Writing compelling emails for progressive causes should be easier than writing persuasive corporate emails. This just makes sense, with some obvious exceptions.

Most progressive causes align with what the majority of people already want. Flight of the Conchords joked about taking a political stand against HIV/AIDS… strangely enough, most people are on board.

That said, what a non-profit is specifically asking for in the short term may not be. For instance, I worked on a campaign demanding more tax on beer. You can imagine how tricky it was to get a crowd chanting that passionately. Similarly if it involves a credible multi-lateral process, it will likely sound unbearably dull to anyone ‘normal’.

The point I’m making is progressive causes tend to have have less need for psychological tricks. They’re not manufacturing desires, using hidden persuaders so you buy another pair of trainers before yours are worn out.

They don’t need you to look the other way from the bonded labour conditions their workers are in. #SeafoodIndustry

They’re not depleting Earth’s resources to a point of no return, so want you to gloss over your kids’ future. #TechCompanies

They’re not using psychological techniques to link products to emotional fulfillment. Well, there is a little crossover with this last one. Progressive causes do want you to feel good about ‘doing the right thing’. They also have to counter the programming that ‘the right thing’ is branded with a swoosh or stripes.

How Can AI Supercharge My Non-profit’s Writing?

In a context where I see non-profits as wildly under-resourced to be competing in the same media spaces as corporations… AI might offer a bit of a helping hand. I’m sure there are multiple ways to achieve what I outline below. The purpose of this is to offer some food for thought.

RAG is the key to it all

To create a useful AI Agent, you want it to have a deep knowledge core specifically related to your objective. This is known as the RAG – Retrieval-Augmented Generation. A RAG consisting of relevant texts can create a specialist agent, rather than a generalist like most default LLMs. This knowledge base needs to be well structured and ‘chunked’ to make it easy to draw on by the agent. The creation of this ‘chunked knowledge’ has been easy for a while. However to date the hosting of it in an accessible way for an agent has been expensive.

It’s just become possible to create shareable GEMs in Google’s Gemini product. With a GEM you can create a form of RAG hosted by Google, that their Gemini agent can draw upon. So I wanted to test how useful an agent could be for writing. So far i’m very impressed, so here’s how you can make your own.

Step 1 – Codify the relevant knowledge

I created a knowledge base consisting of Neuro-Linguistic Programming and Cognitive Linguistics textbooks. I added in digital copies of movement and organising theory books. And finally added one or two about powerful narratives and storytelling.

Step 2 – Instruct your agent

For your agent to understand what you want from them, you have to explicitly state how they should act. I gave a lengthy description, that you could roughly summarise as:

Act as an expert in cognitive linguistics, psychology, and metaphor, providing direct, theory-backed feedback to improve writing. Offer culturally-sensitive solutions using clear, concise language, adapting your tone for the audience. Be collaborative, and state clearly if you have insufficient information to provide a strong answer. Be clear above all else, don't be kind.

Step 3 – Test and refine

Now you can test your agent. Check how well they’re able to apply the knowledge base you’ve given them. By this stage, you should be interacting with an exceptionally talented creative agency professional. One who is expert at employing psychological persuasion techniques. Test a few different subject matters and contexts.

Prompt it to write a letter that inspires activists so much they’ll send it to their MP. And their MP will also feel spurred into action.

Give it the blog / press release you’re drafting. Ask it to make suggestions for how to make it more emotionally engaging.

If it doesn’t deliver excellence, try different source materials for knowledge base. How you describe your expectations of the agent’s behaviour is also critical, so test variations of that.

If your organisation enables Gemini for your Google accounts, you can now share your Gem. I’m not advocating for any particular platform – this is just the context I’m working within. Your organisation should be clear what you’re allowing Google to read if you give it access to your organisational Drives. That said, if you’re using Gmail, you should also assume that is not private.

Do We All End Up With The Midas Touch?

My Greek friends would kill me if I used the Midas reference incorrectly, so here I’m ‘reclaiming’ it for them. If all progressive organisations start using supercharged AI agents educated in humanity’s psychological foibles, will we become too ‘powerful’? Effectively will it turn all our work to gold?

With the mythical Midas, the key message is lost in popular usage. Everything turning into gold included his daughter, and the food he wanted to eat. He begged Dionysus to reverse his wish. It would perhaps be useful in our consumerist society to remember what we wish for often won’t bring us happiness. Taking us neatly back to ‘products sold to meet emotional fulfillment’.

But to answer the question seriously, could this greatly improve all the comms from the non-profit sector? Yes.

Could that cause unforeseen problems. Possibly. Particularly in terms of emotional / empathetic exhaustion for audiences.

I referred to the Vance Packard book above, The Hidden Persuaders (1957). It lifted the lid on the extent psychology and psychiatry were weaponised to manipulate the masses into a consumerist society. That was in 1957. Edward Bernays was Freud’s nephew and brought him to New York to help him start the first PR company. So in my opinion, there is far too much manipulation already happening. You can only imagine how it has been honed and refined over subsequent decades. With digital devices recognising and exploiting desires we don’t even know ourselves.

The majority of people now know the scientific reality about climate change. But corporations were able to afford the suppression of that science showing they’re causing the end of the world. According to the New York Times, Reform UK has already taken more than £2.3m from oil and gas interests, polluting industries and climate deniers. That’s roughly 92% (figure from Novara) of its overall donations.
Can anyone guess whether their energy policies will be guided by science or short-term profit for millionaires?

So do I think a little power-up to the folks trying to do good is problematic? Less so.

Disclaimer

For best practice, I like people to be clear when something has been written by AI. So to model that behaviour – this was written imperfectly by me. I then asked Claude.ai to suggest some improvements. I acted on roughly half of them.