What running publishing businesses taught us about using AI
admin
27 view(s)
Stella Novus runs publishing and education businesses: the Ancient Origins website, Ancient Origins Magazine, Ancient Origins UNLEASHED on Substack and, since 2026, Ancient Wisdom Academy. Like many small companies, we have spent the last few years working out where AI helps and where it does not.
We do not have every answer. But we have learned enough to share a few practical lessons. They are also part of why we set up Nexus AI Advisory.
1. The real value is in mechanical processes
Publishing involves a great deal of mechanical work: moving content from one format to another, tagging and sorting it, checking lists against each other, preparing the same material for different channels, keeping records up to date. None of this needs much judgement, but all of it takes time.
This is where AI and automation have been most useful to us. Not in replacing the thinking and writing, but in taking over the repetitive steps around it, so people can spend their time on the parts that need them.
2. Look for roadblocks and hand-offs
When we looked closely at how work moved through our businesses, the delays rarely came from the work itself. They came from the gaps: waiting for someone to pass something on, re-entering the same information in a second system, a task sitting in a queue because one step needed a person to press a button.
Every hand-off is a chance for delay or error. Automating those mechanical steps means fewer roadblocks and fewer hand-offs, and work flows more smoothly from start to finish. In our experience, that is a more useful question to ask than "where can we use AI?": ask instead "where does work get stuck?"
3. Keep judgement with people
In publishing, accuracy matters. AI tools can produce text that sounds right and is not. We think anything that reaches readers or customers should have a person responsible for it. Automation can handle the mechanics; people should keep the judgement.
4. Your data is usually part of the problem
Many problems we expected to solve with AI turned out to be problems with how information was organised: content spread across different systems, inconsistent naming, old material nobody had catalogued. Tidying that up often makes automation possible in the first place.
5. Think about privacy and rules early
Any business handling customer data has to think carefully about what goes into which tool. In Europe that means GDPR, and increasingly the EU AI Act. It is much easier to make good choices at the start than to unpick bad ones later.
6. Try small, and be willing to stop
Not every experiment works. The mistake is not trying something that fails; it is continuing to pay for it, in money or attention, after it is clear it is not helping. A small, contained trial with a clear question, such as "does this remove a step?", is usually the best way to find out.
7. People matter more than tools
The hardest part is rarely technical. It is helping people understand what a change is for and how their own work will be affected. People who are involved in the decision are much more likely to make it work.
Why this led to Nexus
As we talked with other business owners, we found many asking the same questions we had asked. Nexus AI Advisory brings together experience in strategy, AI research, business change, commercial operations and data governance to help businesses find their mechanical processes, remove unnecessary hand-offs and adopt AI practically.
Stella Novus runs publishing and education businesses: the Ancient Origins website, Ancient Origins Magazine, Ancient Origins UNLEASHED on Substack and, since 2026, Ancient Wisdom Academy. Like many small companies, we have spent the last few years working out where AI helps and where it does not.
We do not have every answer. But we have learned enough to share a few practical lessons. They are also part of why we set up Nexus AI Advisory.
1. The real value is in mechanical processes
Publishing involves a great deal of mechanical work: moving content from one format to another, tagging and sorting it, checking lists against each other, preparing the same material for different channels, keeping records up to date. None of this needs much judgement, but all of it takes time.
This is where AI and automation have been most useful to us. Not in replacing the thinking and writing, but in taking over the repetitive steps around it, so people can spend their time on the parts that need them.
2. Look for roadblocks and hand-offs
When we looked closely at how work moved through our businesses, the delays rarely came from the work itself. They came from the gaps: waiting for someone to pass something on, re-entering the same information in a second system, a task sitting in a queue because one step needed a person to press a button.
Every hand-off is a chance for delay or error. Automating those mechanical steps means fewer roadblocks and fewer hand-offs, and work flows more smoothly from start to finish. In our experience, that is a more useful question to ask than "where can we use AI?": ask instead "where does work get stuck?"
3. Keep judgement with people
In publishing, accuracy matters. AI tools can produce text that sounds right and is not. We think anything that reaches readers or customers should have a person responsible for it. Automation can handle the mechanics; people should keep the judgement.
4. Your data is usually part of the problem
Many problems we expected to solve with AI turned out to be problems with how information was organised: content spread across different systems, inconsistent naming, old material nobody had catalogued. Tidying that up often makes automation possible in the first place.
5. Think about privacy and rules early
Any business handling customer data has to think carefully about what goes into which tool. In Europe that means GDPR, and increasingly the EU AI Act. It is much easier to make good choices at the start than to unpick bad ones later.
6. Try small, and be willing to stop
Not every experiment works. The mistake is not trying something that fails; it is continuing to pay for it, in money or attention, after it is clear it is not helping. A small, contained trial with a clear question, such as "does this remove a step?", is usually the best way to find out.
7. People matter more than tools
The hardest part is rarely technical. It is helping people understand what a change is for and how their own work will be affected. People who are involved in the decision are much more likely to make it work.
Why this led to Nexus
As we talked with other business owners, we found many asking the same questions we had asked. Nexus AI Advisory brings together experience in strategy, AI research, business change, commercial operations and data governance to help businesses find their mechanical processes, remove unnecessary hand-offs and adopt AI practically.
If this sounds familiar, Nexus offers a free 30-minute consultation. You can find out more at www.nexusaiadvisory.com or email [email protected].