If you are going to use AI in your organisation to enhance knowledge work #productivity, what can you really expect as of today? 👀

This post is inspired by the work we have undertaken building 42ai, the RevOps AI Analyst created by the team here at 42 Problems, Inc. My thoughts are just a perspective, i'd love to hear what you think... 🙏

In theory, the disruptive potential of AI is extreme. But in practice something is happening, but we're not sure what just yet. Let's look at it from a few different angles:

Defined input

You know what you want to input (& hopefully why). You have an idea of what you want out of it, but you don't have a seriously tight output criteria that needs to be met. You do have standards though and will try a few times to get it to the level you need (Think along the lines of DALL-E Open Ai, Midjourney, Writer, Jasper and foundation models themselves).

Impact ⭐ 3/5 - These tools are immensely powerful and saving you A LOT of time. But it might feel more like you replaced the freelancer or the intern.

Defined output

There is only one correct answer. If the answer or analysis is incorrect, you lose trust immediately and/or look like an idiot (which was maybe true before AI). A very current example of this are the tools vying to disrupt Looker&Tableau. Essentially making everyone a Data & BI expert by putting a natural language prompt between business decision makers and data. No training or prior expertise required.

Impact ⭐ 3/5 - You might expect a higher rating here. But the reality is, when this use-case works a la ThoughtSpot the likelihood is AI is just there for the natural language interface. Code interpreter is good, but more on an individual task basis and there is a lot of work that needs to be done to make it accurate as an end-to-end analytical tool. In general defined output use-cases require A LOT of AI "hand" holding.

Automation ( in knowledge work)Beyond transcription and text extraction from images, this seems like an area where AI should be playing a bigger role? I like Chat GPT x Zapier but isn't that just the same as Zapier? 🤷♂️ And call summaries are wonderful and all but have you actually read any recently?

Impact ⭐ 1/5 - A few co-pilots aside. COULD DO BETTER.

The CONCLUSION (finally)

If you REALLY want to supercharge your business with AI, you need to be specific. Think about a specific team or role. Enable them with a super tight implementation. You need a product that does all three of the above (DI, DO, Automation). Now, given what else i have said, i'd recommend avoiding a tool that calculates your P&L, creates a presentation and presents it to the board all by itself. For now 🤔

So are there tools out there? A few - my lucky examples for today are: Microsoft CoPilot Clay & Copy.ai - but i haven't used Microsoft since school, so... what I like about Copy & Clay is the mix of AI workflow automation, AI content but with some key defined output criteria... for specific teams.

What are your thoughts?