming back to the identical prompt for specific things. For instance, if you might have a prompt that does rather well like answering FAQs, just keep using that very same prompt, even when its contents are for multiple articles.
- This won’t work perfectly out of the gate. I’ve made dozens of coaching docs at this point and I still have frustrating experiences.
- If any of this seems intimidating, enroll for a paid account with OpenAI to create a GPT without training documents. (Not a sponsor.)
Now get on the market and 10X the written word in a high-quality, human way that matters!
Keep the conversation going…
Over 10,000 of us are having every day conversations over in our free Facebook group and we would like to see you there. Join us!