Leadership is a lot about strategy. Which is basically a business-y word for planning. I'm sure I'm not alone in wishing I had a crystal ball and could predict a few things to help me out at Taught by Humans.
You've probably heard the term Horizon Scanning - basically looking into the future and the impact this will have on your industry or company. I've been made to do it on many a business programme and we have always found it silly. If I can predict things, I would bet on them (both with my money, and my business decisions).
One of our customers, Future Finance - an innovation and leadership accelerator - actually has a framework for this based on their research. Usually they deliver this in a session - which is very hands on and interactive. They wanted to find a way to make this more than a one off, and to allow more companies to make horizon scanning an easy and repeatable business practice.
Delivered By Dotly
When companies use our platform for their own training content (through our Partner Tier), they get some custom Dotlys (our AI learning companion). Basically, their own AI workflows based on a process or framework they already have. One thing I am surprised about is how well these actually work (Dotly for difficult conversation practice, Dotly for writing proposals). It's all to do with how we built this bit of the platform, which I'll admit to being a bit of a geek about. But that's worth its own blog post.
For Future Finance's Horizon Scanning Dotly, we took their existing framework, and gave it to our platform and it can now be delivered to their users via a friendly chat with Dotly. The flow talks the user through the framework - personalising it to them and their company. Probing them to input their expertise and experience. And outputs an easy to read and use report, which can be updated every week, month, quarter - whatever is needed.
Is This Better Than What Another AI Can Do?
We test a lot of tools and learning workflows for customers (it's made me know more than I care to about the probation system). Usually we test things with specific files - both me and Amy have used an Intro to Stats document 100 times to make Dots, or with quick irrelevant-to-us examples. But for this Horizon Scanning tool I noticed we were both using relevant things: workplace learning, AI-assisted coding. And we were chatting about the reports. Maybe Horizon Scanning was actually more useful than I thought?
It wasn't until someone mentioned using the tool and it being better than ChatGPT that I decided to investigate. I'm very much a Claude user - so decided to put Dotly and Claude head to head on horizon scanning.
Claude: Horizon Scanning
I use Claude a lot. So I was concerned going in was how much info Claude has on Taught by Humans, so it was going to be more personalised to our business and needs.
I want to do a Horizon Scan.
Claude came back with three questions - industry, timeframe and output.
Claude asked three questions then started building a document.
I clearly misunderstood the first - I thought it was about wider domain/sector to guide the other questions. Nope - after I answered the three it just started creating a doc. I stopped it to specify I had a topic: AI avatars in learning. And then it just jumped into action again. I didn't feel like I'd given enough info, but I guess I could have written longer prompts.
I stopped it to say I had a specific topic. It just carried on.
6 mins later I had an output.
Honest take - it was really long and wordy. Nothing relevant to the UK or my business. 1,600 words. I actually didn't understand some of the decisions it output. Its sources were random blogs.
To be fair, I tried to give it more context. I sent it my pitch deck and said it was for Taught by Humans (it literally has a very detailed Skill on our business so should change tone of voice, wording and have a decent bit of info on us).
4 mins later.
No back and forth again. I actually got distracted and did something else. If I really was doing a Horizon Scan it would be in the graveyard of my Claude chats, never to be returned to.
This was actually longer - 2,000 words but much less academic and a bit more readable. The decisions were the same - but weirdly the Taught by Humans versions of recommendations, for example protect the humans became taught by humans the name should be more than marketing and inform the product with exactly the same blurb after.
Verdict: I took nothing away, didn't actually read it. Back to thinking Horizon Scanning is a pointless exercise.
Dotly
Dotly knows a little bit about me from previous interactions. The relevant stuff - user is likely senior and interested in AI, ed tech and regulated industries. Well done Dotly. The rest is like user clearly likes learning and is highly engaged, not super relevant for Horizon Scanning, but we are a learning platform after all.
So Dotly's approach to Horizon Scanning is different - it's an ongoing engaged chat. I checked how long after and was genuinely surprised it was 4m 26s end to end. It felt really fast.
Dotly asked why AI avatars, what were my reasons for this horizon scan and what questions I wanted answered, and then went through the framework from Future Finance.
Dotly asked what I was actually trying to figure out.
I uploaded my pitch part way through - and it changed how Dotly thought about the signals (particularly around our customer being regulated industries and our need to be careful).
It already knew the context. The signals were relevant from the start.
The output was 845 words. Easy to read. Relevant, as it had researched the examples I gave. And the recommendations are really actionable - I've actually already done one of them (I wrote some blogs around our approach to adopting gamification, and used this as a selling point in an investor pitch), and assigned some to other team members (establishing research baselines, and setting up alerts to monitor our competitors).
Actionable recommendations. I'd already done one before I finished reading.
With Claude, I had to use some of my prompting expertise to get a mediocre, general output. With Dotly, I didn't have to prompt in a particular way to get the output, just chat.
So Horizon Scanning (Much Like Most Things) Is Actually Useful When Done Right
We have always claimed to be learning first, tech second. And in this case it was just someone else's excellently researched framework - to help people learn how to effectively horizon scan for their company. Our tech just helped turn it into an easy process - but none of this works without Future Finance's research being genuinely excellent in the first place.
We took their expertise. And made it extraordinary.