The AI Route Planner We Built for Our Own Field Team
We built an AI-assisted route planner that turns a CRM full of prospects into focused, drivable days — and owns every byte of the data.
Anyone who spends their days visiting customers in person knows the quiet tax that comes before the first handshake: figuring out where to go. You open the CRM, scroll through a list of prospects scattered across a metro area, glance at a map, and try to stitch together a day that doesn’t have you crossing the same highway four times. It’s fifteen minutes of planning that rarely produces a good answer, repeated every single morning.
We had this exact problem with our own outside sales work, so we built a tool to solve it. We call it an AI-assisted route planner, and it does something deceptively simple: it turns a CRM full of addresses into a focused, drivable day. We built it for ourselves first — which is the only honest way to recommend a thing — and the same approach now works for any business that sends people into the field.
This is a good example of what we mean when we describe ourselves as a Managed Intelligence Provider. The automation sits on top of dependable systems you already have. It doesn’t replace the person doing the work; it removes the busywork that stands between them and the work that matters.
The problem with planning one perfect route
The instinct, when you first think about route optimization, is to ask for the optimal route — one mega-loop that touches every prospect in your database with the least possible driving. That sounds efficient. In practice it’s useless. Nobody visits two hundred prospects in a day. The “optimal” all-in-one route just tells you that prospect 147 is theoretically three minutes closer if you visit it before prospect 148, which is not a decision anyone needs help making.
Field reps don’t think in single mega-routes. They think in days. “I’m going to be up north today — who else is up there?” That’s the real planning unit, and it’s the one we built around.
So instead of solving for one giant loop, the planner solves for a region day. You pick a center — a town, a postal code, a point on the map — and a radius around it. The tool pulls every prospect in your CRM that falls inside that circle, then builds the best same-day loop through them, starting and ending at your office. Tomorrow you pick a different center, and over the course of a week the whole territory gets covered in geographically coherent days, with very little overlap between them. It mirrors how good field reps already work, just without the morning guesswork.
A day only holds so many hours
Here’s where it stops being a map toy and starts being genuinely useful. A circle on a map might contain forty prospects. A workday contains maybe eight hours. The planner knows the difference.
When it builds a route, it accounts for the real shape of a day: when you start, when you want to be home, and how long a typical visit takes. It then schedules only as many stops as actually fit between those bookends — drive time and visit time included. The prospects that don’t fit don’t disappear. They stay in the pool, ready for the next time you plan that region.
That turns a single plan into a rolling campaign. Day one covers the closest, densest cluster. Day two picks up where day one ran out of hours. The territory gets worked down methodically instead of in a frantic, overlapping scramble. You always know what’s been visited, what’s scheduled, and what’s still waiting — without keeping any of it in your head.
Two ways it keeps you from wasting a stop
Driving to a prospect is the expensive part of the job, so the planner works to make sure every stop is worth the drive.
First, it won’t send you somewhere you were just last week. Before it schedules anyone, it checks your CRM’s own activity history. If a prospect was contacted recently, it quietly sets them aside on a cooldown so they don’t clutter today’s route. Walking into an account you touched four days ago doesn’t help anyone; the system simply skips it until enough time has passed.
Second — and this is the part people get most excited about — it prepares you for the door. For every stop on the day’s route, the planner assembles a short brief: who the primary contact is, what’s happened recently with the account, and two or three concrete talking points pulled from what your systems already know about that organization. Where it has rich detail it uses it; where it doesn’t, it falls back to the basics from the CRM so you’re never walking in completely cold. This is the kind of quiet desk research that, done by hand, would take longer than the visit itself. Done automatically, it means the rep arrives prepared instead of improvising.
That second feature is, in a real sense, an AI virtual employee — the diligent assistant who sits at the desk every morning, reads up on each account, and hands the rep a tidy briefing before they leave. It never gets bored, never skips the prep, and never asks for the day off.
The human is still driving — literally and figuratively
We are deliberate about one thing: the automation never acts on its own. It proposes; a person disposes.
The planner builds the day and lays it out on a map, but nothing is committed until the rep looks at it and agrees. The rep makes the visits — the conversations, the judgment, the relationship — because that’s the part no software should touch. When a visit is done, they tap a single button, and the system records it back into the CRM as a completed activity, so the office and the field stay in sync without anyone doing double entry.
Crucially, every one of those write-backs is reversible. The tool keeps an audit log of everything it records, and a single action can undo a day’s worth of entries cleanly if something needs to change. We treat reversibility and a clear audit trail as non-negotiable, not afterthoughts — the same standard we hold for any system that writes into a business’s records. Automation you can’t undo isn’t a convenience; it’s a liability.
This is the human-in-the-loop principle in practice. The machine handles geography, arithmetic, and research — the parts that are tedious and error-prone for people. The person handles the relationship and the final call — the parts that are tedious and error-prone for machines.
Why we host it ourselves
We built the planner on open-source mapping and routing engines — the same class of technology that powers the navigation apps on everyone’s phone — and we run the whole thing on our own infrastructure rather than renting it as a per-seat subscription.
That choice carries through in three ways that matter to any business:
- You own your data. Your customer list, your visit history, and your routes never leave systems you control. For anyone in a regulated industry, that alone is worth the effort.
- The cost doesn’t scale against you. Per-seat field-sales subscriptions charge every user, every month, forever. A self-hosted tool you own has a real build cost and then a modest cost to keep running — it doesn’t quietly tax every new hire.
- It bends to fit you, not the other way around. Because we control the code, the planner works the way your team actually works. A subscription product makes you adopt its assumptions; this one adopts yours.
The routing quality is not a compromise, either. The engines doing the geocoding and the route math are the open, battle-tested ones behind major commercial mapping products. You give up nothing on the optimization to gain everything on ownership.
The same engine, pointed at other problems
We talk about this tool as a sales-route planner because that’s the problem we had. But the capability underneath is broader, and that’s the part worth sitting with.
Strip away the labels and what you have is an engine that takes a list of locations and a set of constraints — a workday, priorities, who was visited recently — and produces a sensible, optimized plan a human can approve and act on. Point that at field-service technicians and it’s dispatch routing. Point it at deliveries and it’s a logistics planner. Point it at inspections, audits, or installs and it’s a way to cover a region without burning hours in the truck. The pattern is the same: pull the data you already have, do the optimization no human enjoys doing, prepare the person, and let them decide.
That’s the thread running through everything we describe as managed intelligence. It starts with the unglamorous foundation — networks that stay up, data that’s backed up, systems that are secured — and then adds automation on top that gives your people time back. The AI is additive. It makes a good team faster; it doesn’t replace them.
See what it could do for your team
If your business sends people into the field — sales, service, delivery, inspection — there’s a strong chance a slice of every day is being lost to planning, backtracking, and prep that could be handled automatically. We found ours, built a tool to reclaim it, and now run on it daily.
The best way to find out where that opportunity lives in your operation is to start with a clear picture of how your technology and your workflows fit together today. That’s exactly what our free business IT assessment is for. We’ll look at where automation can genuinely help, where it shouldn’t go anywhere near, and what a practical first step looks like for your team.
The AI Route Planner We Built for Our Own Field Team
We built an AI-assisted route planner that turns a CRM full of prospects into focused, drivable days — and owns every byte of the data.
Anyone who spends their days visiting customers in person knows the quiet tax that comes before the first handshake: figuring out where to go. You open the CRM, scroll through a list of prospects scattered across a metro area, glance at a map, and try to stitch together a day that doesn’t have you crossing the same highway four times. It’s fifteen minutes of planning that rarely produces a good answer, repeated every single morning.
We had this exact problem with our own outside sales work, so we built a tool to solve it. We call it an AI-assisted route planner, and it does something deceptively simple: it turns a CRM full of addresses into a focused, drivable day. We built it for ourselves first — which is the only honest way to recommend a thing — and the same approach now works for any business that sends people into the field.
This is a good example of what we mean when we describe ourselves as a Managed Intelligence Provider. The automation sits on top of dependable systems you already have. It doesn’t replace the person doing the work; it removes the busywork that stands between them and the work that matters.
The problem with planning one perfect route
The instinct, when you first think about route optimization, is to ask for the optimal route — one mega-loop that touches every prospect in your database with the least possible driving. That sounds efficient. In practice it’s useless. Nobody visits two hundred prospects in a day. The “optimal” all-in-one route just tells you that prospect 147 is theoretically three minutes closer if you visit it before prospect 148, which is not a decision anyone needs help making.
Field reps don’t think in single mega-routes. They think in days. “I’m going to be up north today — who else is up there?” That’s the real planning unit, and it’s the one we built around.
So instead of solving for one giant loop, the planner solves for a region day. You pick a center — a town, a postal code, a point on the map — and a radius around it. The tool pulls every prospect in your CRM that falls inside that circle, then builds the best same-day loop through them, starting and ending at your office. Tomorrow you pick a different center, and over the course of a week the whole territory gets covered in geographically coherent days, with very little overlap between them. It mirrors how good field reps already work, just without the morning guesswork.
A day only holds so many hours
Here’s where it stops being a map toy and starts being genuinely useful. A circle on a map might contain forty prospects. A workday contains maybe eight hours. The planner knows the difference.
When it builds a route, it accounts for the real shape of a day: when you start, when you want to be home, and how long a typical visit takes. It then schedules only as many stops as actually fit between those bookends — drive time and visit time included. The prospects that don’t fit don’t disappear. They stay in the pool, ready for the next time you plan that region.
That turns a single plan into a rolling campaign. Day one covers the closest, densest cluster. Day two picks up where day one ran out of hours. The territory gets worked down methodically instead of in a frantic, overlapping scramble. You always know what’s been visited, what’s scheduled, and what’s still waiting — without keeping any of it in your head.
Two ways it keeps you from wasting a stop
Driving to a prospect is the expensive part of the job, so the planner works to make sure every stop is worth the drive.
First, it won’t send you somewhere you were just last week. Before it schedules anyone, it checks your CRM’s own activity history. If a prospect was contacted recently, it quietly sets them aside on a cooldown so they don’t clutter today’s route. Walking into an account you touched four days ago doesn’t help anyone; the system simply skips it until enough time has passed.
Second — and this is the part people get most excited about — it prepares you for the door. For every stop on the day’s route, the planner assembles a short brief: who the primary contact is, what’s happened recently with the account, and two or three concrete talking points pulled from what your systems already know about that organization. Where it has rich detail it uses it; where it doesn’t, it falls back to the basics from the CRM so you’re never walking in completely cold. This is the kind of quiet desk research that, done by hand, would take longer than the visit itself. Done automatically, it means the rep arrives prepared instead of improvising.
That second feature is, in a real sense, an AI virtual employee — the diligent assistant who sits at the desk every morning, reads up on each account, and hands the rep a tidy briefing before they leave. It never gets bored, never skips the prep, and never asks for the day off.
The human is still driving — literally and figuratively
We are deliberate about one thing: the automation never acts on its own. It proposes; a person disposes.
The planner builds the day and lays it out on a map, but nothing is committed until the rep looks at it and agrees. The rep makes the visits — the conversations, the judgment, the relationship — because that’s the part no software should touch. When a visit is done, they tap a single button, and the system records it back into the CRM as a completed activity, so the office and the field stay in sync without anyone doing double entry.
Crucially, every one of those write-backs is reversible. The tool keeps an audit log of everything it records, and a single action can undo a day’s worth of entries cleanly if something needs to change. We treat reversibility and a clear audit trail as non-negotiable, not afterthoughts — the same standard we hold for any system that writes into a business’s records. Automation you can’t undo isn’t a convenience; it’s a liability.
This is the human-in-the-loop principle in practice. The machine handles geography, arithmetic, and research — the parts that are tedious and error-prone for people. The person handles the relationship and the final call — the parts that are tedious and error-prone for machines.
Why we host it ourselves
We built the planner on open-source mapping and routing engines — the same class of technology that powers the navigation apps on everyone’s phone — and we run the whole thing on our own infrastructure rather than renting it as a per-seat subscription.
That choice carries through in three ways that matter to any business:
- You own your data. Your customer list, your visit history, and your routes never leave systems you control. For anyone in a regulated industry, that alone is worth the effort.
- The cost doesn’t scale against you. Per-seat field-sales subscriptions charge every user, every month, forever. A self-hosted tool you own has a real build cost and then a modest cost to keep running — it doesn’t quietly tax every new hire.
- It bends to fit you, not the other way around. Because we control the code, the planner works the way your team actually works. A subscription product makes you adopt its assumptions; this one adopts yours.
The routing quality is not a compromise, either. The engines doing the geocoding and the route math are the open, battle-tested ones behind major commercial mapping products. You give up nothing on the optimization to gain everything on ownership.
The same engine, pointed at other problems
We talk about this tool as a sales-route planner because that’s the problem we had. But the capability underneath is broader, and that’s the part worth sitting with.
Strip away the labels and what you have is an engine that takes a list of locations and a set of constraints — a workday, priorities, who was visited recently — and produces a sensible, optimized plan a human can approve and act on. Point that at field-service technicians and it’s dispatch routing. Point it at deliveries and it’s a logistics planner. Point it at inspections, audits, or installs and it’s a way to cover a region without burning hours in the truck. The pattern is the same: pull the data you already have, do the optimization no human enjoys doing, prepare the person, and let them decide.
That’s the thread running through everything we describe as managed intelligence. It starts with the unglamorous foundation — networks that stay up, data that’s backed up, systems that are secured — and then adds automation on top that gives your people time back. The AI is additive. It makes a good team faster; it doesn’t replace them.
See what it could do for your team
If your business sends people into the field — sales, service, delivery, inspection — there’s a strong chance a slice of every day is being lost to planning, backtracking, and prep that could be handled automatically. We found ours, built a tool to reclaim it, and now run on it daily.
The best way to find out where that opportunity lives in your operation is to start with a clear picture of how your technology and your workflows fit together today. That’s exactly what our free business IT assessment is for. We’ll look at where automation can genuinely help, where it shouldn’t go anywhere near, and what a practical first step looks like for your team.