Anapoly AI Labs

Category: diary

A public record of the development of our ideas

  • A pivot

    Our initial idea, prompted by Kamil Banc’s writing on practical AI use, was to run a small, local club. Somewhere people like us could meet in person, experiment with ChatGPT, and see what we could actually do with it. A “non-threatening, friendly environment,” we called it at the time.

    But the concept developed, and the name seemed too cosy. A reference to Google Labs brought up the idea of a lab as a place to experiment with tools and ideas. This resonated, so we pivoted to thinking of ourselves not as conveners of a club but as facilitators of a sandbox: a safe space to try things out and see what works.

    Our sandbox would be friendly and exploratory, but with a clear purpose: to model the use of general-purpose AI tools in everyday working environments. It would enable a number of labs, each modelling a different working situation, where we could try things out, see what helps and what doesn’t, and work out how to get better results.

    Hence Anapoly AI Labs: one sandbox, many lab setups.

    sandbox: a safe play area where computer programs can be used without affecting the operational system; useful for experimenting with or testing new software.

  • Our stance

    Stance: a way of thinking about something, especially expressed in a publicly stated opinion.

    We don’t claim to be AI experts. We’re practitioners exploring AI in real problems faced by professionals like us. We’re testing, documenting, and improving – in public. That’s our value.

  • Initial assumptions

    Having decided that the idea of an AI Club was worth pursuing, Dennis and I co-opted Ray into the initiative and set out its underlying assumptions.These are listed below.

    Assumption: there is a market for an AI Club amongst the local population of active and retired professionals, small business owners, and the like. These people are aware of AI’s potential, curious about it, but unsure how best to make use of it.

    Assumption: the social aspect of our club will make it a suitable, informal setting for people to learn how AI tools can improve the quality and productivity of their work.

    Assumption: Although different people will find different aspects of AI useful, there are some common purposes. These include:

    • supporting personal or professional development;
    • getting more done in less time;
    • creating better quality output;
    • improving the the quality of service offered to others; and
    • saving money.

    Assumption: a face-to-face, small-group format with peer interaction, real-time demonstrations, and a narrative focus will be more appealing than virtual courses or corporate-style workshops.

    Assumption: people will be willing to attend in-person sessions and contribute a modest fee once value is evident. They will think that the experience is more useful, trustworthy, and rewarding than online alternatives.

    Assumption: at present, no equivalent offering exists locally that blends live demonstration, peer learning, and practical AI support with such a low-barrier to entry.

    Assumption: Anapoly can deliver sessions using current, general-purpose AI tools such as ChatGPT, Perplexity and NotebookLM without a costly technical infrastructure, relying on existing facilities and minimal setup.

    Assumption: Anapoly, as a local consultancy run by experienced professionals, will be trusted by the audience and seen as non-threatening, practical, and thoughtful.

  • Exploring the idea with ChatGPT

    Once Dennis and I had come up with the idea of a small, local AI club, I used ChatGPT to help me explore its possibilities. I explained the idea to the AI and began the chat by asking: Is there a market for such a club in Plymouth? The answer was a helpful analysis of the main issues and a qualified yes; but the pitch would need to be “pragmatic rather than evangelical”. We would need to keep things down to earth.

    From there, we explored aspects ranging from possible niche markets to session formats. We talked about why people might want to use AI. We looked at different types of users and what each might want from a club like this.

    The conversation broadened my thinking about the viability of the idea and brought the issues into better focus. What started as a vague idea of “a club for people like us” grew into something I began to think we could actually try.

    As an aside, this research with AI was also research about AI: a simple use case, a small step on my path to AI proficiency.

  • The idea

    The idea for Anapoly AI Labs began with a newsletter.

    Kamil Banc wrote a thought-provoking piece on SubStack under the title “My Top 10 ChatGPT Features That Actually Matter At Work”. Early in the piece he wrote:

    “Most professionals approach ChatGPT like tourists at an all-you-can-eat buffet. They sample everything, master nothing, and walk away wondering why they feel unsatisfied. The harsh truth? Not all ChatGPT features deliver equal career value. While everyone else is busy playing with voice demos and testing the latest gimmick, the quiet power users are building workflows that make their bosses take notice.”

    I was in danger of falling into the “buffet” approach, too. But I had some points of focus. I am writing stories from my family history, and from time to time I needed to research how to produce specialist documents, for example a RAMS: a Risk Assessment and Method Statement for some building maintenance work. These tasks needed me to do more than “browse the buffet”. I began to think seriously about and to research how to get the best value out of ChatGPT.

    Dennis had been experimenting too, and Kamil Banc’s assertion resonated with us both. We felt sure that many people must be like us: interested inusing AI, but uncertain how to do so effectively.

    We felt, also, that many people,rather than do one of the many online courses or get bogged down in YouTube, would prefer to try AI in a more social environment.

    Hence the thought: what if we run a small club: get a few like-minded people in a room from time to time, talk about the real tasks AI can help us with, try out a smalll number of AI tools, and compare notes. That was the seed of the idea.