Inside Renada's own AI stack: autoscheduling, call notes and workflow audits

Hello and welcome back to this AI series we're doing. This video today is a little bit nerve-wracking for me, honestly. I'm ripping back the curtain, if you will, and showing you some of the things that we use AI for in Renada and giving all our competitors the lovely advantage of copying us. Although, as my kids would say, sharing is caring. And as Mendy would say, a rising tide raises all ships. So I think on brand we are giving away all of our knowledge. Not going to tell you how to do all of these things because some of them are crazy complicated, but I'm just going to sort of plant some seeds really and show you in our production build what it is we do on a day-to-day basis to run Renada.

Before we continue, need me to shout out Dylan for this video because he's the one who's got to go and sanitise any client data from this entire recording as I whip through it showing you what we do in Halo for our clients. So let's jump into it. The first thing I'm going to talk about today is our autoscheduler. So how do we get tasks from our client into our diaries? Let's go. So here we are. Welcome to our Halo production environment. I don't think there is any client data here. I'm just running through it. That's the team who do the task. That's this. Okay. I think we're good. We're probably not this. I already feel sick fiddling, but here we are. So what you're looking at here is a task.

Now, I'm not going to talk you through how a task goes from an idea to actually being scoped and generated. But this board you're seeing here or this ticket dashboard is kind of setting the scene. So what we basically do at Renada is whenever we get a task in, we scope it. Okay? And when we scope it, we give it a time budget. So how much time do we think it's going to take to deliver the thing that you want? Okay? An hour, two hours, four hours, a month, you know, whatever. Then we set a due date. So the due date currently is when we can deliver that, unless a client gives us a priority. This is a really critical thing. Can you get it done? But typically we don't work like that. It's we will get it done as soon as we can. And then available hours. This client has loads of hours actually which is probably why I'm blasting through some tasks now. Left hand side you see available hours. This is very important. This plays into this automation to be clear.

So what we do at Renada is once a task is scoped and it is approved by the client, we then schedule it into our diaries to do it. Right? Fairly straightforward. And we leverage the data here or some of the data here to make those decisions. So what does it actually look like? Well, what we do, you'll see here, Renada. And you'll see this information. So it says scheduling is successful. Appointment created today. So this is the 11th. I'm doing this video instead of doing the task. I'll get it done. This is when we're going to do it. So in my calendar right now will be this appointment to do. I'm doing this instead because I'm my own worst enemy. Who's going to do it? The task for this is just the title from the task and then scheduling details.

So we've got a big AI prompt really that says well, a lot of constraints really. Don't book more than two hours at a time if we can help it. Again, we've got two hours and fifteen minutes here, but that's allowing for buffer. So don't allow more than two hours if we can. Don't allow more than like three a day. Don't have back-to-back. Allow a break. Nine to five, et cetera et cetera et cetera. So we've got a big prompt of AI which is basically saying here's information from the task and go and schedule it. And you'll see down here below we have the scheduling details. So selected this time slot as it was the earliest available slot that met all the constraints. So again it's doing a constraint validation. Meets a two-hour lead time. So again, whenever we have a task approved, we don't schedule it straight away because I could have, you know, mentally earmarked the next hour or two and it's really annoying when a task is approved and then it's in your diary. So we give a two-hour window.

Fits within the working hours. So I work nine to five. It validates the agent's working hours and their calendar and their availability. Respects capacity constraints. So two hours forty-five minutes within a four and a half hour consultation limit. I restrict four and a half hours of consulting work a day. So actually business work a day. I know, best boss ever. But the idea is that four and a half hours of actual work is way more than what actually we target here. So I think that works quite nicely. Schedule before the due date. So you'll see down here this is the due date, the absolute latest date we can deliver this by. Again we always try and deliver this first. So you'll see here this is when's the due date of this? On the 21st. That's ten days from now but I'm doing it today. Again it's just saying we can get it done by a date as opposed to I will do it on that date.

I think the motion, marketing, which kind of annoys me, but I get it. And it says change due dates. No, change due dates into due dates. And I'm like, no, you want to do it before the due date anyways. It's the whole point. And then appointment ends at twelve, leaving the day available for work. Cool. And then it's just a slight reminder. So this is basically saying, um, for us, rather than me writing a crazy algorithm to or maybe JavaScript script snippet really, um, it's where we started, by the way. Rather than me trying to write all this logic in and do all the validation, I go actually AI is pretty good at the reasoning for this and it doesn't cost us a lot of money. I think it cost us about like twenty pence or something or fifteen pence to schedule this in. It's quite a big prompt and quite a lot of data got to go through. Which for me the time saving is great.

Now there's a few other things that come with this just for reference. We also have up here like this do later button. So in this case for now I mean I've got the rest of the day free for some miracle. But if I didn't want to do this now I could click this appointment and if there's multiple when it chunks it there will be multiple appointments. I can click it and then I can click duration to push. So I can say an hour, two to three hours, four to five hours, or six plus hours. Again this is just saying I don't want to do it now. I don't want to do it today. Maybe push it four hours or six hours or whatever. We don't use this a bunch to be honest. It was kind of more work in progress this really. We started with the days and then that just got a bit too heavy. So this is just hours. But the idea is we send the data back to Claude, to the AI we use, and say here's the appointment, here's a time and away you go.

So yeah, AI for scheduling our tasks. We actually leverage n8n to run this automation purely because it can take quite a long time to actually go through the entire prompt. Something you need to be aware of in Halo is if the AI takes longer than I think it's one hundred seconds to process, so one minute forty, good maths, then the AI or the runbook in Halo will time out and fail. This can take on average two minutes because some of our tasks are maybe six to eight hours. So we're having like three chunks. It's got to make all the appointments. It is generally quite long and heavy. So we've actually opted to do this within n8n for this runbook or this automation. But yeah, that's this one. This is the autoscheduler. Hope you enjoy it.

Number two, what are we talking about now? Let's talk about the biggest game changer we've ever had at Renada. So I'm just going to get the scene set up here. So please do bear with me. Called transcription. So this is the biggest game changer that we've ever had at Renada. And the amount of time it saved for us is just unbelievable really. This one's going to be hard to show because this is actually a live recap I've done because I wanted to actually show you the true value it gives us.

So one of the things we do is that we actually use Teams premium at Renada. And that's because personally for me, these systems like Otter and Fathom and stuff like that, they hold all the recordings for our conversations. And the problem I have with that is we chat about really in-depth business details of a company. We go through their finances. We'll often chat about staffing issues. We'll chat about many things really that I don't really want to be held by another company. I don't know what they're going to do with that recording. And for me it's always just felt a little bit annoying when a bot attends the call and it's doing stuff you don't know and it's sending you data god knows where. And I'm just like, so we decided because of me, things that we're going to build it ourselves, baby. And actually that was the best decision we've ever done.

Now it took an inordinate amount of time. I'm not going to lie. Again, we use n8n for this because an hour and a half or an hour of me chatting on is a lot for AI to process, but it does some really good things. So it removes some personally identifiable information. It removes any filler crap, right? We might talk about how your holiday was for five minutes because, you know, we're humans. It removes all of that. And it's really tailored around our tone and it's clear and it's concise.

And let me show you. So I had a session the other day with a new client. This is actually an onboarding, a little bit of an oddity for Renada or a rarity if you will. And it was a scoping session, right? And I was basically going through and you know, we're not, we don't hide the fact we use AI to write these notes, right? The question is do you want to pay me half an hour to write up the notes or do you want it for free? And AI will do a good job of it. We all know what's going to be selected here. So the way we've done it is we send the Teams transcription to Claude AI. Claude processes it and then it writes back all of the data to the ticket into custom fields.

So here what you'll see is what did we do is the first section. Introduce your consultation workflow, include an automation session recording or automatic session recording. There you go. AI generated transcripts. I haven't made this up by the way for this video. Did an overview of our portal. So how our clients use our portal, blah blah blah blah blah. I'm not going to scroll down any further, but this is basically chunked into sections of what we did. It's a bit of reading, but not crazy amount of reading. Then we say what's planned for next session. So before I end any call, we say what we're going to do on the next call so we know who needs to attend, right? So we're going to configure the QuickBooks Online integration. We're going to set up the Pax8 integration. We're going to begin billing configuration. And we're going to review and approve service desk configuration because we're building that out for them. A bit of a unique onboarding this.

Additional notes down here. Dylan, you will have to sanitise a little bit down there below. I'm just going to jump cut back in the recording. Look at this. It's like magic. Then you can see, um, client homework. So this is just a list, bullet point list of all the things that we've told the client they need to do. We then provide links in here if we need to. So prepare the customer list, decide whether the product list, et cetera, et cetera, et cetera. And then there's also Renada homework. We also then attach down here the recording link. And this is automatically populated from our automation.

And here's the genius bit. Because we've said there's Renada homework here, I have actually slightly changed this wording just for this video. But what it does is it then generates all of these custom fields down below. Now, what we currently do is we don't want AI to decide if it's a task request or an internal task or a support or a road map item. The lines are very much blurred with stuff like this. So what we just say is show all the tasks. So we said we're going to build the service desk out. We're going to configure the dashboard. I actually decided I don't need to make a task for the workflows or all the ticket types. So we can just ignore these. But the idea is they're really quick ways to make the tickets.

Now, what you're probably sat there thinking is that's great, but what's it actually going to do? Is it just going to give you a ticket title? Fear not. This is where we do a little bit more post-processing of this data. So one of the things that we discovered at Renada is recording a call is great. You get all of the notes, but reality is nobody's going to sit and read all the notes. I'll show you in a minute how they look. By the way, one of the things we often say as consultants is we'll go and do that or we'll follow up with that. One of the biggest complaints we actually get from any new prospect is I worked with X company and they never did anything after the call they said they were going to do or they never followed up. It is a problem we faced as well. It's not unique, right? You've got one hour in, it's high octane, it's high intensity, you're flying through stuff, and then you're like, "Oh my god, what did I say I was going to do?"

And back in the day when we wrote notes with our hand on our keyboards, we sometimes forgot, right? So what we actually do is we say, right, make everything here a task. Whether we do it or not is irrelevant, right? This holds us accountable to the data. And what we actually do, cue Dylan here, making some magic, is actually write to this tasks tab, all of the tasks that we've potentially going to do. Now, again, Dylan's going to have to sanitise this. We're just going to take this top one, so we'll blur anything beneath, Dylan. This still needs some love. It's not perfect, right? This is AI and it's definitely an iterative process. But let's take one of those tasks. So I said on the call that we would go and build the service desk configuration with the ticket types. Okay. So one of the things we always ask internally before we do any task is what problem are we going to solve? Because if there's no problem to solve then why are we doing the task?

So what we have here is a slight bit of blur about the problem and we typically rewrite these a little bit but this problem will be taken from the conversation if that makes sense. So client is merging two organisations. Dylan, please merge these, blur these names. Needs a clean production ready service desk for January go live. The compressed timeline requiring to take ownership of building the core configuration rather than extended training session as the client needs to start delivering services through the platform immediately in the new year. So this is the unique constraint with this one. The client has to go live quick. I didn't really want to take on the project but the client knows this by the way is back. But I knew we could deliver it if we change the scope a little bit which is we're building a lot of the core functionality out.

Then it's the task overview. So what did I describe on the call? Build out the services configuration on the client's Halo instance. Triage incident, request, alert, project, task. I did say those things. Technical details. So I spoke a little bit about what it means. Restrictions. So do not add custom ticket types beyond the call system without client's approval due to compressed timelines. Focus on MVP product. Again, this is really clear context from the call. I was saying to the client, whilst it's important we're building what you need, meeting in the middle, remember we've got a six week window. So if you start adding in all these bespoke ticket types, we have to really tailor it back for MVP. And again, the reality is I'm not always the one building all this out in Halo.

So what we're doing here and what we're saying at Renada is set up your colleague, give them as much information as we can, and it's actually me who will scope this task in. And then we have critical information. So time budget was not confirmed. It's high priority. And the due date is before the next session on Wednesday. Not one hundred per cent true this. And we do actually have a bug in this one here. It says readiness status is ready. This is something actually we're playing with at the minute. What we're trying to actually get to, and it's challenging, but we're trying to get a readiness status. So when a task comes in, and I'll show you this in a minute, when a task comes in to be scoped, it will go, is this ready to be scoped or not? Meaning any of our colleagues can go yes or no, and it saves any questions.

And there's a bit of fun we're playing with at the minute. So the consultant commitment quote. Sorry, I just moved that. Dylan just go and sanitise again. Connor says, "What we'll, we'll probably frontload this by the way, after this call we can just get all the service scoped and built. I wouldn't be surprised if Elaine has it done by the end of the week to be honest. I know Elaine's doing it. She's on the call. So the next two weeks will be from a service side of things ready for internal testing."

Connor discussed during project scope and timeline section. Again, this is an actual thing that I said on the call. It's a bit of fun really just to really wind up Robbie to be fair. I never said that. Yeah, you did because I'm mean here. But the idea is that it holds us accountable, right? Our values at Renada: trust, consistent reliability, and embrace challenge. The point I was trying to make there is formulate a nice sentence, but it's really important that we're consistently reliable in the things that we are doing. So if we're saying we're going to do something, we best believe we're going to deliver on it and you best believe we're going to hold ourselves accountable if we don't.

So that's kind of the task overview. And then what we do, jump cut a second while I speak, while I set the scene, if Dylan's got to work some magic, what we then do is press this complete session button. And what we do in Halo, and you'll see this in a minute, is we then email this to the client. So I'm just going to go ahead and press send over here. And then basically what it does, if I can just scroll down a little bit, is sends an email a little bit like this. So Dylan please do sanitise the top emails here please. But it basically is it then formulates it and puts it in a nice email template.

So we ask for feedback on the session. Really important to make sure that if you're having a session in Renada that you are having the opportunity every session to give feedback. Sometimes it's too late after a month to ask for feedback. They've already lost a client. We've never actually had that happen at Renada, but again, it's a mechanism. If it was too quick, if it was too high level, if it was too rushed, if the consultant was late, just feedback, you know, give us feedback on it. We're trying to improve all the time. And also, if you give good feedback, it plants a tree. Go figure.

Then what did we do is a section I'm just going to slightly whip through this super quick, but it's going to be blurred unfortunately just because I'm not asking Dylan to sanitise all of this. What is planned for next session? So you can see again from the action, it's just a snippet here. And if we scroll all the way down, blur, blur, blur, blur, blur, blur, blur. What we then embed is the session recording. So we automatically add that recording from Teams and then provide our booking links, which again, we use Calendly actually for this. But again, this is actually our unique booking links for the agent that has finished the call. Booking with Connor for the next session, half an hour or an hour to go through it.

So again, what we're saying here is one of the biggest time kills at Renada was writing all the notes, providing the quality, then going and making all of the tasks. I reckon we're saving between half an hour and an hour. So this one, at least an hour per task, per agent now. And we'll often do maybe three or four calls a day per person. No, maybe not. Maybe two to three per agent. So let's say we're saving, I don't know, let's say there's four of us, have two calls. That's eight hours. We're saving between four and eight hours a day on this. And the quality on the back end is so much better than it ever was before. So that's number two on how Renada uses automation.

Speaker: Between what creating these tasks does. Um, now, so let's jump cut to that section now. So again, what we do when you're seeing that task table, we don't actually have a look at that. The idea is never to go look at it, but it takes all the data from those sections and then pops it in over here. So when we are scoping our tasks every day, all the context from the call is passed across.

And then all we've got to do really is make it look a bit more human, adding any technical detail. So if I just go down here slightly. There we go. Um, so you know, Bobby's done an error here, but there we go. Um so looking at you know how long is it going to take us, a client approves every task by the way, we do approval process here, the due date, the type of work, the technical details, this is what we then internally write: dependencies, hand over, hand over, um et cetera et cetera.

Let me just jump up a second, Dylan might need to sanitise that. God, this is a nightmare. Um, but yeah, so the idea is is that we scope all of our tasks really robustly now and define a really clear path for success. One of the things that um I think it's really important with any businesses is you're really clearly defining expectations. Um, so for us the better we have our inputs the better the outputs are going to be and the fact everyone's on the same page. So we are telling you what we're going to do, what problem it's solving, technical details for the team, any dependencies, requirements, what platform, what type of work.

Um, when we're doing budgeting, we have a budgeting table. Right? So this is just for us internal really to try and understand where the time's going, how much are we spending time scoping it or implementing it or documenting it or handing it over, and then how does that total time formulate? Um you'll have to then also hide Dylan the approver if you see that popup. Um, but yeah, the idea is is that um it's just really improving the way we work at Renada and taking out all of the boring bits. All of my team love the doing. None of us like really writing all of the notes and slogging through this for hours every day. So AI has enabled us to massively speed up this process. So that is number two at Renada on how we use AI within the business.

Okay, we're changing pace a little bit. I was going to show some more automations we're doing in Halo, but I'm going to show you something that we use of a cloud project that is actually related to Halo, but isn't actually directly built in Halo just to just to change the pace, I think, a little bit. So um, live demo time. Here we go.

Uh, not that one. I have two orgs because I am a nerd. So, we have the workflow analyser. Okay. And at some point, we'll make this into our internal tool. Um, but basically what we're doing is um in our Halo we can go to uh a report we've written called workflow debug and details and I'm just going to pick my opportunity workflow. Okay. And basically what we can do is I can go um how do I use this project? Okay, live demo. We will jump cut because it takes a minute to run. So what it's telling me to do is export your HaloPSA workflow configuration as JSON. Now what we've written actually is this very high level report here, honestly um with a bunch of data which basically is telling me stuff like um what is the workflow name, what is the stage name, what is the step name, what is the button name, what is the outcome description, the system use blah blah blah blah blah. And what we do is we export it to JSON. Okay.

Then what I do is I simply copy that. So over here I'm just going to copy this JSON payload um and then grab it here. Here is the JSON. Uh, use a random logo. You'll see why in a minute. And what that does is this has a big prompt that I've curated over time. And it goes and looks at that workflow and then creates a report for me which I can give to client which basically is here is a workflow analysis of any gaps we think you've got within the workflow. So um, and you'll see that whenever this runs in two minutes exactly how that works. So, we're going to have a quick jump cut. Um, and when we get back there will be this will be finished. I won't touch my keyboard. You'll have to just trust me on this one. Um, as how all this works and looks. So, we'll be back in one minute.

Okay, we're back. So, that took like two or three minutes. What I normally do is just leave it off on the other on my other screen. Now, I'm not saying something we just blindly go and give to our clients, by the way. I'm not stupid. Um, but what it does is it takes a lot of the you know initial thinking out of us um out of it for us to then go and really dig into it and again do the things we're good at. Right? What you don't want to do is spend—maybe you do, but we don't really enjoy it—is spend loads of money with us writing out documents and you know detailing all the information. What you really want to do is just get the output. So what we try and leverage AI for is just a tool to help us speed up some of the adminy parts of it and some of the thinking. Right? So this is basically saying this has analysed our internal Renada um CRM opportunity workflow. Now this is our probably our worst workflow because it's only me that uses it and I don't do any reporting on it or anything yet. I don't—get me started. It should be better. Um, the ones we give our clients are way more robust. But the idea is is it takes that report and generates this document. So let's just have a quick look at it together. I'm just going to pull this over here. So what is it saying to me?

Well, uh, that's just a summary. Who cares right now? Um, context capture on project handover action. The project handover button, which moves when opportunities from handover set to closure, currently has no fields configured. This means opportunities can be handed over to delivery team without capturing important transitional context. Fine. Recommendation. Consider adding the following fields: summary, expectations, implementation, timeline, prime implementation, and other considerations. In most scenarios, that's really valid data. I happen to know in hours it just sends an email to the client and goes, "We'll reach out in three to five days." However, if this handover was actually to a service delivery team, then great catch clause. Makes perfect sense. Then as we go down, we have like the time tracking enhancements, right? And part of this prompt is um make sure we're capturing our time consistently. So, it's saying here that the current state, the email action used across qualification, proposal, et cetera, captures time really well. However, um consider whether other communication and coordination activities also warrant time tracking. So recommended review whether actions like capture business information, email vendor and raise quote should include time taken field. Well, raise quote is a bit of anomaly, but email vendor it definitely should have. So the fact we don't have that means we are losing time tracking capabilities. So again, really good catch. I'm not going to bore you through all of this, but um it basically goes through the workflow and looks at the data and drafts us quite an extensive report basically.

Okay. What we then do with this is we then go and validate the claims and add or remove anything that doesn't seem to make sense. Again, if we was to do this manually and you know, just to give you some context, jump cut super quick. This is my opportunity uh workflow within Halo for transparency. If I had to go through every single action within this workflow, look at it. Is it emailing? Is it tracking time? What email is it sent to? Where does it go afterwards? This would take me hours and hours to go through this. And the reality is, yes, AI might miss some piece of information. But so am I. Right? If I do this manually, I'm going to miss information. I'm going to have to walk through it loads of times. Whereas giving it hard data, giving it a set of rules to abide by, look for these things. This is what we expect. Then getting that output as you've seen there is saving us so much time at Renada. And I think the whole point of this series is is what problem are you solving? Well, the problem for me is none of us liked the inputs of these tasks. Having to manually go and look at all of this took ages, right? And you're not paying me for that time. You're paying us for the knowledge that we have. And the output is here's the recommendations. That's what you want. You don't really want all the blurb around all the crap it's found. That's what we need as consultants, right? We need the context. All you need as a partner is go and do these things to make it better. And again, we're trying to speed up the point of delivery.

We're doing the same now with our daily huddles. I get an automated email each day giving me a summary of the business. Um, when we make projects, we're automating all the task creation. When we do an audit, we automate all the audit. So, we automate the entire audit findings through the entire audit at AI and it gives us a complete summary um with prioritisation which we then throw back at Halo and it makes all of the road map items out. There is honestly so much that we're doing at Renada and I I don't want the video to be six weeks long but the point is is it's taken us years to get to this state and we take it project or task at a time and reality is we take it problem at a time. So here's very much um me opening up Renada to you all today and showing you how we're leveraging it. Um, one of the things that I always um, bang on about, I suppose, is is the unique nature of Renada. Um, we are not consulting eight hours Monday through Friday. All of my staff consult on hours of two to three hours a day at a maximum. We are spending loads and loads of time internally building process, building automation, learning AI, building a little bit of internal tooling. That's a debate for another day. But what we're really trying to do is I'm not trying to have a business with thousands of people in it. I'm trying to have a really well-oiled machine with just a few of us. And I believe that is the future of our industry whether you like it or not. I think companies like Renada will last the test of time because we're building really solid foundations. And what we're not trying to do is just hire, hire, hire, higher, higher, because more people—I was gonna use the meme, but more people equals more problems, if you know the meme. Enjoy it. But yeah, that's that's a snapshot into Renada. Um, this is how we're using AI for some of the pieces you've seen today within the business.

Um, I feel vulnerable, so I'm going to have a shower. So, um, yeah, there you go. I've been Connor Fagan. We are Renada Solutions. If you've managed to get through this entire video, I appreciate you. Please do subscribe below. If you see this on LinkedIn, please do share it. We're trying to make more awareness around this AI stuff, right? Um, and yeah, have a great day. Lots of love. Goodbye. Bye.