[0:08] Nirav Chheda: awesome this is this is what you came here for talk all about your problems you guys got costs how many of you of you work for a transportation provider either own or work for a transportation provider please raise
[0:28] Nirav Chheda: your hands high up in the air okay a good lot of you know leave the hands up in the air everyone leave the hands up in the air now over the past thank you over the past few years
[0:41] Nirav Chheda: um I want you to tell me um you know all of these costs that you have your fuel your 99 Cents every time your vehicle moves you have these costs you also have these costs on a daily basis
[0:53] Nirav Chheda: have your costs been increasing the over the past few years leave your hands up if they are okay now leave your hands up if your revenues your reimbursement rates have not been increasing at the rate to match your
[1:11] Nirav Chheda: there we go uh luckily for you all I went to MIT and I crunched the numbers on this it turns out if your costs are increasing and your revenues aren't going up at the same rate so what do
[1:32] Nirav Chheda: we do about this right I mean this is the whole name of the game you all operate private for-profit businesses right you heard a lot about it probably from Plenty of software companies route optimization schedule optimization optimize everything
[1:51] Nirav Chheda: let's optimize all right let's optimize whatever to increase your profits what are those optimizers actually doing and people say oh decrease your deadhead miles this app at the end of the day what is the ultimate metric that these
[2:08] Nirav Chheda: algorithms are actually optimizing for well they're trying to maximize the number of trips that you can do while at the same time minimizing the total distance that's driven by your entire fleet that's pretty good right Mr Mark Porterfield
[2:30] Nirav Chheda: why is that good and uh does it also help you to project how many trips that you can take miles reduces that wear and tear Mark can you come up we're gonna play a quick game here any empty
[3:06] Nirav Chheda: company for you this is just work uh as you do on a daily basis but actually here yeah you're good over there mark this is your anynt company congratulations you operate with three vehicles and uh the top is
[3:19] Nirav Chheda: the start time of a trip the bottom is the end time these are all trips and uh guess what this Friday you're doing 18 trips pretty good right with three drivers yeah but today's only Tuesday I mean you
[3:32] Nirav Chheda: only know as of today that you're going to be doing 18 trips on Friday of course tomorrow and the day after you're gonna get more trips called in for Friday but as of now you just know it today's
[3:43] Nirav Chheda: Tuesday on Friday 18 trips have been called in simple enough called in what are you gonna do and wherever I can think I can make it work exactly and who are you going to have do that or try
[4:05] Nirav Chheda: to figure that out and uh she guesses at it how good is she a software software can tell you that hey these trips don't fit in properly into your schedule and you should reject them pretty good right and
[4:34] Nirav Chheda: uh guess what now today's Wednesday you got five more trips called in three of them fit in there and two more don't fit in and software tells you reject these two and take those three instead of your guesser
[4:47] Nirav Chheda: pretty good right yeah it is pretty good Mark it is pretty good um what we've got and the best of AI based route algorithms questions afterwards but thank you for your this is the best of what schedule optimizers
[5:07] Nirav Chheda: do right now and that's pretty good right you buy it I'm thinking it's going to be doing he's a smart man Mr Mark Porterfield I'm gonna do you one better mark I'm gonna bring you into the future in
[5:27] Nirav Chheda: the future we're not going to be maximizing for those things we're going to be maximizing for profit how hard is it to calculate profit at a per trip level right now so it's time consuming it's time why because
[5:46] Nirav Chheda: there are various costs that go in 30 40 costs I mean you have all of these operating costs that we talked about before that you have to calculate at a per trip level and then all of these factors
[6:01] Nirav Chheda: that the algorithms are figuring out for scheduling have to be married with these costs to just figure out whether it's profitable to do a trip or not it's pretty hard right but thankfully with statistical models we can actually
[6:15] Nirav Chheda: do that so let's go back to our previous example here today's Wednesday you're still projecting Friday's trips what if Mark software said hey you should put an extra vehicle on the road and actually do these four trips instead
[6:29] Nirav Chheda: of rejecting them because it is more profitable to do these trips than not your costs are lower than your revenues for putting this driver on the road and all of your depreciation schedules for your vehicles your labor costs
[6:45] Nirav Chheda: your fuel costs it calculates everything and tells you it's get a bunch more trips and it tells you hey add another driver that's pretty cool right but the thing is you're not going to be on Wednesday saying Hey
[7:03] Nirav Chheda: Driver can you extra driver can you drive on Friday oh it's Thursday I need another driver can you drive you guys roughly have the same number of drivers every day what if software were able to project how many
[7:15] Nirav Chheda: trips you're actually going to have by the time Friday rolls around and it tells you days in advance this is how many drivers you should plan to have on the road that day do you like that thank you
[7:28] Nirav Chheda: I like it too Mr Mark borderfield you all operate companies where stuff happens day off right let's talk about those real-time dispatching additions and changes um Mr Howard berquist you you live and operate in real time why don't
[7:56] Nirav Chheda: you come up help me out with this one we've got yet again three drivers and um again the top of a trip is the start time the bottom is the end time and uh we've got these three drivers
[8:14] Nirav Chheda: and Driver a is taking passenger Bob driver B's taking passenger Laura so on so um the thing is now a trip from Margaret got called in where are we gonna place that trip see who said that she is
[8:40] Nirav Chheda: green okay thank you but you're wrong but you still get chocolate why can't driver see do Margaret driver sees on the other side of town yeah Margaret I mean driver C has availability during that time but they won't
[9:03] Nirav Chheda: be able to get to Margaret do that trip and then come back in time for now the thing is Margaret would be perfect to be taken by driver a but why can't driver a do it same time as
[9:17] Nirav Chheda: Bob thank you who said that could have raised their hand I would have thrown you chocolate but okay let's say we put Margaret under driver a now what do we got to do with Bob where do we place
[9:34] Nirav Chheda: Bob Howard where do we place Bob trick question nobody knows where we can play Spock but software knows software can figure out we should Place Margaret on driver a which would move Bob over to driver B and in
[9:56] Nirav Chheda: order for driver B to do Bob we'd have to move Laura over to driver C which works out day where you get new trips called in things change and you need to move around your schedules who's figuring this
[10:12] Nirav Chheda: out for you let me guess yes you're a human being dispatcher is figuring out all of this for you and guess what they're not just doing one of these they're trying to figure out 10 of these every day
[10:30] Nirav Chheda: that's insane Howard is that insane that's insane you should have software doing this for you instantly and this exists this is the latest an AI and this is only going to get better thank you Howard for playing and
[10:47] Nirav Chheda: uh you've been a great contestant oh and opportunistically there we go let's just simply recap what we just saw new trips get called in what do you do do it for you driver gets stuck for 15 minutes the
[11:16] Nirav Chheda: trip that they were about to do they're going to be late for you're going to try to see if another driver can do it so that trip doesn't happen late and if that driver has availability if anyone if
[11:25] Nirav Chheda: we what do you do to figure that out it's a it's a hazardous world we live in any EMT you're used to it the beginning of the day your driver had nine trips on her schedule and now she
[11:50] Nirav Chheda: can't do any of them and you got to figure out if those nine trips can be moved on in an optimal way to enter the rest of the drivers on your Fleet what are you gonna do but hey
[12:12] Nirav Chheda: you know what's better than click to reoptimize not clicking and still getting reoptimization when your algorithms are continuously running in the background figuring out hey this change just happened with this driver this driver got delayed and we can
[12:31] Nirav Chheda: see all that information in real time and your dispatcher just gets a notification saying hey um I'm the software I just re-optimize for you based on things that are happening in real time is this cool with you dispatcher
[12:44] Nirav Chheda: says okay that's cool with me no clicking necessary boom that is the future company is required to give your drivers break times right you want to come up and play with me know it yet Mike what do you
[13:15] Nirav Chheda: want to do about your driver break times that you're required to give them that how does that work how do your human being staff make that happen well unfortunately you must have been having a slice them at the
[13:34] Nirav Chheda: beginning that day and then you know they're difficult to move so we wind up sub-optimizing what if you could click a and the driver brakes were placed optimally for you based on wherever the drivers had downtime them manually
[13:57] Nirav Chheda: they were placed optimally awesome awesome hey um while you're up here Mike does your company happen to do a lot of will call trips Hospital discharges that can pop up at any time 85 wow that's a lot of
[14:16] Nirav Chheda: percent yeah so how do you organize your fleet's capacity to handle these trips that could pop up when you don't know when they'll pop up with difficulty how much difficulty sub-optimized okay predict the future nobody can predict the
[14:53] Nirav Chheda: future but we can help a little bit with something called machine learning and don't be scared by this phrase at its simplest level machine learning is essentially this you have two points you draw a line and that line
[15:05] Nirav Chheda: helps you figure out where the third point is Mike what if I told you that 90 of your trips that you have our software is going to analyze all those will call trips and see you know all the
[15:20] Nirav Chheda: factors like time of day of week the weather conditions patient characteristics and it's going to take all the information and make patterns out of it and those patterns are going to tell you hey here's the typical wait times
[15:33] Nirav Chheda: and the typical length of appointments at you know X and Y facilities would that be helpful in the great difficulty your people face you know it's gonna be even more helpful when we turn those paddles into predictive models
[15:48] Nirav Chheda: okay and those models are going to keep getting this data and the accuracy in predicting the timing of the will call trips is going to enhance over time pretty awesome right where does that move from difficult to where
[16:04] AUDIENCE MEMBER: are you at right now I mean you know right now we basically try to do predictive modeling for like we have a holiday how many vehicles do we use for holiday so I at least okay
