Proxima

For Indian D2C and 3PL operations teams

Operational cost review and controlled AI automation for Indian D2C and 3PL teams.

For a reference brand shipping 3,000 parcels a day at 55% COD, the modeled leak across avoidable returns, follow-up hours and support cost runs to ₹15–25 lakh a month. Proxima starts with a free two-day review of one process, computes your number from your own data, and automates only the follow-up work that proves safe — behind your team's approval.

Free 2-day review Impact modeled in rupees No software commitment
NDR trace / PX-24081 Review active

NDR raised on a COD parcel

DEL → PATNA / ₹1,199 COD / first attempt failed

RTO risk
09:08
NDR event received"Customer not available" — Delhivery portal, read within minutes.
System
09:14
Recovery window openNext re-attempt slot closes at 18:00 today; at 28% RTO this parcel becomes a ₹200 loss.
Handoff
09:19
Recommended recovery actionConfirm availability on WhatsApp, capture a landmark, book the re-attempt for tomorrow 10:00–12:00.
Suggest only
09:24
Supervisor review readyCustomer message, courier instruction and operating limits attached.
Approval
Human approval requiredMaking a change in a live system remains blocked until a supervisor approves.

Start with one repeatable process

Where Proxima applies: NDR recovery, COD confirmation, and delivery-status support.

COD-heavy brands

COD confirmation on WhatsApp

A large share of COD returns is decided in the five minutes after the order is placed — confirm the buyer's intent, verify the address, offer prepaid — while the phone is still in their hand.

Proxima runs the confirmation flow, structures every reply into an order-level decision, and escalates only the unusual orders to your team.

3PL and support teams

Where-is-my-order queue

40–55% of support tickets are customers asking what your courier tracking already knows, at a fully loaded cost of ₹70–90 per ticket.

Proxima answers from live courier data across Delhivery, Xpressbees, Ekart and Blue Dart, measures resolution by no repeat contact, and routes genuine complaints to people.

2 daysFocused process review
No chargeNo software commitment
1 processClear scope
3 outputsProcess map, cost in rupees, next step

What the review shows

What the two-day review covers.

The review begins with queues and handoffs rather than with the technology: where work waits, which exceptions recur, and how many hours the team loses each week. On that basis we separate the cases AI can handle from the decisions that should remain human.

Illustrative shipment review

Exception recovery process

A representative process-review format showing how findings would be presented. Customer outcomes must be measured with customer data.

74Process readiness
out of 100
01
Cases wait too long for an ownerMost waiting time occurs before a team accepts the exception.
Measure first
02
Routing can be tested without changing live systemsThe system recommends actions but does not execute them; recommendations are compared against reviewed team decisions.
Suggest only Shadow
03
Your team approves customer promisesSupervisors retain control until quality and the agreed rules and limits have been proven.
Approval gate
04
Live system changes need an agreed recovery pathLogging, rollback, access and escalation are defined before making a change in a live system.
IT review

How it works

Deployment stages: observe, recommend, approve, automate.

The system operates with full transparency: each new responsibility is granted only after the system demonstrates its accuracy against a threshold your team agreed to in advance, and every consequential action waits for human approval.

01

Understand the work Observe

We map one process on a read-only basis — queues, handoffs, exception rates and the cost of manual follow-up — without making any changes to your systems.

02

Recommendations without execution Shadow

The system recommends actions but takes none of them; every suggestion is scored against the decision your team actually made.

03

Supervisor approval of each action Human approval

A person reviews and approves each action, with the supporting data, agreed operating limits and recovery path visible.

04

Automation of proven categories Controlled automation

Routine actions that have met the agreed quality threshold may run automatically within defined limits; if accuracy falls below that threshold, the category reverts to human approval.

Commercial model

Engagement structure after the review.

  • 01Free process review · 2 daysYou receive a process map, an assessment of the hours being lost, and a recommended next step. There is no charge and no obligation to proceed with automation.
  • 02Deployment sprint · target 6 weeksAn implementation engineer and an industry specialist work with your team through observation, suggestions and human approval. Each action category advances only after it meets the performance measures agreed for that process.
  • 03Controlled rolloutOnly routine cases that meet the agreed quality level may run automatically, and logging, recovery and escalation requirements are defined before production use.

How the engagement works

Terms, exit conditions, and handover.

If automation is not the right answer, you retain the process map and the recommendation at no cost. Where it is, the next step is a fixed-scope implementation that your IT team can review end to end, and subscription terms begin only after an action category has demonstrated its accuracy.

Free review first Agreed quality gate Terms after proof
  • Scope and price in writingThe implementation scope, deliverables, assumptions and price are agreed before work begins.
  • Exit at every gateIf a category does not reach the agreed accuracy level, the engagement stops at that gate, and you retain the maps, the data and the analysis.
  • Customer-owned evidenceThe proposal defines credential ownership, operational-data boundaries, required logs and where those records are stored.
  • Handover plannedOwnership of connectors, decision rules, documentation, support and escalation is agreed for each implementation.

Systems and security

Integration and security approach.

Proxima is designed around approved interfaces, customer-specific controls and inspectable decisions rather than a rip-and-replace deployment.

Integration approach

Integration through approved APIs and data feeds.

Proxima can connect to approved APIs, databases or file feeds across transport, warehouse, ERP, banking and point-of-sale systems. An implementation engineer builds the required connections for the agreed process.

  • Data inAdapters connect to your APIs, databases or file feeds. Events flow in a standard format.
  • Approval-first changesDuring supervised testing, actions on live systems remain behind human approval. Later permissions are agreed by action category.
  • Audit requirementsBefore production, we agree which inputs, rules, approvals, actions and recovery events must be recorded.

Security and data handling

Data handling and deployment boundaries.

Proxima is designed to run in your cloud or an approved on-premises environment. Before production, we document what data is processed, where it is stored, which service calls are permitted and which operational metrics may be shared.

  • Data residencyCloud, on-premises and external-service boundaries are reviewed with your IT, legal and security teams. Regulated processes require customer-specific RBI and DPDP review.
  • Access controlCredentials, service identities, secret storage, expiry and revocation are agreed for the approved architecture.
  • Reviewable by ITThe InfoSec appendix covers data handling, access, retention and the questions your IT team should ask.

Proof you can inspect

Review materials available before you commit.

All materials are available for review in advance: what the two days include, what the readout looks like, and the questions your security team should ask.

Start here

Request a free process review.

The initial process review is provided at no cost and computes your numbers from your own order, courier and support data: the monthly leak, the hours lost, and which steps merit automation. A fixed implementation scope and price follow only where the evidence supports them.

Figures on this page are modeled for an illustrative reference brand shipping 3,000 parcels a day at 55% COD. Your review is computed from your own data.