Last month, we introduced the Lantern Leadership Studio, a governed AI environment designed to help school leaders establish the correct policy-based professional footing of a situation before deciding how to act. The post – Can AI Help School Leaders Think more Clearly has a leadership challenge and a downloadable response from the Lantern Leadership Studio.
This second part reflects on what happened when a real school used it.
The Belthorn pilot
Belthorn Academy, a primary school in East Lancashire, became Lantern’s first pilot institution. The brief was simply to use it in real situations, on real issues, and tell us honestly what happened.
Kelly Drugan, Deputy Headteacher, was among the first to use it regularly. Her account cuts to what matters most in a leadership support tool.
“I was amazed at how quickly it could generate clear, structured responses to a wide range of issues. It immediately felt like a powerful time-saving tool that uses all the school information, policies, and procedures. I’ve used it in multiple situations, most memorably to advise on what information could be shared with a parent regarding a SEND issue. The email it then helped me construct led to an agreeable resolution for all involved. Overall, I’d say it saves time while providing both assurance and confidence.”
But there’s a moment from the pilot that tells a quieter story about adoption. A WhatsApp group had been set up to log usage during the trial. After a few weeks, it went quiet. When Kelly was asked if everything was alright, she replied matter-of-factly, “Sorry, I use it every day now, so I keep forgetting.” This comment carried as much weight as her overall assessment of Lantern as a tool because it hinted at something else beyond in-the-moment support. Early-stage institutional adoption.
Melanie Furness, Headteacher at Belthorn, identified something the pilot had not been designed to test — the cumulative effect of a consistently structured output.
“I’ve used various AI tools but Lantern has a very different feel about it. Every response is structured in the same way, so you start to develop a real grounding in the situations you’re dealing with. But it’s also the detail it goes into. Everything from the issue classifications and policy hierarchies to the pressures you’ll likely feel and the defensible next steps feels controlled and trustworthy.”
An independent test
Before beginning a Phase 2 pilot, the updated Studio was put in front of someone with no prior involvement in Lantern’s development: Philip Mooney, a National Leader of Education with eighteen years of headship experience and a track record as a CSI Lead Inspector.
He applied it retrospectively to two real situations from his own career; cases he had lived through and whose outcomes he already knew.
“Having retrospectively applied Lantern to two real situations that reached a conclusion, it was impressive to see how it quickly summarised key points, provided timescale and guidance options for immediate application as well as defensible ways forward. In both cases Lantern would have helped me conclude things more quickly and, in all honesty, more effectively for all parties.”
His broader assessment framed what the pilot had been working towards.
“Lantern gives headteachers direction without instruction and options that are realistic. It provides a layer of protection which is based on policy and reason, with the ability to cut through issues very quickly, bring control to situations and remove potentially harmful influencers such as emotion and fatigue.”
The phase 2 pilot is now open
Lantern is now inviting schools to take part in the second pilot phase. Phase 2 pilot places are limited. Up to five schools can be accommodated, each receiving full access to the new Studio in exchange for documented experience that will inform Lantern’s ongoing development. If you’d like to be among the first schools to put governed AI to work in your leadership practice register your interest below and we’ll be in touch personally. Link: https://lanternaisystems.com/#contact
Technical Note – The move to Claude
The original Studio ran on a custom GPT environment within ChatGPT. It worked. But as the pilot progressed, it became clear that Lantern AI had more to give. If Lantern’s true potential was to be realised, it required a different foundation.
The migration to Anthropic’s Claude was a natural evolution rather than a correction. Claude’s capacity to handle high instruction density without degrading output quality, combined with the room for manoeuvre a Claude Project environment offers over a custom GPT, made it the right platform for what Lantern’s configuration layer demands. Anthropic’s published commitment to responsible AI development also aligns more closely with Lantern’s own governance philosophy.
The migration also accelerated Lantern’s instruction architecture’s development. Version 5.5 — the most refined iteration to date — was built for the Claude environment, not simply ported into it.
@LeadingLearner 
Discussion
No comments yet.