Name: Reworking Work
Uploaded: 2025-12-02
Duration: 15 min 28 s

AI is taking aim at the hidden workplace inefficiencies that slow companies down —from automating routine and repetitive tasks to generating insights to streamline decision-making. But how does this actually change businesses, and what becomes possible when teams finally get their time back? Executives from leading tech platforms explain how companies can use AI to provide intelligent document processing, create consistent step-by-step guides for staff onboarding, and better recruit and retain employees. What’s actually working? What do leaders get wrong about increasing productivity? Where does human judgment still matter most?

Reworking Work

Conferences

So you just heard there, AI promises to end some aspects of drudge work at the office, solving inefficiencies by automating repetitive tasks and redesigning the company workflow. But how will that transform the way businesses operate day to day? Joining us next are executives from two companies that are taking on the big challenge of reworking the way we work. Scribe is a firm that uses AI to create consistent step by step guides. For staff on boarding as well as document complex operational processes within companies, and GoodNotes is a note taking app with AI enabled digital capabilities that was named Apple's 2022 iPad app of the year. Joining us are Joseph Linn, AI champion lead of Goodotes, and Jennifer Smith, CEO of Scribe, who will be participating virtually. They will be interviewed by Fortune's editorial director for live Media, Kristen Stoller. All right, good morning everyone. Joseph, thank you so much for joining me and we have Jennifer here all the way from California. Thank you for being here. So I want to start off by getting a sense of the problems both of you were trying to solve when you started your companies, when they started, and how you are actually solving them now. Jennifer, let's start with you. There's a joke in Silicon Valley that every company wants to change the world. We want to change how the world works. So if you think about how most people spend their 9 to 5 right now, it's a gross misallocation of time and talent, and I would. The most powerful force in the universe is human potential, what people are able to do, and so we wanted to make it that when people show up to do work they are doing their best, highest calling. Hm, very cool. And explain to us exactly how scribe works. Subscribe will watch experts do work like people do the things they normally know how to do every single day and it'll automatically capture that know-how and turn it into a bunch of different formats. The basic primitive is a step by step guide with screen. Shots and instructions on how do I do this process so you've now essentially like sucked out of somebody's head what they know how to do like the true knowledge where the rubber meets the road. What am I doing when I'm fingers on keyboard every day trying to create value for this team, this company. And now I'm making it in a format that we can deliver directly to a colleague, a client, a customer, anybody else when they're going to do that same workflow, they'll now do it as good as that expert the first time they do it every time they do it. They're basically making knowledge like completely fluid across a team or across an organization, so no longer having to have someone just sit next to someone and and watch physically. Yeah, which happens a lot, right? We we're replacing the proverbial shoulder tapping or the quick zoom link share or Slack message of, hey, can you just show me how to do this thing? We're trying to make it seamless and automatic across everybody. Very cool. And Joseph, what about you? Good notes before I start, can I just say I also use Scribe and I'm a huge because documentation is so important to being able to leverage AI effectively. So what Goodnos is trying to do is we want to build the best medium for human AI collaboration. Think about it. So the pri primary way we use AI is through chatbots, but chatbots, we have to be able to do something better. Than that, right? We are now at the Nokia 3310 era of AI and then we believe the future of AI interaction has to be much more intuitive, much more pleasurable. So, um, good notes, we are a handwriting stylist-based. A and then we believe there is a lot of undiscovered richness to be found in trying to go for a stylus and voice-based AI workflow. And who are your users? Is it companies, certain types of people, who are you seeing most often using it? Students, teachers, uh, architects, uh, crime scene investigators, people of all walks of life. So imagine you're an architect, you're walking through the site with your civil engineer and you start the transcription, right? So that your discussion is captured and that will be a minute. At the same time you import the PDF of your site plan, you mark. It up, you scribble things in the margins, right, and then something doesn't look right, so you just snap a picture and then you can scribble another sticky note and then when you are done with the whole thing, you can just use the Ask good notes, good notes AI feature and then it'll just look at all the handwritings and the PDFs and the discussion and the everything else and give you a An answer based on that. So that's, I think, a more fluid natural way to use AI. Interesting. Now I'm going to ask you both a question that I asked this to a lot of CEOs, and they either laugh in my face or they shudder and get very scared when I ask this, but thank you both for bearing with me, which is how exactly are you quantifying success using AI right now? Is it how exactly are you measuring? Is it productivity? Is it revenue? What to you is a form of success here, Jennifer? Um, so I'll have two answers. One's internal how we measure it inscribe, and the second is how our customers are measuring using scribe. So our customers will apply it to whatever they're trying to do. The lowest common denominator, easiest way for folks to think about it is by time saving. Right, and so our customers have done really detailed time studies on their own, very obsessed with efficiency, uh, of what Scribe has meant for them, and they found they've saved anywhere between 35 to 42 hours per person per month of time so that's a work week a month of time back that they found for folks. Um, and then depending on the use case folks will say, hey, I'm trying to on board, uh, new hires faster, so how much faster am I able to ramp people or I'm trying to improve customer satisfaction. What's, uh, what's happening to my CSAT scores or I'm trying to drive error reduction. What's happening to my compliance or the accuracy of the tasks that are being done, um, and so that's really fun and interesting for us to work with different customers. We're trying. Achieve different things, um, and seeing how they measure that and helping them through that excellent for ourselves internally, oh, our internally we're, we're really fast growing company and so it's always how do we do more with very limited resources, um, because our biggest bottleneck is always hiring, um, and, uh, and having enough talent and so anything we can do to, you know, help our people just get more done in the day is a big win. Sounds like we have a lot of very similar problems, Jennifer. Uh, so internally we basically track AI based on time saved and tasks executed. So if in the past you used to have to do like a due diligence for procurement and then you need to look up all these things and then compile them into a report, nowadays you should be able to run one deep research query and that it should be done in no more than 5 minutes, right? And then, uh, at the same time, Uh, the number of tasks executed, if each task only takes 5 minutes now, you should be able to do more in the same given amount of time. So, uh, the marketing team, whereas in the past you might have 3 to 5 variations of social media copy, now you might be expected to generate 50 in the same span of time, and that goes to the other part, which is actually like AI. Really great, but it also really doesn't change the way business is done. We are trying to solve for the same pain points that customers have, and so a lot of the success metrics remain the same. Are users happy with our app? Are they saying good things about us? How many support tickets are we getting? How quickly can we resolve those queries? And all of those things, none of those things actually change. It's just that we expect the bar to be raised. Because now everyone is empowered with AI. That's a really good point because I, my next question was, you know, when humans are getting their time given back to them, what should they be doing instead? You said building, you know, more tickets, more marketing things, more social media company. Jennifer, is there something in your experience that you, you know, advise the companies who use your platform like, oh, they can be doing more X or more Y or where do you think humans still fit into the mix? I think humans will always maybe this is a contrarian thing to say, humans will always still fit in the mix, and I think what we're doing is we're just going up a level each step, and humans are gonna go up a higher and higher level of abstraction, right? So if you take the example of tickets, for example, it's gonna go from how many tickets did you handle today to did you, how many agents are you handling and how many tickets are they able to handle and then you're doing exception handling or you're pro. the agents to do something different or maybe now you're thinking about how you will avoid tickets entirely and so I think you're just gonna see folks kind of continue to climb up the curve and maybe start to do more of the work that their managers did, for example, yeah, yeah, can I add to that. So, uh, speaking of like what will humans be doing in the future, I think there are 3 things. So, uh, first it's executive function. And then there's common sense, the smell test, and then finally it's long horizon projects. So in terms of the executive function, you can curate, let's say, you know, a robust prompt, a few 1000 words. You need to decide what goes into it, what are the relevant contexts that an AI, a large language model needs to know to execute this job competently. It actually takes someone with a reasonable amount of seniority experience to know what's relevan. And then what's not relevant, right? And then at the same time, uh, no matter how beautifully you write your prom, or however advanced the models will get, like there is a bit of common sense check at the end of the generation. Someone needs to look at it and just say, hmm, it feels a little bit weird, right? And then you tell the AI to give you 10 more options. And then, so, and the final thing that models cannot do is long horizon planning. So right now, the cutting edge. Models can reason, can work effectively for about 20 minutes, right? So for any task that is beyond that duration, you still need humans who can think in terms of quarters and months and years in order to make those decisions. So to, to wrap up everything, I would say high impact, high consequence tasks will be within human's purview, whereas everything else, let's, you know, let an AI take care of it. Now both of your companies have been really successful in uncovering these hidden bottlenecks, time wasted, but you're you're solving for that. Where is the next bottleneck that you're going to see? What is the next problem that both of you want to solve? And Joseph, let's start with you, OK. Actually, there, there are lots of bottlenecks that will never go away, right? So AI is going to remove digital bottlenecks, information processing, uh, just text transformation related bottlenecks, but anything physical, AI can't do that, um, knowledge diffusion, approval chains, uh, resources. Allocation, your teams are too busy. There's a skills gap, just general knowledge management within the company. All of those will remain bottlenecks in the foreseeable future unless you know something very exciting happens. And then we would need to reassess everything. Yeah, so you don't see those being solved by AI right now. Would they be in the future, do you think? We will require several profound AI breakthroughs in order to solve for those. So, uh, physicality, shipping, I mean, uh, yes, they are robots, yes, they are self-driving cars, but at the same at the end of the day, it takes time to move things from A to B, right? And then people they only. have so much time in a day. If you want to communicate a message across your company, right, you need to think of really useful, really good tricks to get everybody to pay attention. And those are more human and physical real world problems that AI can help, but it's not going to solve. Yeah, Jennifer, what about you? Um, I, I think that on the knowledge diffusion piece, like one of the one of the biggest challenges with AI is that it can't improve or act on what it doesn't know or can't see. So if you think about a lot of work, talking about knowledge work in a company, a lot of that's actually invisible. And what I mean by that is it's not documented or written anywhere in a way that an LLM can search through it and synthesize and apply, right? It's knowledge that lives in people's heads as institutional know-how and so this is the problem that we're really focused on, which is how do you take that knowledge out and then put it in a format that is readable by both. Human and LLM because now once LLMs actually understand the context of how work is done at this particular team in this particular company, what are the real work flows like what are people doing? What are the lifeblood of this it can then help those people be better at how they work and in the absence of having that data and that understanding. Yeah, it's kind of like shooting blind, you know, it's almost like having a reference librarian who can go get you anything, but the bookshelves are empty. There's nothing to go get and so we're trying to fill all of those bookshelves with all of the relevant knowledge from the company and then make it so that you have that librarian who not only will get the book but then will come over to you and say, I actually don't think this was the right book like you should. Reading this other book instead or or the way that you were working right now actually you could be automating some portion of that and you should be doing this instead and so I think one of the things that'll be really interesting to see AI start to do is a little meta help people use AI um because the pace of change, uh, within AI I thinks faster than anything we've ever seen before. And I think it's gonna outpace humans and particularly organizations' ability to rock change and so we think a lot about how do we actually use AI to prompt people as they're working for ways that they could be better in using the tools that they have AI and non-AI based tools and so how do we use it to understand the context of what you're doing. Take the best of what you know how to do and then help you actually improve upon that and be better than you could do that before and then kind of guide you through as you're working what that means for you and how you should change your work I think that's where you unlock kind of the next big step change and that and that's the problem that really excites us and that we focused on my quick comment is. There's so much truth in what Jennifer has said. In fact, whenever we try to automate a new workflow, the first step is always to get the whole team sit together, and then we have like 1 hour, 2 hours where we just brain dump everything that everybody knows about that process and then we transcribe that and then we, that is the source. With which it's necessary for us to turn that into workflows into diagrams, into due diligence, into everything else. So just getting people in a room, force them to speak their part is so essential. Yeah. Oh, I'm sure that makes humans feel very much more essential. Joseph, thank you so much, Jennifer, thank you so much for joining us. We really appreciate it. Thank you. Thank you.