Gen AI – our slope of enlightenment
After a fantastic couple of weeks away recharging the batteries, we’re back with a vengeance and brimming with a bucketload of ideas about how we can help clients in the run up to Christmas (128 days – just saying).
This week we’re back on the topic of Gen AI. It has dominated the LINAR summer in a variety of areas. One consistent thread we’re hearing from clients is that they are all struggling to find the jumping off point for using Gen AI within their day to day activities. As a result, quite a few are languishing in the trough of disillusionment and wondering how to start climbing the slope of enlightenment. If you’re struggling with this terminology, watch the guys at Gartner give a brief intro to their hype cycle.
This week’s Three Things describes the LINAR Gen AI journey to date and the fun we’re having along the way. Enjoy.
#1. Learn by doing
There are a plethora of companies out there at the moment with some really smart people harnessing the power of Gen AI to solve everyday business problems. The great thing is that most of these companies have free trials where you can test their products. We’ve been having some fun recently with the following:
- fireflies.ai: an AI notetaker, fireflies.ai joins every online LINAR meeting and captures an accurate record of what’s been said, agreed and the follow up actions. Using the keyword functionality, we’re starting to train fireflies.ai to pick up on certain terms and phrases so that the action lists produced are more accurate and require less editing. The sentiment analysis can also show trends for our sales calls and predict outcomes based on factors including the amount of time we’ve spent talking rather than listening.
- Superwhisper: hands down, in our view, the best gen AI dictation device on the market currently. We really like the ability to choose what you’re doing (email, note, general etc) and get different outputs based on the model you choose. In fact, the base for today’s Three Things was dictated as I was walking around Richmond Green yesterday lunchtime with the dog!
- kore.ai: allows low code/no code development of chatbots (AI assistants) based on sources of data defined by you. One thing we’re playing with is an AI assistant (chatbot) to help with new quotes/pitches. We’ve loaded in anonymised versions of all previous quotes and then ask the chatbot to generate a new quote based on a structured set of questions and historical data.
#2. Listen to your clients
Nobody we’ve spoken to has cracked Gen AI and/or implemented it successfully across all parts of their business. Firms around the city are at wildly different ends of a spectrum that begins with fear and rejection and continues through to cautious adoption.
What has become clear during our client conversations is that most firms recognise that Gen AI is not another fad. The technology is here to stay and its use will start to permeate the roles we all perform, whether we like it or not.
From a LINAR perspective, we’re building a bank of interesting use cases based on the conversations we’re having with clients and also asking ourselves the simple question, “How could Gen AI help with this?”, in every job we do. We’ll share this list with you in a later edition of Three Things.
#3. Be transparent
We’ve been thinking A LOT about the future and how Gen AI will shape the work LINAR will be doing. Our GenAI lead, Sam Stamp, has developed a simple categorisation to show transparently where we’re using Gen AI outputs. Going forward, all our client facing work will be assigned a category:
- AH (All Human): good old fashioned human grunt. Not a chatbot or LLM in sight.
- PA (Part AI): we’ve used a commercial Gen AI platform (e.g. Chat GPT, Gemini etc) to help us with a piece of work. A lot of the time, this takes the form of what Sam calls a “blank page problem”. You have an idea in your head of what you want to do but are struggling to start putting it down into words. Chat GPT is BRILLIANT at providing a first draft. Equally, we could have prompted a platform to conduct some competitor research, define a Total Addressable Market (TAR) or help us design a social media plan. We think there are a LOT of exciting possibilities in this category.
- AA (All AI): an output totally generated by Gen AI with no human correction. Obviously, like most other sensible businesses, we are aware of the risks associated with pushing content out without a human pair of eyes so have nothing in this bucket, yet. However, with the advancement of agentic AI (see short 7 min article here), we’re very conscious that areas like marketing and sales are ripe for a well trained Gen AI platform to start interacting with targets independently of the human LINAR team. For example, platforms like La Growth Machine are already seeing success in this area.
In case you were wondering, we’ve categorised this week’s Three Things as PA (Part AI) because I used Superwhisper to create my draft!
