Latest / How I Work / 8 tasks you should never do manually again
Transcript
- 0:00Speaker 1: Hey, before we get going, you know that feeling when
- 0:03Speaker 1: you're using AI but you're still doing some tasks manually,
- 0:06Speaker 1: over and over again. Well, Neo and I built the
- 0:09Speaker 1: AI Agent boot Camp to fix exactly that.
- 0:12Speaker 2: Four weeks fully.
- 0:13Speaker 1: Virtual and you walk away with agents already built and running.
- 0:16Speaker 1: Spots are limited, and the details are in the show notes.
- 0:20Speaker 1: AI has promised us so much when it comes to productivity,
- 0:23Speaker 1: and yet most of the people that Neo and I
- 0:26Speaker 1: speak to are still doing a surprising number of tasks
- 0:29Speaker 1: completely manually, like manually scrubbing through needing transcripts for action items,
- 0:35Speaker 1: tabbing between LinkedIn Google News and the inbox before every
- 0:38Speaker 1: sales call, spending twenty minutes writing an executive summary for
- 0:42Speaker 1: a paper they just finished writing. And this is not
- 0:44Speaker 1: because they don't have access to AI, because they do,
- 0:48Speaker 1: but they're.
- 0:48Speaker 2: Just not using it for the right things.
- 0:51Speaker 1: So today Neo and I go through eight tasks that
- 0:54Speaker 1: you should never have to do manually again, and by
- 0:59Speaker 1: the end of today's so you will have a really
- 1:01Speaker 1: clear picture of where agents can start turbocharging your productivity
- 1:05Speaker 1: so you're not spending time on manual grunt work ever. Again,
- 1:14Speaker 1: Welcome to how IAI with me, doctor Amantha Imba, and
- 1:18Speaker 1: Neo Applin, head of Inventium AI. Each episode we share
- 1:23Speaker 1: one practical way to use AI better at work and
- 1:26Speaker 1: in light, no fluff, no tech jargon, just things you
- 1:30Speaker 1: can use straight away.
- 1:32Speaker 3: So a lot of people are using AI in their
- 1:34Speaker 3: day job and they're getting good benefit out of it,
- 1:37Speaker 3: but not enough people are using AI to take things
- 1:40Speaker 3: off their plate entirely. So what we're going to go
- 1:44Speaker 3: through today is some things that knowledge workers should probably
- 1:49Speaker 3: not really be doing ever again, just by themselves. This
- 1:53Speaker 3: might be a you and AI thing, but certainly how
- 1:56Speaker 3: do you actually get rid of things off your plate?
- 1:58Speaker 3: Now we've been talking about agents, so these are really
- 2:01Speaker 3: replicable things that you could build into an agent to
- 2:05Speaker 3: help make your day job better. All right, So, Amantha,
- 2:09Speaker 3: where do you think that most people should be using
- 2:11Speaker 3: agents to get rid of those things so they cannot
- 2:14Speaker 3: do them ever again.
- 2:16Speaker 1: Well, one of the areas where we spend so much
- 2:20Speaker 1: time of our working life is meeting. So the average
- 2:25Speaker 1: knowledge worker spends around about twenty one and a half hours.
- 2:29Speaker 2: A week in meetings. That the most the latest stats
- 2:32Speaker 2: I have read.
- 2:34Speaker 1: And even though there's so much talk around how powerful
- 2:38Speaker 1: AI is when it comes to meetings. Certainly in terms
- 2:41Speaker 1: of preparing for a meeting, during the meeting, and following
- 2:44Speaker 1: the meeting, I find that with the meetings I'm in,
- 2:48Speaker 1: very few people are using AI in those meetings. I
- 2:51Speaker 1: would say less than five percent. I am using AI
- 2:56Speaker 1: in one hundred percent of my meeting. So I use Granola,
- 2:59Speaker 1: which is software that we've spoken about on the pod before,
- 3:02Speaker 1: which is just quietly transcribing everything that is being said,
- 3:07Speaker 1: just in the background. It's not joining, it's not making
- 3:09Speaker 1: a big announcement saying Amantha's AI note taker is here
- 3:12Speaker 1: even if I'm not. It's just doing it in the background,
- 3:15Speaker 1: which means that I'm literally using this workflow for every
- 3:19Speaker 1: single meeting. So certainly, if you are not using AI
- 3:23Speaker 1: for meetings, this is manual task that you should never
- 3:26Speaker 1: be doing again. In terms of identifying what were the
- 3:29Speaker 1: actions for all parties to do at the end of
- 3:32Speaker 1: a meeting, what were the key decisions that were made,
- 3:35Speaker 1: what is a summary.
- 3:37Speaker 2: Of the meeting?
- 3:38Speaker 1: And I will add, while I do have an agent
- 3:40Speaker 1: to do that, I will always use human judgment as well,
- 3:43Speaker 1: particularly for summaries of meetings, because I find if I
- 3:47Speaker 1: am not going through that meeting summary with human judgment.
- 3:50Speaker 2: It's not quite what I want it to.
- 3:53Speaker 1: Be because inevitably, you know there'll be chitchat that happens
- 3:56Speaker 1: in a meeting where you go off tangent, of course,
- 4:00Speaker 1: and I don't need that in the summary to make
- 4:02Speaker 1: it a useful summary. So that's one thing for me. Neo,
- 4:07Speaker 1: what's something that you are not doing manually?
- 4:11Speaker 3: I'll just jump onto the meetings part if I can.
- 4:13Speaker 3: The meetings are really interesting sometimes if you've got lots
- 4:16Speaker 3: of people talking about things in different orders, and I've
- 4:18Speaker 3: found that sometimes the human check is really important because
- 4:21Speaker 3: you might say something about the task that needs to
- 4:24Speaker 3: be done, but not who's going to do it, and
- 4:26Speaker 3: then there's conversation and then oh yeah, NEO can do it.
- 4:30Speaker 3: And then that's about three minutes later, and then a
- 4:32Speaker 3: couple of minutes, oh yeah, maybe it's going to be Neo,
- 4:34Speaker 3: maybe it's a manther, and then okay, we'll need it
- 4:36Speaker 3: done by Thursday. And there's three different things that are
- 4:39Speaker 3: being mentioned within probably a five minute period. AI is
- 4:42Speaker 3: not great at putting those three together succinctly. So a
- 4:45Speaker 3: bit of a tip for meetings if you'd like internal
- 4:47Speaker 3: team meetings, just summarize what's been agreed. You can just say, oh,
- 4:51Speaker 3: Neo's going to do it by Tuesday, right, and you
- 4:53Speaker 3: get a much better summary at the end to a
- 4:54Speaker 3: better result.
- 4:56Speaker 1: Now, another thing with meetings is that I don't know
- 4:59Speaker 1: about you, Nao, but quite often in my work life
- 5:02Speaker 1: I am meeting with people that I've never met before.
- 5:04Speaker 1: So I look after a lot of the sales and
- 5:06Speaker 1: business development for Inventium and Inventium AI, and I want
- 5:12Speaker 1: a quick briefing on the person that I am meeting with,
- 5:15Speaker 1: And yes, I could do and used to do this
- 5:17Speaker 1: very manually. I'd go to LinkedIn, I'd have a little
- 5:20Speaker 1: look at the person's profile. I'd do a Google news
- 5:24Speaker 1: search to see as this company been in the news lately.
- 5:29Speaker 2: It's a lot of work, you know.
- 5:30Speaker 1: I might go through emails to see have I emailed
- 5:33Speaker 1: people with this company. I might look in our CRM
- 5:36Speaker 1: to go what have we done previously. But this is
- 5:39Speaker 1: a really great use case for building an agent that
- 5:42Speaker 1: goes through the steps automatically that you would normally have
- 5:46Speaker 1: to do yourself in a very manual way to get
- 5:49Speaker 1: a summary in the way that you like it with
- 5:52Speaker 1: the knowledge and points that are important to you in
- 5:55Speaker 1: your context. So I think that is another excellent use
- 5:58Speaker 1: of agents and things that you should certainly not be
- 6:01Speaker 1: doing manually.
- 6:03Speaker 3: Again, for me, another one is industry news. I like
- 6:07Speaker 3: to keep up to date with things we've talked about,
- 6:09Speaker 3: researching and things like that, but sometimes, particularly if you
- 6:12Speaker 3: need to know things about like up to the second
- 6:15Speaker 3: up to date, it's great to get that industry news
- 6:17Speaker 3: dropped in automatically if you'd like. So what you can
- 6:20Speaker 3: do is say, show me what's happened in my industry
- 6:23Speaker 3: for the last week or with this particular topic in
- 6:26Speaker 3: the last week. Of course, be more specific about that. Heck,
- 6:29Speaker 3: you could even grab the research stuff we talked about
- 6:31Speaker 3: in the previous episode and put that into your thing.
- 6:34Speaker 3: But you can. That's a great way to keep up
- 6:36Speaker 3: to date with industry knowledge. Rather than it's being a
- 6:40Speaker 3: generically briefing, this one becomes personalized for you, your job,
- 6:44Speaker 3: your role, what you need to know and for that
- 6:46Speaker 3: last period of time, like the last twenty four hours,
- 6:49Speaker 3: forty eight hours, two weeks, one month, whatever. I encourage
- 6:52Speaker 3: you to do this regularly, but certainly do it when
- 6:54Speaker 3: you get back from holidays. What happened over the last
- 6:56Speaker 3: couple of weeks after I was on holidays. Very handy
- 6:58Speaker 3: thing to do.
- 7:00Speaker 1: Now, Nia, do you have an agent for comparing options
- 7:03Speaker 1: before a purchase decision?
- 7:06Speaker 3: Oh? God, I do this so many times, but each
- 7:08Speaker 3: time I need to compare slightly different things. If I'm
- 7:10Speaker 3: going to search for I use hiking boots as an example,
- 7:13Speaker 3: because I bought one of those last year. If I'm
- 7:16Speaker 3: searching for hiking boots, that's a very different thing than
- 7:18Speaker 3: if I'm going to be searching and comparing software systems
- 7:22Speaker 3: and things like that. But I certainly do have a
- 7:26Speaker 3: comparison agent there. It's actually a pretty simple prompt. The
- 7:31Speaker 3: tip there is compare these two things. You may want
- 7:34Speaker 3: to be specific about the columns. You want it to
- 7:37Speaker 3: be comparing in a table, and I really love tables
- 7:40Speaker 3: for comparison. So compare A with B. Here's the columns,
- 7:44Speaker 3: and then what it'll do is to a comparison there
- 7:46Speaker 3: and give me the pros and cons and things like that.
- 7:49Speaker 3: Pro level tip. If you want to say I want
- 7:51Speaker 3: to use rag r ag red, amber green emojis as well,
- 7:57Speaker 3: it'll also give you a nice little green orange read
- 8:00Speaker 3: flag next to each of these comparison to show you
- 8:03Speaker 3: in which of these dimensions that thing is stronger or
- 8:05Speaker 3: not as strong. So yeah, comparison is amazing.
- 8:09Speaker 1: Using agents, now I have got well, look, I've built
- 8:13Speaker 1: a few agents that do editing and proofreading for me.
- 8:17Speaker 1: I found that this was particularly useful when I was
- 8:20Speaker 1: working on my fifth and most recent book, The Energy Game,
- 8:23Speaker 1: which comes out on July seven, twenty twenty six, and
- 8:28Speaker 1: I created agents that would help me with editing in
- 8:34Speaker 1: terms of looking for things where my argument didn't flow,
- 8:40Speaker 1: where paragraphs weren't as engaging as they could have been,
- 8:44Speaker 1: where just the writing felt a little bit flat, and
- 8:47Speaker 1: then I could use that editorial direction to then go
- 8:51Speaker 1: back and do the edit obviously myself as a human.
- 8:55Speaker 1: I also find that proofreading is really good and one
- 8:58Speaker 1: thing I find very very useful. And this is where
- 9:02Speaker 1: it's really important the instructions that you give to the agent,
- 9:05Speaker 1: because if you don't give good instructions to your agent,
- 9:09Speaker 1: then the risk with using AI for editing and proofreading
- 9:13Speaker 1: is that it can turn something that is human generated
- 9:16Speaker 1: into something that looks like it was AI generated, which
- 9:19Speaker 1: course is going to turn a lot of people off
- 9:21Speaker 1: and potentially do reputational damage as well.
- 9:24Speaker 3: That's huge.
- 9:25Speaker 1: It's really really huge again, which is why knowing how
- 9:29Speaker 1: to build an agent, and again I'm going to do
- 9:31Speaker 1: a plug for our AI Agent boot camp which inventium
- 9:34Speaker 1: AI is running kicking off in July, is really important
- 9:38Speaker 1: that you build these well. So for example, with my
- 9:41Speaker 1: proofreading agent and also my editing agent, I want it
- 9:45Speaker 1: to catch where I am using phrases or words or
- 9:52Speaker 1: structures that feel AI.
- 9:55Speaker 2: And I must say, I find this.
- 9:56Speaker 1: Really really annoying because there are things that I do
- 10:00Speaker 1: I'm writing, if I'm doing a first draft as a human,
- 10:03Speaker 1: as I always do, there are.
- 10:06Speaker 2: Things that AI does that I do as a human.
- 10:09Speaker 1: And I got to say, this shits me so much
- 10:13Speaker 1: like the it's not X, it's Y, or variations of
- 10:16Speaker 1: I think that's called corrective antithesis.
- 10:18Speaker 2: I use that in my writing, and now I have.
- 10:21Speaker 1: To deliberately edit my writing me as a human that
- 10:26Speaker 1: I would naturally say, and I have to change that,
- 10:28Speaker 1: and it feels very artificial, but I know that I
- 10:32Speaker 1: need to do that because if I use too many
- 10:34Speaker 1: corrective antithesises in say a chapter, people are going to
- 10:39Speaker 1: read that and go oh, that was written by AI
- 10:41Speaker 1: when actually it was written by me.
- 10:43Speaker 2: But I need to use that agent to actually clean
- 10:46Speaker 2: that up.
- 10:47Speaker 1: To get rid of those Tellesides, I don't know if
- 10:50Speaker 1: I explain that clearly, but it's certainly a very big
- 10:52Speaker 1: frustration for me as a writer, and I've heard that
- 10:55Speaker 1: a lot from a lot a lot of other writers
- 10:57Speaker 1: as well. What other agents have you built or even
- 11:03Speaker 1: do you intend to build, neo, because you know the
- 11:05Speaker 1: power of agents. I know you've built probably thirty or
- 11:08Speaker 1: thirty five over the last few months, and I feel
- 11:10Speaker 1: like every week you're building a new agent. What's like
- 11:13Speaker 1: on your list even of agents that you're like, oh,
- 11:16Speaker 1: this could be solved, This manual job could be solved
- 11:19Speaker 1: by building a great agent.
- 11:21Speaker 3: This is one that I've not needed as an agent,
- 11:24Speaker 3: but in my old jobs I would have needed. I
- 11:27Speaker 3: believe executive summaries should not be written by humans. How
- 11:33Speaker 3: many times have we written a big analysis paper whatever,
- 11:35Speaker 3: and then you've then got to write the executive summary. Now,
- 11:38Speaker 3: some people say write the execu summary first. If that's you, great,
- 11:42Speaker 3: do that, But if you've already written the entire paper,
- 11:45Speaker 3: is it really worth your while to spend twenty minutes
- 11:48Speaker 3: crafting and awesome executive summary, or is it better to
- 11:51Speaker 3: get AI to craft them the way you want to
- 11:52Speaker 3: craft them, and then you look over it, tweak it
- 11:54Speaker 3: and improve it. And I reckommen. For a lot of
- 11:56Speaker 3: knowledge workers we're writing papers, that's a great one for me.
- 11:59Speaker 3: The next thing I'm going to be building is a
- 12:01Speaker 3: learning agent. And what I do is I often get
- 12:05Speaker 3: some weird and wacky term. There's this thing where maybe
- 12:08Speaker 3: it's an industry thing because we consult with different industries
- 12:10Speaker 3: and I need to know those terms. And I'm a
- 12:12Speaker 3: generalist and I like to learn lots of these things.
- 12:14Speaker 3: So it's kind of like it'll give me. I give
- 12:17Speaker 3: it the thing I want to learn the context of
- 12:20Speaker 3: what I want to learn about it. Because you don't
- 12:21Speaker 3: want to just say plumbing, you know to be plumbing
- 12:24Speaker 3: because I've got a blocked thing. You know, you need
- 12:26Speaker 3: to say why, and also to learn ask you where
- 12:29Speaker 3: your base is. So once you say I know this,
- 12:32Speaker 3: I want to know that it will then draw for
- 12:35Speaker 3: you the learning along the way and help you to
- 12:38Speaker 3: learn that. So because I'm always doing that manually, and
- 12:41Speaker 3: I just realized as we're chatting, that's something I could
- 12:44Speaker 3: build into an agent, a learning agent to help me
- 12:47Speaker 3: to learn.
- 12:49Speaker 2: Now, one final thing I would say is with people's.
- 12:51Speaker 1: Inbox, and the caveat here is that your AI needs
- 12:54Speaker 1: to be connected to your inbox, so you know, within
- 12:58Speaker 1: the Microsoft ecosism, this would be having co pilots in
- 13:02Speaker 1: your outlook. If you're on Gemini, that's pretty straightforward because
- 13:08Speaker 1: it will connect to your Gmail. If you're in Claude,
- 13:12Speaker 1: that can definitely connect to the Google ecosystem, and it
- 13:15Speaker 1: can connect to Microsoft three six five chat chip.
- 13:20Speaker 3: It's called a connector if you want to have a
- 13:22Speaker 3: look at it, and you can do a connector with
- 13:24Speaker 3: CHUTCHBT and Claude so they can both see your emails
- 13:27Speaker 3: and calendars and other things as well.
- 13:29Speaker 2: And so Neo.
- 13:30Speaker 1: If you were advising people what kind of agents are
- 13:34Speaker 1: most useful for conquering the inbox, which is a source
- 13:37Speaker 1: of pain for most people, what would you recommend?
- 13:42Speaker 3: Treat it like it's the assistant you wish you had.
- 13:45Speaker 3: But this a system can't do anything. I can just
- 13:47Speaker 3: tell you things, and so it may be give me
- 13:51Speaker 3: a list of the last emails that came in the
- 13:53Speaker 3: last twenty four hours or the last week. Maybe there's
- 13:55Speaker 3: different agents you want to do those things, and you
- 13:57Speaker 3: can say which ones are important. Maybe these are my bosses,
- 14:00Speaker 3: or my colleagues or my stakeholders. Make sure they're put
- 14:04Speaker 3: at the top, and even then give me recommendations on
- 14:07Speaker 3: what I need to do with these emails. Maybe it's
- 14:10Speaker 3: draft them. Whatever you want your assistant to do for you.
- 14:14Speaker 3: That's what you program your agent to do, and you
- 14:16Speaker 3: might have multiple agents for that. It might be tell
- 14:18Speaker 3: me about my last twenty four hours of emails. Maybe
- 14:20Speaker 3: that's one task for an agent to do, and then
- 14:24Speaker 3: it might be for the sales emails. Help me respond
- 14:28Speaker 3: back with each of these things, and so you can
- 14:30Speaker 3: bring your agents into the conversation to help you to
- 14:32Speaker 3: solve those problems with your emails. Because I don't know
- 14:35Speaker 3: about you. I hate emails and I get stuck with them,
- 14:38Speaker 3: but getting a quick pass of them is very useful sometimes.
- 14:42Speaker 1: Noia, I feel like we could just go on for
- 14:44Speaker 1: hours listing different agents that can replace manual tasks, but
- 14:48Speaker 1: we are not going to do that. Instead, if you
- 14:51Speaker 1: are keen to dig deeper into the world of agents
- 14:55Speaker 1: and know how can you even identify where in your
- 14:58Speaker 1: day to day tasks in your workflow. Is there the
- 15:01Speaker 1: opportunity to really reduce manual work with an agent? Then
- 15:06Speaker 1: hop into the show notes there is a link to
- 15:08Speaker 1: enroll in our AI Agent boot Camp, which I can
- 15:12Speaker 1: guarantee will absolutely transform how you are approaching these annoying
- 15:18Speaker 1: manual tasks, the grunt work. You will never have to
- 15:21Speaker 1: deal with those things in the same way ever again.
- 15:25Speaker 1: How AI was hosted by me, Amantha Imber and Neo Applan.
- 15:29Speaker 1: A big thank you to Martin Imber who does our
- 15:31Speaker 1: sound editing, and Jim Rubio for production support, and thank
- 15:35Speaker 1: you to John Kilby who composed the theme music.