How a 1,000-Person Company Actually Uses n8n with Sahar Rahmani

🎁 Link to Enterprise Playbook 👉 https://go.n8n.io/n8n-enterprise-playbook

Most companies trying to roll out AI get stuck at five or ten workflows. Fullscript, a health tech company with over a thousand employees, built more than 1,500 in under a year (with over a hundred in production) and saved tens of thousands of hours doing it.

Sahar Rahmani is the Director of Internal AI at Fullscript. She walks through exactly how they went from engineering-bottlenecked automation requests to company-wide self-service AI adoption using n8n.

The conversation covers their hackathon launch strategy, the template library and global credentials trick that removed barriers for non-technical teams, and the highest-value workflows they built across sales, finance, legal, and customer support. If you're looking for a practical enterprise AI automation playbook, this is it.

Sahar also shares how they shifted their messaging from "AI first" to "people first, AI powered," and why that cultural reframe was the key to getting real buy-in across every department.

🎙 Chapters:
00:00 - Intro and Fullscript's AI results
02:12 - The state of automation before n8n
03:56 - Workshops, templates, and credentials
06:33 - Why the templates actually worked
09:44 - The company-wide AI hackathon
13:31 - Hackathon advice: start small
20:05 - Sales lead scoring automation
24:37 - ROI: tens of thousands of hours saved
27:20 - Finance invoice reconciliation
32:48 - The Director of Internal AI role
33:48 - Overcoming employee resistance to AI
38:00 - Customer support call automation
42:22 - Legal contract review with local AI
44:56 - Keeping momentum after launch
48:42 - Final advice and where to find Sahar

Get 30% Off n8n Cloud Starter or Pro Plans!
Want to get started with n8n? Visit n8n.io/pricing and use code 2025-N8N-PODCAST-729C416E at checkout for 30% off your first month or year.

Transcript

SPEAKER_01 0:00

Today we are learning how to empower every employee in your company to be able to build more than 1500 workflows in less than a year. We have tens of thousands of our sales. We have a workflow that we build for compliance. That alone saved like 3,000 per year. We give access to everybody, didn't ask questions, limit anybody, everybody could use it. Last year we were AI first and people were like, but this doesn't align with our culture. We change it to people first AI powered. My role, I am a director of internal AI at Furasque, and my main goal is to make sure internal teams, everybody like including legal, customer support, marketing style, know how to use AI, use AI or bring solution for them.

SPEAKER_02 0:48

Most companies trying to roll out AI at their org usually get stuck at five, maybe ten workflows. The engineering team is a bottleneck, and the non-technical teams don't know where to start, and nothing makes it to production. FullScript is a health tech company with over a thousand employees and compliant constraints. They handle PHI and PII information, so not every AI tool is safe to use. They have built over 1,500 workflows, with over a hundred of them making it in to production. Tens of thousands of hours saved. Just one compliance workflow has saved them over 3,000 hours this year alone. Sahar Romani runs this. Her title is the Director of Internal AI. And most companies don't even have this position. Today she is walking us through how they did it. The hackathon launch, the template library, the global credentials trick that removed the biggest barriers for non-technical teams that supports every builder. And the cultural shift from AI first to being people first, AI powered. If you want to save tens of thousands of hours like Sahar did, we put together a free playbook with a step-by-step guide on how you can bring this into your company. Plus an ROI calculator so that you can run the numbers for your own team. The link is in the description. This is the Innate In Enterprise Playbook. Let's get into it. Hey Sahar, so what are we learning today?

SPEAKER_01 2:14

Today we are learning how to empower every employee in your company to be able to build more than 1,500 workflows in less than a year.

SPEAKER_02 2:24

Incredible. That is a lot of integrations. Most people never make it past the first five or 10. So I would love to figure out how you've been able to do this in less than a year. So can you walk me back to the beginning? Um, what was the state of automation at Full Strip before you got involved? And like what were the problems that you were like trying to figure out to solve?

SPEAKER_01 2:46

So our state of automation before NA10 was mainly like the teams ask engineers to do automation for them. The engineer recue them and then prioritize them based on their capabilities. Like some of those will see the life, some of them wouldn't because, well, like we have a limited amount of engineers. And that sparks with AI raising, this sparks to bring the tool that help us give access to everybody and help them to build themselves. And that's why we chose NITE.

SPEAKER_02 3:19

Amazing. And the thing about it, I think a lot of people might get a bit of a technical overwhelm uh when it comes to trying to get up and running uh within it and using AI in general. I think a lot of people can can chat Chat GPT things really quickly or maybe cloud code some things, but actually to have, you know, 1,500 workflows built and I know well over 100 in production is incredible, which is you've been able to achieve. So talk me through how do you actually implement and educate and democratize this information to this many people? Like if other companies are listening to this and they and they want to empower all their employees with this type of technology, how did you do it?

SPEAKER_01 3:57

So we make it the job of somebody to help people, right? That's uh we have a group of team people that help the team to through this journey, right? We understand, as you said, like some people get overwhelmed, it's not easy to jump in, like there are so many technical things to still, it's not very, very uh zero technical like chat GPT. Um first of all, we give access to everybody in once it's like we didn't ask questions, we didn't like limit anybody, everybody could use it. Um, we had lots of workshops. We put the like how to open anything from the like first how to build uh workflows. We create our own templates that works for our company, which was a success. Like we have lots of template workflows that they can just copy and paste it in their own um environment and use those templates. We also provide lots of global um credentials to help with the credential onboarding. I noticed lots of non-technical people get overwhelmed on the credential setup. So we was like, oh, just click on this, and it's like you have access to Gemini. Click on it, you have access to Chat GPT, or you have to access to a Slack. That helped a lot for team uh to take it open in it and then start exploring, right? And that's a main thing you need.

SPEAKER_02 5:29

Start getting started. I love it. And this is one of the biggest things that I think is important to know for anybody looking to integrate AI and automations beyond just the typical chat inside a window and then get an output is how do we get people up and started? And you talked about two criteria: one, customized templates based on use cases, and two, credential access. And so you can say, here's a simple workflow for getting started, and this is how you connect everything up and without you know needing to necessarily go into the back end of some system to find an API key or off-bear credentials. What I'm curious about this is when you're when you said the templates were success, what about this made them success? Did you divide them up in departments? Did you divide them amongst use cases? Like, like what about this actually got people to use them and start and get started to adopt them and integrate them into the company?

SPEAKER_01 6:21

We started with something very simple for template, especially, something that lots of people need or ask us. How can I automate this information and send a Slack update? Right? I want to inform a team with a Slack update. So we had a template that you gather data from X solution, and then this is how you send a Slack update. Or how can I um, for example, summarize the project's update and send it to a specific channel so everybody knows what's it? So we create that, we we create uh credential and how to use our uh project management tools, how to summarize it, and then how to send to a Slack message. And then very in a very template, it's very with sticky notes. We clearly said you just need to just give a Slack ID here, update based on update AI here based on your criteria, and then um link to your project. And then not only that in NATO, we also in our internal documentation we create lots of how-to's, like for each template, there is a how-to's that take your screenshots and goes a step by step what you do, even like control, like do control C, control A, like even as a step by step as possible, so the team can feel comfortable to use it.

SPEAKER_02 7:53

And with these educational pieces, are they marked as sticky notes inside of N8N? Or where is this? Is it parallel? Do you have some sort of knowledge base like a Jira or something else where they're living?

SPEAKER_01 8:04

Yeah, like we we have a group, like we have lots of group cards, like for each template. We have a sticky note. If you want detail how to use it, go to this group card. And we have both the sticky notes and group cards that complement each other.

SPEAKER_02 8:20

I'm sorry, are you saying uh group cards? What are you saying?

SPEAKER_01 8:24

Group is a vendor name, like it's okay, okay, okay.

SPEAKER_02 8:26

So it's uh internal documentation, yeah. Got it. Okay, I was not aware of that one. Thank you for explaining. Great, fantastic. All right, so then you started breaking these things up inside of these different categories and say, what are people asking for the most? Well, everyone talks on Slack. So how do I make my life easier that I need to basically data aggregate and then put these things on Slack? Okay, well, great. Now that I have that together, I got to figure out credentials. Uh now I have the templates because I might need credentials from WhatsApp, you might need it from Gmail or Microsoft, whatever it might be. And then now people will have the ability to have that same, I would say, pathway to data. And a lot of these people are having the same pathways to data. I'm I'm trying to bring data from a source to one central form of truth, which is Slack sometimes. Slack is generally a good communication channel, terrible for a knowledge base. So it's uh it is a good use case to say, I'm gonna bring the information here, but maybe it's also a better use case of then ship it uh somewhere else to be stored. When you're coming up with the patterns that you're seeing, like with Slack, everybody needs to bring something into Slack. How did you surface that insight? Like, did you do surveys? Did you whiteboard things up? Did you did you sneak into people's computers when they weren't looking and look at and track them? What did you do?

SPEAKER_01 9:43

Two things happen. First of all, shortly after our N8N big launch, it was coincident. During that hackathon that everybody was, well, this is new tool, and it's starting exploring as a solution. We noticed that lots of conversations like I want to send a slide, I want to send a slide. So it's kind of survey-ish, but um not the official survey because like lots of solutions ended up. We want to send a Slack message. The other thing is we have a dedicated help channel that people can ask the AI-related questions, which lots of time in our case become N8N questions. So I regularly go and see what type of questions surface a lot, and then um I write them. Like, for example, another solution. We have like I build a template about it. We have our AI recording tool that records uh like our uh meetings and transcribe it. And I notice lots of people want to take that transcriptions and put like convert it to a document or a slide message, okay. So uh we um another template is that how to gather information from that uh AI trans meeting transcribe and convert it to different types of documents, like in Google Doc.

SPEAKER_02 11:19

With the hackathon, was this online or was this in person?

SPEAKER_01 11:23

Um both.

SPEAKER_02 11:23

Like we okay, so you did both. Okay, and what's amazing is so you had this big, more or less uh serendipity timing for a giant group onboarding. All right, we're gonna have a company hackathon, aka you now need to tinker. We put some dedicated time, and that's one of the biggest challenges. And the things like I've uh in the previous life, I used to run many, many hackathons on high technologies for major universities and brands. And by carving out a dedicated period of time, 28 to 48 hours, you now get everyone to say, okay, look, it is your job to now tinker and try and do. And if you ever tell people to do it in their downtime, a lot less likely. And so having this in place, you all of a sudden basically said, our intention is AI and automation. And that's when you hit what I love about this, this basically entire company feedback system of saying, well, we've got 436 requests for Slack. Uh, we've gotten, you know, 200 requests about sales agents, we got uh uh 400 around recording messages. Cool. And so you're starting to see these patterns of behavior coming in, whether they're they're coming in via an agent that they're chatting with or they're just you know DMing um all of the engineers on the team. Hey bro, can you come help me? You know, yeah, yeah.

SPEAKER_01 12:40

Um we have a actually during the hackathon in our office, there were a room, specifically like we sat, like a specific group of team, like we call it uh GST team, get shit stuff done. And then that team sat, and their job was to help people to build NATO and workflows. And um, obviously, I was new at that time for a company and with Nathan, and most of us were new using NATO, so we all learn and upscale together on how to uh use NATO.

SPEAKER_02 13:14

Rising AI tide raises all ships. So, with that, if you were to go back in time and give advice to your younger self on how to how to get the most out of doing a company-wide NA in AI automation hackathon, what advice would you give?

SPEAKER_01 13:31

Prepare, like that was when we learned pre-pair uh templates and credential ahead of time, because the biggest thing we hit, and again, it was new for us, was like how to build credential for this tool, how to to get credential for that tool, and like you were like, oh, we don't know. Um like it's been a year and the whole NHL is much, much more powerful system, AI is a much more powerful system. Um, something I do now is lean way more on a chat in NATL or like other AI coding tools where I can to help me start building uh workflows. So I wouldn't I ask everybody to not come with the blank page or blank workflow. You build something using AI or like chat in the NHM, and then we continue from there.

SPEAKER_02 14:28

So it sounds to me like one of the biggest things, again, was how you got them started in the first place is the best thing you can do to prep them up for the hackathon is one, how do we identify what templates are the most useful for us to build? And then two, what credentials do we need to get ready and put in place so this whole thing doesn't come crashing down around us uh in a small window of time? Uh so what makes me think is almost if you had a series of questions, and you can anybody can go do this and you tell me this would make sense. Uh go to your favorite AI LLM of choice, pick and choose, and say, I want you to generate me a list of questions uh to help uncover the insights for the most useful AI and automation workflows for my role and my position, and then list them out. Then send them off to the team members, have everybody fill all that out in their favorite LLM of choice, and then send back those results. Then take all that information, feed that into some AI and go, okay, based on the uh sub a thousand employees, uh, here's what we're seeing as common problems. Uh, what are the credentials you're gonna need is X, Y, and Z. And then we can roll out some sort of system that hits the ground running based upon the AI's insights that we're then using AI to surface those to then turn those into templates, workflows, and credential pathways. Does that sound like uh a way to kind of expedite what you're talking about?

SPEAKER_01 15:42

Yeah, that's way better than what I was thinking. But yes, like I was thinking like you go AI and tell them give me an A JSON file. And then like they come back sure.

SPEAKER_02 15:54

Well, it'd be well, I guess that's yeah, that's another plan. I didn't think about that way. I was thinking about you coming with bringing them templates, but you're saying no, you bring your own templates. So essentially we have to have okay, so then we could use uh I know we just released, I mean it's brand new out the gate, uh, but we just have the new MCP feature that just came out right now. You can both create, uh not just only view and execute, but both create and edit workflows. But then I know we're updating our AI agent on the inside. So um, whether you use our internal AI agent or you use something like Claude that can generate these JSON files, but you can but then the biggest thing is great. Come in with your own templates ready to go. Maybe give us a heads up on credentials you might need so that we can get those in place so that we because part of being in the enterprise is that you can have you know global credentials, you can have these global variables in place that would allow people to be able to onboard because maybe the employee doesn't have access to those credentials, and maybe you you don't want 836 perplexity accounts. You know, maybe you just want one. And so it might make it a little bit easier if you're seeing that these are the same problems coming across. Uh, so I love this this back and forth because if people were to start this out the gate, I guess, you know, knowing this, and if you were to do this process again, because you've you've already been able to build 1500 workflows in less than a year. What do you think the the shortening of the cycle, if you were to do this process again and integrate these lessons, how much quicker do you think it would be for you guys to get up and running from time to value, from you know, just getting started with here's NADN starting to actually producing value in the company? Because I do want to segue into the value that you're getting with these automations and applications.

SPEAKER_01 17:30

I would say again, time to value would be lean on AI, right? Especially now, lean on AI, build your first version, build your um NVP, and then come to like build a smart, a small MVP, and then if you want to make it bigger, come to us, come to engineers, or like we help you to like shape the problem. Because um one thing I noticed, especially in non-technical folks, technical folks are very good on hey, this is the small solution, and then I built on top of that to make a full solution. Non-technical folks need the handhelding on that, they have a great idea, but they don't know how to build it in a small bricks, right? So that's make them like a little bit paralyzed when they want to build an amazing solution, right? So I always ask people, lean on AI, build the smallest version, and then I help you, like we help you to build, make it bigger and like make it to your food vision.

SPEAKER_02 18:34

All beginners, man. No matter what industry, no matter what space, uh, no matter what you you do, you always beginners always bite off more than they can chew. And uh when I I was originally before this industry, I was in high-tech virtual reality. And the first application I ever made was inside Unity for people that know it, don't know. It's totally fine. It's a game engine. I want to make this whole beautiful world. So I started planting thousands and thousands of trees inside of this application. And anybody that knows anything about building virtual reality applications is well, if you put a thousands and thousands of objects in an environment and you hit run, the whole thing crashes. And I was like, oh crap. And they're like, yes, you got to start small. Start with squares and blocks and circles, and then you know, as little as possible and build up from there. But you only get that by going through the lessons. And engineers know if you if you if you try to, you know, recreate uh Facebook, Slack, and eBay all in one application that somebody has the bright idea of combining, you might be a little hard to get started. So I love the idea of that. Also, I love the idea of what's making me think is going back to this AI agent that if you could give it to them and say, hey, your job is to help build this JSON workflow for these people, but your job is also to gut check them to say, make this really simple. Like, don't let them, you know, you you get you get five nodes to work with.

SPEAKER_00 19:51

Exactly.

SPEAKER_02 19:52

What can you do to make with five, man? That's it. It's your budget. We're cutting you off.

SPEAKER_01 19:57

We don't cut people off, but we help them to like, yeah, start with five and then see what they can do.

SPEAKER_02 20:03

Yeah, I love it. I love it. So talk to me about value and talk to me about some of the things like this. Uh, we talked about what are some of the most impactful uh AI automations that you've built with N8N in any in any sector or division of the company.

SPEAKER_00 20:17

Oh, so many. It's very hard to choose. There are so many. Let me switch it.

SPEAKER_02 20:21

Let me let me go with sales. People want to know more about sales. So talk to me about the most impactful sales automations that you've created.

SPEAKER_01 20:28

So for sales, like we have that's the most recent one. I'm I'm really fond of it. When our sales team and customer success team want to go to new place, new conference, new event, they have a list of participants. And then it took them a long time to just uh cross-reference that list of participants with our current customer base, with the people like we already have a communication with them. They are not part of our customer base, and then brand new people, right? It takes them like each each list would take like almost a week of time. Or so what we did, we build the automations using um that takes that list and goes to our uh different vendors, different solutions we have, and identifies hey, these are the practitioners or people or our active customers. These ones are we had we send them emails, we send them like advertisement, we send them commercial materials, but they never convert or like they are not part of it. And these ones are brand new, you never talk to them, and it helps them to prioritize, bring their focus, like uh prioritize and optimize their um sale or customer success uh solutions. And well, from that to like the work that takes like maybe a week, now it's just uh run their workflow. Like they uh they upload the list and that's it.

SPEAKER_02 21:56

Awesome. So this is more of like a lead scoring and ranking. Automation that lets you know which one of these people should I absolutely dial in the next five seconds and which ones can we just wait a bit to get to? And so it's being able to go through, parse through, and find out, you know, what are the ratings of these, of these, which is real value. And especially if instead of this taking a week for a sales agent, this is something that can be done in seconds, that response time goes dramatically up, which I think is incredible. Then when you're building these things out, you're talking about starting small, like this application. Uh, I don't know if this started small or or if it grew into this. Like, how do you expand it? Like, because if you have this many automations working together, like what is the what is the process for building upon it? Is it going to the engineer team? Is it like, you know, screaming at your keyboard in the night? What happens?

SPEAKER_01 22:46

It it really depends on uh technicality of the folks and then uh complexity of the solutions. Uh, it could be both ways. Like sometimes we build, like they build a small version, like they give the idea and say, okay, go try. I always tell them, go try that. I never build from scratch. I said, go try that to see how it works. And then, like, for example, they say, well, go try with this specific like marketing and tool only and see how it works. And then we add, we help you to add the sale tool data and like other thing data. Uh, but the other thing is like if it's super complicated and we built it and like it ended up a big workflow, doesn't matter. We transfer back, like we work with them to help them how to run it, how this workflow looks like. We educate them like how you can update it now, and then somebody from business take ownership of it and then continue building and adding materials to it. We again then we become support. So it's it's it's both ways. So they build something, we help them to scale it, we help them to clean it up, and then pass it, them pass it to them so for them to own it and continue growing.

SPEAKER_02 24:10

So it sounds like yeah, part of this is one, having someone on the biz team uh take ownership over it and iterate with it, uh, reaching out to the engineers as need be. And then there's a bit of continuous education going on. It says this is this is how we learn and evolve and grow with this. In terms of this, I I do have a question about other workflows that you built that have provided value. Um, but do you know in in total, like have any idea of how much hours saved or how much how much money you've been able to save by replacing software applications or or money earned from these things? Do you have any ROIs on the on these, you know, 1500 plus uh workflows at and over a hundred in prod uh that you can think of?

SPEAKER_01 24:52

Yeah, we have tens of thousands hours saved. Like I am, and I'm like regularly go check with people. Hey, like make sure you have an hour save on that run, like especially if publish. We have lots of brand new capabilities that we build with um NA10. And if it wasn't an A10, we have to build like a buoy vendor solution or uh like um bring a new solution. So that adds to ROI as well. But even just in terms of our saves, like combined, we have tens of thousands. We have a workflow that we build for compliance, that alone saves like 3,000 per year because it just goes and listen to talks and remove all the PHIPAIs from the calls. So that alone has like thousands, like thousands of hours.

SPEAKER_02 25:48

Got it. So these compliance applications in terms of the back end, especially because you're, I mean, you're in the healthcare space uh with Full Script as your company, and um a lot of that is, I mean, you're you're you're you're you're using real data, Salesforce data, on-call management, SOC compliance reportings, PII information. I mean, this is like real business risks. And so, like, how you how do you decide what's appropriate for like the citizen builder versus what needs to stay with engineering?

SPEAKER_01 26:16

Oh, we don't. Like, we let them like that's that's a meaning of democratizing AI and solutions, right? Um, we don't like it's usually when it's so complicated, the team comes to us for help and they will learn about it. I also have like monitoring systems that manage the complexity of the solutions. So if I see the very complex solutions, I check it out. Um, we have our controls that uh between security and us like that. How make sure like um what type of the data goes in the solutions, like make sure um everybody provides the correct access, like they don't give access to something that we are not have a BAA with it, like we have not contract with it. But um no, we don't like if if you're a non-technical people person, you can use AI and NATO solutions to build to solve one of the most complex solutions you have. That's a huge win for me.

SPEAKER_02 27:17

Like, sure, sure. So I mean it's it's true trust and democratization of the of the use of the technology, which are great, which is great. What other things if companies let me see, because you you've built a number of workflows here. If companies are looking at like, look, I want to get up and running with these things and I run another thousand-person company, right? Okay, this there's this lead rating and reviewing. Um, there's removing the PII informations. What are some other high value workflows that you've been able to produce in the company that you think are are really useful that you would tell a company, hey, maybe look at this, these one, two, or three, because they've been really beneficial to us.

SPEAKER_01 27:55

So we did lots of work with our finance team. Uh that we they gather lots of invoices from different solutions, and we have to put it in our uh current system. So we have like invoice to constellation solution. Um that's one of our successful solutions that we even pass it to the finance team to manage it, which is uh really nice things. We democratize data gathering, data like uh uh cleaning up for some of the smaller uh projects that we don't need, we don't want to have lots of dependency on data team.

SPEAKER_02 28:33

With a quick question about the finance side of things with this, um, you said that it was so successful that you passed it off to the finance team. Uh, was this and I know you had like invoice parsing and extracting. Uh, I know you had pre-processed invoices, reconcilization. What were the ones that made them worth it? Like what were the like these were really useful uh finance workflow automations that is now a central piece of what they do?

SPEAKER_01 28:59

Uhway on it, right? Like how much time it's saved like per run or per month, like how much um, yeah, like that's that's practically our decision point. So if you you want to reconcile, sorry, like your question is like what workflows are they though?

SPEAKER_02 29:21

Uh I know you're talking about time saved, but maybe I maybe I mis asked how to say it. Um maybe I misheard is like, yeah, what workflows? Is it the invoice parsing ones? Is it the is it recognizing?

SPEAKER_01 29:34

Uh in most reconciliations, right? The the rest of them is just smaller projects that ended up to that big project invoice reconciliation. That's the one that um like saves hours and hours per month um for finance statement.

SPEAKER_02 29:52

I could see that. I I don't like reconciling my own bank accounts myself, you know. I I I I go through and I okay, I'm gonna organize these expenses, I've got that. But as soon as I hit the reconcile, I'm like, yeah, I'm gonna do that tomorrow. Amazing. Okay, so those that was the big mover, right? And now, was that did that come from the finance team themselves? That like, hey, this is something that we want to automate. Did they think of it and come up with it? Or did you was that like a top-down?

SPEAKER_01 30:22

I get a little bit of both. Like we have a team of engineers works with our finance team, and then the finance like finance provides the need, like we need to solve this problem. And that team of engineers was like, okay, we can use an NA10 to solve it, and then become a more collaboration, like any other engineering solution. Like, what's your workflow look like? What are the input, output, what do you do? And especially for something so complicated as invoice reconciliation in our right, like uh there were lots of back and forth, and we built it and then we trained the person in finance to help.

SPEAKER_02 31:06

Are you so with the reconciliation? Uh part of that is are you also generating um reports or if you know, like monthly audit reports or profits and loss statements, or anything around that? Uh in order you're able to reconcile these um accounts. Um, I'm curious also if there's anything around like dashboards or things that you've made that pipe data into these things that allow them to visualize what are happening. I know uh a lot of people in the finance especially love dashboards and be able to see, you know, green good, red green bad, buy sell, buy sell. Is there anything that you have on your side that that are around the dashboard related or visualization of this data?

SPEAKER_01 31:44

Uh not on the finance part yet, but we have um like you tested it. We we are testing and we are again like some of solutions, like the finance team goes like, okay, let me see how a dashboard works. But we have lots of building, lots of dashboards, and actually we use webhook. I I personally use webhook heavily to build my own dashboards. Like I have AI dashboards, like internal AI dashboards on and it and I build my own websites. Like one of my fun projects I did was at the end of the year, I create the AI wrap. What is like AI look like on 2025 for full escrive, like how people like use it, what is our big wins, like and stuff like that. And then I use anything web purch to just serve it to the team and like share it with the team.

SPEAKER_02 32:38

Which is so it's awesome. I what might be helpful too is we haven't we haven't got there. Can you please talk about your role and job description of what you do at FullScript?

SPEAKER_01 32:48

So my role, my role, I am a director of internal AI at FullScript, and my main goal is to make sure internal teams at FullScript, like everyone, including legal customer support, marketing sales, um so on, um use AI, know how to use AI, use AI or bring solution for them. So it's partly education, partly strategy, identifying tools, and also solutioning, like in terms of um if there is a high-level solution needed, uh solution those still.

SPEAKER_02 33:30

Got it. Yeah. So you're the uh I would say internal champion of AI and automation, and going, look, we are gonna AI fy this company. And he's come in with that force of nature, which is part of the reason why you you've been driving with adoption and you seem to be very successful at your job, uh, getting all this adoption in place, which is amazing. How do you overcome? There's usually sometimes there's inherent like resistance from internal employees that are saying, uh, AI is gonna take my job. I don't want to tell you how I do my job because you're gonna replace me. I'm gonna come in. There's gonna be a uh a robot uh that's gonna be wearing my tie, and it's gonna have a picture of me and my wife, uh AI generated next to itself, it's gonna be doing work on my behalf. What do you do to overcome this general resistance that employees have to adopting this technology?

SPEAKER_01 34:18

So um overall, like one of the core values of our company and one of the missions of we have is we are people first AI powered. And this message going across the whole company. Actually, the big the best part of it is like we last yeah we were AI first, and people were like, but this doesn't align with our culture, and we are like, you're right, we changed it to people first AI powered, and um that's everybody says like AI, any solution we are bringing, anything we are bringing is augmenting to our current team. We we are people first, we want to enable our people rather than replace them. And we can see in the tools choosing we have, we can see in the education materials, we can see in um what we are doing uh across the all the solutioning that we are just empowering you using AI. Um let's keep people first, and um it helps. Like obviously, there were lots of questions, especially the first I asked, I joined people ask me, but because I'm true believer of AI cannot replace humans, and I think that shares that value shared and shows across the company, right? We can automate things, but we cannot automate human like that. Human touch cannot be automated, and we don't want to keep it.

SPEAKER_02 35:54

Yeah, I like the people first powered by automation. A lot of people get scared by this. The the here's the challenge though. The challenge is this is that everybody wants value, right? And everybody wants people to be more valuable. And if I said, hey, I'm gonna come work for your company, but I'm not gonna use the internet. I just refuse. I'm not gonna internet nothing. I would be less valuable because I'm not willing to use it. Humans are great. We use technology, we we use hammers and fire, and that's the that's what gave us the ability to really dominate this planet when really we're just little middle-of-the-food chain creatures running around that barely can clothe themselves. So having people use AI and automation definitely empowers them to be more valuable. And I'm curious about the, you know, letting them know, hey, we're people first powered by AI. Um, and is there ways that you've seen employees become more valuable because they've adopted AI? Do you have any use cases or stories around Sarah? Uh, you know, she was not doing any of this type of stuff. She started doing this thing and she was now had a way bigger impact because she could do more with AI and automation.

SPEAKER_01 37:02

Um, not in the like same wording, but I can say some examples. Like we automate um some of the like for our customer uh support team, we automate some of the closing call activities, like uh summarizing. After each call, they have to summarize, they have to like tag and like some of those capabilities. We we help them to automate that process. That alone, how about customer support team and that's a real feedback they got is like it helps us to be more focused on the call rather than being worried to take a note or just like do the like do the paperwork after that, right? And that's that's a value of AI. That's what people first mean. We want that people touch, right? But the AI does lots of other work that it helps them to be more focused, actually, right?

SPEAKER_02 38:02

So that one, if I'm understanding it, at the end of the call, it tracks it, it summarizes it, it might be organizes the call, loads it up to some knowledge base somewhere that allows you to analyze. And if you didn't have that, then the call center team members would be taking notes, would be uploading data, and they wouldn't be present because they'd be too busy documenting, not enough time being focused on the human-to-human connection. Is that accurate?

SPEAKER_01 38:27

Yes, exactly. Right. It's not like they wouldn't be present, but like because they have to take a note, because like they would be less present present than now that their whole focus is in on the call, right? They don't need to be worried about documentation. So that's one example that I love, and one feedback I love that hey, here it is. We are not replacing customer support, like we are not replacing that call agents, we make them more present using AI.

SPEAKER_02 38:58

Yeah, you're figuring out the things that are distractions because really what we one of the number one things that we have is attention. Like our greatest asset and value is time and attention, and that's what we got. And if you have to do 50 different tasks, you're not gonna be great at that one task. So, how do we get 48 off the plate? So, okay, that's great. So we've talked about call centers, we've talked about finance, we've talked about some sales. Uh, are there any other ones, maybe one more that comes to mind that you think is like any company that might be in a growth position would be interested in having, like, you know, let's say around a thousand employees. You say, okay, you might want to look in this department or this workflow because this has been really beneficial to us.

SPEAKER_01 39:39

Um, legal, another example for us that we help them again in the company, it's like us that we work with the health data. There are lots of uh contract review, contract review, and then we help our um AI, sorry, legal team to theorize the solutions, like uh help them to make sure every first task goes by AR. So again, it gives them more attention, more focus on the work they uh the complicated solutions or the complicated work rather than um reading all the SMS we are sending, like chat we are sending, messages we are sending, right? So that's another solution we are using AI for, and uh it is very successful.

SPEAKER_02 40:33

Got it. Yeah, so you're running through legal so that it can catch any of the low-hanging fruit of any like legalese contracts or oh gotcha. Because I think a majority of people don't actually read contracts, they just mostly sign on the data line and and and pray. So that's a really good use case as well for the legal side of things to be able to look at that. And that's is that done uh with that AI? Is it done with like uh a local language model, or is this one that is actually being you're using like one that's more of like a chat GPT or name another one out there? I know some people are a little hesitant to put contracts through um language models.

SPEAKER_01 41:10

No, it's a it's a local one, we control the things. And then also we actually that's one of the cases we use evaluation node heavily, to be honest, like in the ATE, uh, which is another favorite note of me for me. Um no, we use uh our road solutions and then um also like any solution we give our data to AI, like we have a contract with them, they don't read it, they don't like um any AR solutions we have. Uh we we have a contract that they don't use our data for their own chain. That's gives us a little bit um less concern, uh, you know, fairness, but uh yeah.

SPEAKER_02 41:59

Yeah, that's the thing. You can use a local language model and do very well. And the cool thing about Nitin is that you can do it on-prem, right? On on edge AI, and you can you spin up a you know, whatever you got, a Mac mini or you know, name your favorite hardware device and load it up there and be good to go. Okay, beautiful. So we talked about that. Uh again, you you you actually mentioned so we talked about that. Is there any you said favorite nodes? You said favorite nodes was an AI evaluation node. Do you have any other favorite nodes that come to mind?

SPEAKER_01 42:25

Oh, that's the webwick node is one of my favorite nodes as well. Like I use it a lot. Yeah, like all the G Suit ones are very useful. Like, I really like the G suit ones. Others, Iterable nodes are famous. I really then there's a new one. It's called iterable. Oh it's a new, yeah.

SPEAKER_02 42:47

Iterable.

SPEAKER_01 42:48

Others I use, yeah.

SPEAKER_02 42:50

I'm actually not familiar with that node. I feel like I should.

SPEAKER_01 42:53

Yeah, I randomly found actually, it's one of those things that AI agents show me. I I usually use HTTP request ones, and then like once I used AI agent to build some workflow, I was like, oh, there's a iterable node, so I didn't know that.

SPEAKER_02 43:09

Okay, fun fact, everybody. I didn't know that one either. I'm gonna go take a look at that. I know it's just we need a better visualization of the nodes that are available. Sometimes you're like, oh, that exists.

SPEAKER_00 43:18

Uh yeah, exactly.

SPEAKER_02 43:20

And there's one, yeah, there was one node that I found out about called compare data set node. I don't know if you've heard about that one, but it's incredible. I I gotta tell you about it just real quick in terms of this was it's a it's a data set node that you can put in two different data sets, and it's gonna look at it and go, okay, and outs it will automatically parse out which ones are the same and which ones are different. And so you just put it in there and it just automatically does that for you. So you don't need to like have any fancy logic or anything else. Two data sources, okay. Uh which one of these names are the same or numbers are the same, or which ones are different? And so it's a very useful node that I was shown from an NAN team member, and I was just like, uh Many times I could have used this thing in my past, and I'll have to do some like code node that's then looking for differences or you know compare then. So, anyways, yes, yes, it's very useful.

SPEAKER_00 44:10

I immediately have three like both full in my mind that I'm going to switch it.

SPEAKER_02 44:16

Amazing. Amazing. All right. So, okay, so you talked about this, we talked about the different nodes that you have, your favorites, you talked about some of your favorite workflows. Uh, in terms of if you were to map out a path, right, for uh another company that wanted to get up and start to use this stuff. We talked this about a little bit about doing a hackathon and getting templates and credentials, right? That's kind of sounds like phase one, right? How do you how do you switch this out? And we talked about this a little bit. Talking about this, what would phase two and phase three go from just not only the creation, but the implementation and the iteration. So that these things cannot just turn from cool concepts that somebody did at 5 p.m. on a Friday into living in prod.

SPEAKER_01 44:56

Yeah, well, we do a few things. Um, first of all, the education part continues, right? I try to every week or every two weeks release a new template, either people use it or not, but like just keep reminding that this is another solution. This is another way. Um I encourage everybody who builds to demo it, like lots of demos to showcase solutions. Like in our town halls, we do demos as well to just showcase hey, this can be done by NATO as well. The team built this one. It's not only by me or engineering team, with everybody who, especially if non-technical teams are there, they go like first online of the demos because we want to the team does like our security team does lots of uh workflows, like build lots of workforce with NA10, and they like we always encourage them to demo as much as they can to showcase the power of the NA10. Um we continue the workshop, like you decide with again uh the beginner to like more advanced, and then the team uh we have a like the whole record of them, so NA10 101, 102, and continue. So we keep the energy alive, like it's part of my job to make sure the hype is alive, like everybody still knows this is here, people using it. Um we report on uh success and I monitor the usage, and then like right now, like you are 77 active users per month. Um and then it's not the same people I switch, obviously, and then I keep making sure to drive engagement on that as well. Like I go to the different department, talk to them and participate in their uh on-site or events to have a focused solution for them and then onboard them and show them the capabilities and so on.

SPEAKER_02 47:04

So we need to hire someone like you in every company. That's what we need. We need a Sahara in every company, that's what we need. All right, you heard everybody. It's just gonna be multitasking all these jobs.

SPEAKER_01 47:14

That's what I said. Make a job of one person, right? Make a job of person to drive adoption, otherwise, it's always like like I don't know, it's a hundred percent of it.

SPEAKER_02 47:24

What you talk about is this. So if we have we need an internal champion, we need someone who's gonna be focusing on automation, and also because you become a force multiplier. Because now it's not that one job that you have, you're now dedicated on being able to empower all of these other people with hey, here's the education, here's here's templates to get started, here's credentials, here's 101, here's 201, here's the engineers building, inspiration of what's possible, connecting ownerships to these people, all these things that you're you're creating this ecosystem and this culture around having people to become engaged with AI and automations and to where it bleeds into the company, and also adjusting the company culture to where it's like, yes, AI is super exciting, but don't lose your values. You want to have this people first powered by AI that allows you to be able to have people take ownership over it without the fear of them jobs are gonna get replaced. So it's it's an amazing ecosystem and culture that you've created. And it's been able to save thousands of hours and tons of time, and I'm sure generated lots of revenue with this lead sourcing. Uh, we know if you respond to your lead within five minutes, you triple your close rate. So that lead rating and scoring, I'm sure, increased the value a ton of the company as well. This has been amazing to have you on the show. Is there anything else you'd like to let people know about before you tell them how to get a hold of you?

SPEAKER_01 48:42

The other thing I want to say, like, it's not one person, like I champion, but every department, every person has to step up, and I'm really happy that the team accepted it. People like we have champions from different departments that learn and take ownership and like bring the business mindset, and we have to encourage and help them as well. Also, I'm really thankful to the group of engineering again, Team GSD, that helped to support the team as well. Like we provide support. If you have a question, no matter how big advanced or how um beginner you are, you feel welcome. We respond to you, we jump on the car, we help you. And that helps a lot of adoption as well. So invest on people, let them do their selves. They are smart, everybody wants to learn new things, everybody wants to grow. And if you let them, they do.

SPEAKER_02 49:42

Inspiration, education, support. Fantastic. Uh, and if people want to find out more about you and your company, what do they do?

SPEAKER_01 49:49

Uh, for me, like you can go to LinkedIn, you can find my LinkedIn very easily, Sahar Amani. Uh, and then please go to fullersweep.com. There are lots of educational material about the Hell's Health Take Supplement Labs, Wellness, Whole Care in the um Fuller Sweep.com. And yeah.

SPEAKER_02 50:10

Beautiful. Sahara, thank you for the time. It's been an honor and a pleasure. Much love, and I'll see you on the other side.

SPEAKER_01 50:16

See you. Have a good one. Bye.

SPEAKER_02 50:18

All right, all right. Bye now.