A logistics head in India may describe his control center to you as a system where large screens show maps of movement with small dots moving. To the sides of the center area would be a large wall of red and amber alerts and a few people in the center area deciding which alert to call someone about.
A decade ago, such control centers would have had all of their team members functioning as operators, relying on phone calls and spreading out information in spreadsheets. However, the dashboard will slowly but surely move away from the center of the control center over the next 5 years. Although visibility will be crucial, solving the issue shown on the screen is only half the battle, and most of the time and money is lost in the other half.
The dashboard era did its job
Track-and-trace, GPS, e-way bill integration, and carrier APIs are some of the latest tools to enable Indian shippers to take a collective glance at their freight in motion. For a long tail of small firms with a large number of operating vehicles, this is a hard-won right.
Each notification requires some form of action where someone needs to notice it, make sense of it, assess the relevance of the information, check for alternative solutions, and then execute on them. At low volumes, a good planner can often react instinctively to such updates. However, as volume increases, all such updates would translate into high-volume triage, with the shipper’s system being governed by the ‘loudest’ or most ‘recent’ update, not necessarily the most critical one.
We wrote about this in our piece on why visibility was step one and the time to decision is the next level of supply chain optimization. The short version: knowing at 2 pm that a truck is late helps much less than knowing at 2 pm what to do about it.
Why the decision is now the expensive part
For years, India's logistics cost was quoted at 13 to 14 percent of GDP. The NCAER study for DPIIT, released in 2025, put it at 7.97 percent for 2023-24, which is close to what developed economies report. That is good news, though I wouldn't celebrate too long. The government's release on the study noted that smaller firms face significantly higher logistics costs. And a Business Standard column on the findings points out that road transport accounts for about 42 percent of the total, which keeps the whole system sensitive to fuel and congestion.
Road-heavy, fuel-sensitive, and fragmented. In a system like that, the margin lives in thousands of small daily choices. Which vehicle takes which load. Whether to wait for a return load. Which carrier gets an urgent shipment when the first one falls through? Each choice is minor. Added up, they are most of the game.
No dashboard improves those choices by itself. A person has to make them, and people run out of hours.
What software does that look like?
"Systems that act" can sound abstract, so here are the places I expect it to show up first.
Exceptions. On a busy morning, hundreds of trips are in motion, and a fraction of them have a real problem. Today, software flags all of them, and a planner sorts through them. The next version ranks them by actual cost and urgency, and for the routine ones, simply handles them. A late vehicle on a flexible delivery shouldn't need a human at all.
Planning. Right now, a planning engine proposes a plan, and the planner spends the morning overriding half of it. That's a lot of skilled time spent correcting software. The shift is a system that commits to the plan inside the rules the business has set and sends people only the cases that fall outside those rules. This is the idea behind our capacity and route planning engine, and I expect most planning tools to move in this direction.
Recovery. When a vehicle breaks down or a carrier rejects a load, the usual response is a flurry of calls. An acting system re-tenders or reroutes within cost and compliance limits, then records what it did and why.
Learning. This is the one people underrate. Every time a planner overrides a recommendation, that override carries information. Either the system was wrong, or the planner knows something the system can't see yet. Software that treats overrides as feedback gets better every week without a retraining project. Software that ignores them keeps making the same mistake and annoying the same person.
None of this requires science fiction. It requires clean execution data, explicit business rules, and a system that sits on both planning and execution instead of one or the other. That last point is why it matters whether your transport management system is a record-keeper or the place decisions actually happen.
How we expect the next five years to unfold
Predictions are cheap, so treat this as a working view and not a forecast.
Now to 2027: better decision support. Most systems will still wait for a human to click. The gains will come from fewer, better alerts and recommendations people actually trust. Many companies will find out that their data is messier than they thought.
2027 to 2029: bounded autonomy. The low-risk, high-frequency decisions get handed over first. Retendering within an agreed rate card. Rescheduling dock appointments. Rerouting around known delays. Humans move from approving each step to setting the limits and auditing the results.
Analysts see the same direction. Gartner predicts that by 2030, half of cross-functional supply chain management solutions will use intelligent agents to execute decisions on their own. That is a global projection, not a promise about any one market, and analyst timelines tend to run optimistic. Still, I find the direction hard to argue with.
2029 to 2031: coordination across companies. This is the part I'm least sure about. If shippers, carriers, and warehouses all run systems that can act, the next friction point is how those systems talk to each other. I'd expect standard ways of sharing capacity and status and some early, awkward versions of software negotiating routine terms with other software. Treat this stage as a reasonable guess.
Where people stay in charge
There is a lot of misinformation out there, such as “software is taking over." The honest version of this story is that there is a split in who makes the decision, and it has to be managed.
I would keep people in charge of financial exposure above a certain threshold, things that have large compliance or safety implications, and customer interactions that cannot be captured by numbers. A simple rule of thumb: things that cost little to reverse and are made quickly by a system are fine to have a system make a call. Things that cost a lot to fix or are public and therefore embarrassing should have a human sign off.
There are two key elements that make this split workable: 1) auditability of all automated decisions (i.e., record of decision, data used, rule applied), and 2) trust in the input data to the agent. Stale vehicle location information or incorrect rate card data is far worse in the hands of a human who can quickly make a wrong decision with certainty.
Many automation projects stall for the reasons mentioned above and not because the model is not good enough. Governance and data quality are hard to automate and therefore typically take longer to set up than the actual model.
What to do this year
You don't need to wait for 2029 to start. A few moves pay off immediately.
Measure your time to make a decision. Pick one common exception, say a delayed pickup, and track how long it takes from the moment the system knows to the moment someone acts. Most teams are surprised by the number.
Choose one class of decision to delegate. Start small and boring. Write down the rules a good planner uses without thinking: the cost ceiling, the preferred carriers, and the cases that always need a call. If you can't write the rules down, you aren't ready to automate the decision yet, and that's useful to learn.
Fix the data that decisions depend on. Vehicle master records, lane rates, e-way bill status, and delivery windows. Dull work, and it decides whether any of the rest functions.
And ask your software vendors a pointed question: when your system flags a problem, what does it do next? If the answer is "it notifies the planner," you're buying a dashboard.
The control room of 2031 will probably be quieter than today's. Fewer red alerts, fewer people watching screens, and more time spent on the small share of decisions that really need judgment. The measure of good logistics software will shift from how much it shows you to how much it takes off your plate and how reliably.
That's the bet we're building toward at Libera. The dashboards aren't going away. They just stop being the point.