Unlock the power of voice memos and AI with our step-by-step guide. Learn how to automate your thoughts into actionable tasks using Zapier, OpenAI, and Claude.
Great to hear the piece gave you inspiration for some possibilities on how to apply this!
As to your questions:
1. Paid subscriptions
Zapier: You'll need a Zapier subscription to create zaps with more than two steps, which the workflow I presented requires. They do give you a 14-day free trial that includes multi-step zaps, so that's one way to try it out.
Another option would be to do your transcriptions and the uploading to an AI manually first to see if you like the results, before building the entire workflow.
I really like MacWhisper, which is macOS app that only charges a one-time payment and does a great job at transcriptions. You could use that, and then attach the voice memo transcription to an interaction in ChatGPT or Claude's consumer chat interface.
Claude and OpenAI APIs: you'll need credit on Claude and OpenAI accounts if you do build out the entire workflow. The good news is that those don't require subscriptions. You just put $10-$20 in the accounts, and that credit will last you a long time if all you're doing is things like these voice memo workflows.
Another option is to use OpenAI's models for both the transcription step and the processing step of your voice memo. That way, you only have to put credit in an OpenAI API account. (The only caveat being that I personally think Claude's models are currently better for this kind of task.)
2. Feedback on proposals
Usually there's an aspect of a proposal I'm thinking about, and I'll talk the AI through it in my voice memo. E.g., "A friend of mine has asked me to join their startup as an advisor. Their proposal is X, but I'm concerned about Y and Z. I'm considering to do A or B."
Of course, if you want specific feedback on the complicated terms of an agreement or something like that, you'll need the attachment. In such cases, I might still record a voice memo with my thoughts and questions, mention that I'll later share an attachment, and then when I get to my computer I'll just copy my original voice memo + the AI's initial output + the proposal as attachments into Claude or OpenAI's consumer chat interface and run that through for its feedback.
TL;DR: I think a good way to test out the value of all this is to start with recording some voice memos, transcribing them manually (for example with MacWhisper), and then attaching them to a conversation in Claude (or ChatGPT) and see the results you get. If you like it, then it's worth setting up this automated workflow.
Happy to answer more questions or thoughts if you have them!
Hi Tim, fantastic add-on explanation, thank you. I think as a consultant or anyone giving feedback, writing, advise or ideas in any field, this will be very interesting and worth while. Of course it will all eventually put together into one easy app, but until then it’s up to the pioneers to test and try, but also have the potential technological advantage over those not willing to experiment. However I am missing the quality of an experienced advisor in this way of working, especially when thinking of context. I have a few friends who are experts in their field. I go to them for advice, yes it’s not as quick as AI, but it is however perfectly in context and personally adapted to my situation and needs. Not everyone has these kind of connections, so there is a market for AI. It’s interesting to see where this will go.
Yes that’s what I mean, at least for now, my opinion is likely to change over time when AI has developed further. But specifically when context is of importance or specific situations that are out of the ordinary / not generic. The human factor allows for finely tuned context, a little flair, originality and for instance being bold when the chance arises and in context. For instance doing something bold and unexpected in a brief to a client, it might just be that factor that wins over the client.
Fantastic article Tim,
My mind is spinning a thousand miles an hour thinking about the possibilities for this. Some practical questions though:
You mention feedback on proposals. How can I get feedback on a proposal that I wrote if I cannot upload an attachment with the actual proposal?
To set this up, do I first need to have paid subscriptions on several platforms such as Zapier? Or are there free plans on all needed platforms?
Just weighing the investment versus returns.
Regards John
Hi John,
Great to hear the piece gave you inspiration for some possibilities on how to apply this!
As to your questions:
1. Paid subscriptions
Zapier: You'll need a Zapier subscription to create zaps with more than two steps, which the workflow I presented requires. They do give you a 14-day free trial that includes multi-step zaps, so that's one way to try it out.
Another option would be to do your transcriptions and the uploading to an AI manually first to see if you like the results, before building the entire workflow.
I really like MacWhisper, which is macOS app that only charges a one-time payment and does a great job at transcriptions. You could use that, and then attach the voice memo transcription to an interaction in ChatGPT or Claude's consumer chat interface.
Claude and OpenAI APIs: you'll need credit on Claude and OpenAI accounts if you do build out the entire workflow. The good news is that those don't require subscriptions. You just put $10-$20 in the accounts, and that credit will last you a long time if all you're doing is things like these voice memo workflows.
Another option is to use OpenAI's models for both the transcription step and the processing step of your voice memo. That way, you only have to put credit in an OpenAI API account. (The only caveat being that I personally think Claude's models are currently better for this kind of task.)
2. Feedback on proposals
Usually there's an aspect of a proposal I'm thinking about, and I'll talk the AI through it in my voice memo. E.g., "A friend of mine has asked me to join their startup as an advisor. Their proposal is X, but I'm concerned about Y and Z. I'm considering to do A or B."
Of course, if you want specific feedback on the complicated terms of an agreement or something like that, you'll need the attachment. In such cases, I might still record a voice memo with my thoughts and questions, mention that I'll later share an attachment, and then when I get to my computer I'll just copy my original voice memo + the AI's initial output + the proposal as attachments into Claude or OpenAI's consumer chat interface and run that through for its feedback.
TL;DR: I think a good way to test out the value of all this is to start with recording some voice memos, transcribing them manually (for example with MacWhisper), and then attaching them to a conversation in Claude (or ChatGPT) and see the results you get. If you like it, then it's worth setting up this automated workflow.
Happy to answer more questions or thoughts if you have them!
Tim
Hi Tim, fantastic add-on explanation, thank you. I think as a consultant or anyone giving feedback, writing, advise or ideas in any field, this will be very interesting and worth while. Of course it will all eventually put together into one easy app, but until then it’s up to the pioneers to test and try, but also have the potential technological advantage over those not willing to experiment. However I am missing the quality of an experienced advisor in this way of working, especially when thinking of context. I have a few friends who are experts in their field. I go to them for advice, yes it’s not as quick as AI, but it is however perfectly in context and personally adapted to my situation and needs. Not everyone has these kind of connections, so there is a market for AI. It’s interesting to see where this will go.
John
Do you mean that AI doesn't feel as qualified as an experienced (human) advisor?
Yes that’s what I mean, at least for now, my opinion is likely to change over time when AI has developed further. But specifically when context is of importance or specific situations that are out of the ordinary / not generic. The human factor allows for finely tuned context, a little flair, originality and for instance being bold when the chance arises and in context. For instance doing something bold and unexpected in a brief to a client, it might just be that factor that wins over the client.
That's definitely true. Though sometimes AI can come up with unexpected ideas, so it's like having an additional sounding board / brainstorm partner.