Go back to Market Research
Pando supplies transportation management software for manufacturers, distributors, and retailers. Its applications cover freight procurement, load planning, carrier assignment, shipment tracking, and freight audit and payment. Fleetx acquired the Pando TMS business in August 2026, retaining the Pando brand and leadership. (1, 3)
Overall supply chain score: 5.2/10.
| Dimension | Score |
|---|---|
| Supply chain depth | 5.4/10 |
| Decision and optimization substance | 5.2/10 |
| Product and architecture integrity | 5.0/10 |
| Technical transparency | 5.4/10 |
| Vendor seriousness | 4.8/10 |
Pando has substantial freight execution functionality and a documented optimization method. Its main weakness is the distance between these bounded tasks and the much broader promise of autonomous supply chains. The public numerical formulation concentrates on transport charges; it does not establish decisions valued against joint demand, transit-time, capacity, and inventory uncertainty. Agent-generated correspondence and faster reporting can remove administrative work without resolving that economic gap. (1, 5, 6, 8)
Corporate history also changes the interpretation of the AI evidence. Pando separated its India TMS and US AI businesses in April 2025. Freehand launched in February 2026 as the continuation of the team and technology behind Pi. Pando’s website still carries Pi material alongside the Fleetx-owned TMS. Those overlapping pages describe a shared history, while current purchasing and product responsibilities span distinct businesses. (2, 3, 4)
Pando versus Lokad
Pando operates close to the freight transaction: preparing a load, obtaining a carrier rate, following a shipment, and matching an invoice. Lokad’s supply chain optimization starts with the economic choice: what to buy, produce, allocate, move, or discount, given uncertain future demand and supply. The perspective developed in Introduction to Supply Chain connects those choices through their financial consequences rather than optimizing each departmental workflow separately. (1)
Consider a shipment that can wait for consolidation or leave immediately in a partly empty vehicle. A loading engine can identify a cheaper physical arrangement. The economic decision also depends on whether waiting jeopardizes production or sales, how uncertain the next arrivals are, and what alternative uses exist for the inventory. Pando’s disclosed delivery-cost formulation captures a useful transport subproblem; it does not establish this broader calculation. Lokad’s approach requires modeling these consequences explicitly and revising the model when actual failures reveal omitted costs or options. (5)
The operating burden differs as well. Pando maintains a transaction platform, carrier workflows, customer configurations, and associated support. Lokad’s Supply Chain Scientists maintain the customer’s numerical decision model and its data preparation. Both involve continuing engineering work. The relevant productivity measure is the cost of reliably producing and improving decisions across the business, including that work—not the number of screens or messages an assistant can traverse. (6, 12)
What the optimization actually computes
A concrete loading model, with a narrow economic objective
Pando’s December 2023 announcement links its machine-learning claim to US11537977B1, granted in December 2022. The patent uses deep reinforcement learning to combine three-dimensional loading and delivery assignments, masking actions that violate compatibility, stacking, or access constraints. This addresses a real interaction: the cheapest route and the best packing arrangement cannot generally be selected independently. (5, 16)
The illustrated per-vehicle cost is:
delivery charge = vehicle charge + (number of consignees − 1) × drop charge
The two charges are fixed inputs. Training rewards loading sequences according to their contribution to the resulting delivery cost. (5)
That objective can suit a contracted tariff. It leaves important business choices outside the illustrated calculation: accepting an additional day’s delay, protecting a scarce component, splitting a high-value order, or tolerating a more expensive carrier to reduce a costly tail risk. Better optimization of the stated charge cannot recover a consequence absent from the model. This is the Holimization problem: the formulation itself must improve as the business learns, rather than merely searching its existing action space more intensively.
No reproducible customer benchmark accompanies the patent. Its relationship to each current planning function is unspecified. The current TMS separately advertises consolidation, route constraints, carrier selection, predictive arrival times, and three-dimensional loading. Its public description does not expose how those calculations share uncertainty or reconcile their objectives. (1, 5)
Procurement scenarios do not establish risk-adjusted decisions
Pando’s January 2025 procurement article describes clustering lanes, comparing bids, and adjusting allocation scenarios across cost and carrier considerations. The shipper accepts a recommendation before the counteroffer proceeds. The Pi procurement page extends the workflow into carrier correspondence, bid clarification, and contract updates. (6, 7)
These steps can reduce the effort of running a tender. Economically, however, a cheap award is only useful if the carrier can deliver the required service under the conditions that actually occur. A scenario comparison needs calibrated probabilities for transit times, tender rejection, capacity shortages, and future shipment volumes, including their dependencies. Public Pando material does not show those distributions entering a common expected-profit calculation. Predictive ETAs and external risk information are inputs to that task, not its completion. (1, 7)
The procurement page’s promise of 100% service reliability has no defined observation window, failure denominator, or contractual remedy. Physical transportation retains uncertainty after an RFQ has been automated. An absolute service headline obscures the very trade-off that a freight decision engine should quantify. (6)
AI productivity and the work that remains
Pi automated real administrative steps
At Pi’s February 2025 launch, Accuride described extracting supplier emails and packing lists into shipment records ready for review, reducing a task from days to minutes. That is a concrete document-processing improvement. It does not measure the remaining review effort or prove autonomous freight decisions. (19)
An undated Pi customer story describes an unnamed industrial components manufacturer with more than $5 million in freight spend. Pando reports quarterly procurement events falling from 8–12 hours to under two hours, more than $200,000 in annual freight-payment leakage eliminated, and a 90% productivity improvement in execution workflows. The first measure gives a task and a before/after duration; the broader productivity figure supplies no total labor baseline, implementation effort, or ongoing support denominator. A quarterly task saving alone cannot establish a 90% reduction in the work of running logistics. (9)
A 2025 JBF Consulting paper documents a gradual adoption method: agents handle small processes under substantial initial oversight, then become parts of larger workflows. The examples include requesting a missing receipt, flagging a rate discrepancy, and sending a weight mismatch for master-data correction. It also describes confidence scores and customer feedback used to refine the system. These are useful controls, but the paper gives no calibration of those confidence scores or measured false-action rate. (20)
Current Pando analytics preserves a human decision process
In August 2026, Pando’s head of product and engineering described a specific answer to inconsistent conversational analytics: a useful analysis is reviewed, locked to a definition, and reused with consistent execution. The knowledge graph supplies operational context. This reduces the danger of a fresh LLM interpretation silently changing a recurring financial report. (8)
The resulting workflow still moves a person from a business question to an answer and then to a decision. It improves analytical access. It does not demonstrate a sustained decline in the number of interventions required per shipment or in the labor needed to revise operational policies. Nor does a record of past decisions establish that the system can distinguish a sound policy from a repeatedly accepted mistake. The next material improvement would be a measured process for turning operational failures into revised, tested decision logic. (8)
Technology history and roadmap integrity
| Date | Public claim or delivery | Technical significance |
|---|---|---|
| April 2018 | The founders presented networked delivery management and raised $2 million in seed funding. | The original diagnosis concerned fragmented logistics relationships and systems. (14) |
| November 2021 | Zydus selected 10xTMS, with AI/ML dispatch planning, visibility, and automated invoicing among the promised functions. | The announcement describes intended deployment benefits, without a measured AI contribution. (15) |
| December 2022 | The loading and consignment-delivery patent was granted. | A concrete technical disclosure. (5) |
| December 2023 | Pando promoted autonomous logistics, predictive ETAs, and the patented algorithm. | The scope of the autonomy claim became wider than the disclosed optimization formulation. (16) |
| September 2024 | Pando promoted blockchain for logistics documentation and payments. | The technical explanation included an incorrect account of pharmaceutical tracing requirements. (17, 18) |
| February–April 2025 | Pi launched; the TMS and US AI businesses then received distinct operating models and roadmaps. | Workflow agents became a new product direction, followed by an organizational separation. (2, 19) |
| February–August 2026 | Freehand launched from the Pi team; Fleetx acquired the Pando TMS business. | Current AI ownership and product continuity need to be traced through the split. (3, 4) |
The 2024 blockchain article is a revealing failure of technical precision. It says DSCSA mandated EPCIS standards “including blockchain.” FDA’s guidance instead recommends EPCIS as a data-exchange standard compatible with different technological approaches; it explicitly allows choices suited to the trading partner. Blockchain is not part of a mandated technology stack. Pando attaches the fashionable implementation label to a regulatory requirement it does not substantiate. (17, 18)
The stronger parts of the history are the loading patent and the current explanation of reproducible analytics. The weaker parts repeatedly expand the language—from networked execution to autonomous supply chains—without comparable expansion of the published economic model or validation. The patent supplies a specific method against which the earlier AI claims can be assessed; their broader economic promises remain unverified. (5, 8, 15, 16)
Freehand explicitly calls itself Pi’s evolution and now targets wider supply-chain spend management. Those claims belong to Freehand. Fleetx’s acquisition announcement promises a broader fleet-to-freight stack; it does not document the completed integration, shared decision model, or operational gains from that combination. (3, 4)
Software bill of materials and security
Components that can be identified
| Layer | Public evidence | Implication |
|---|---|---|
| Application backend | Pando’s August 2026 engineering vacancy assigns Node.js development and documentation. | A maintained application backend is directly evidenced. (12) |
| Data and interface hiring | The same vacancy requests MongoDB/PostgreSQL experience; Vue.js is an additional qualification. | The database alternatives and optional UI skill are hiring requirements; the deployed combination is unspecified. (12) |
| Cloud infrastructure | The January 2026 security self-assessment names AWS EC2, S3, and KMS. | Compute, document storage, and key management depend on AWS services. (13) |
| AI analytics | The August 2026 engineering article describes a knowledge graph and approved reusable analyses. | This adds maintained analytical definitions to the application. (8) |
Pando’s public component evidence is too limited to estimate hosting cost or the effort of upgrading the full application. The more visible ownership burden is semantic: maintaining carrier identifiers, rate cards, contractual exceptions, document extraction, integrations, and the meaning of recurring reports. A shared interface does not remove those dependencies. The 2025 business split and 2026 acquisition add a concrete product-boundary question about which organization maintains each part. (2, 3, 6, 8)
Security depends on enforceable boundaries
Pando’s published security self-assessment describes separate administration hosts protected by MFA distinct from normal corporate credentials, with logged access. It also describes S3 versioning and KMS-backed encryption. These are identifiable mechanisms. The tenant-separation answer combines a general statement of logical segregation with EC2 hypervisor isolation; it does not explain enforcement between customers sharing an application or database. A network-defense answer refers to encryption-key policy rather than explaining attack detection or containment. (13)
That leaves consequential questions for freight agents. A carrier email or uploaded invoice is an untrusted input; an agent allowed to update rates, dispatch instructions, or payment records has authority beyond reading it. The public Pi documentation describes business guardrails and human validation, but does not expose an independently enforced boundary between document instructions and executable authority, restricted outbound access, or recovery after an incorrect external write. Human confidence is not a substitute for those controls. (6, 20)
Configuration maintenance, ownership cost, and accountability
Pi onboarding requires enterprise data preparation, use-case selection, decision flows, guardrails, and human validation before autonomy increases. Each new carrier practice or contract exception can change these definitions. The permanent cost is therefore the labor required to detect stale assumptions, implement corrections, test their wider effects, and keep the automation running as the business changes. (6)
Pando’s August 2026 customer-success vacancy assigns value measurement, success plans, business reviews, training, issue coordination, renewals, and expansion. That is a continuing service relationship with explicit responsibility to quantify savings. The public offer does not quantify the customer and vendor hours consumed by it or show the cost per correct operational decision over time. (12)
The Accuride record has useful commercial substance. Its June 2023 selection announcement identifies visibility and inbound execution as the initial problem. A later freight-audit article attributes more than $1.5 million in annual savings and recovery of startup costs within four months to the phased implementation. The latter does not publish the freight-spend baseline, calculation, or separation of software effects from rate negotiations and operational changes. It supports a reported customer benefit, with limited causal attribution. (10, 11)
For recurring accountability, shipment and invoice histories should connect each recommendation to its eventual consequences. A bad allocation should lead to examination of the price, service-risk estimate, and omitted alternatives that produced it—not merely a request that the planner review more exceptions. The published materials do not trace such a failure through model revision and retesting, or define who bears the economic cost when the promised savings fail to appear. (6, 8, 12)
Supply chain score
Supply chain depth
| Criterion | Score |
|---|---|
| Economic framing | 6/10 |
| Decision end-state | 6/10 |
| Conceptual sharpness on supply chain | 6/10 |
| Freedom from obsolete doctrinal centerpieces | 6/10 |
| Robustness against KPI theater | 3/10 |
Average: 5.4/10.
Economic framing — 6/10. Transport costs and invoice recoveries have direct financial meaning. The disclosed objective omits important consequences of waiting, shortages, and uncertain service, limiting its reach across the supply chain. (5, 11)
Decision end-state — 6/10. Loads, tenders, allocations, and payable records are actionable outputs. The boundary between autonomous completion and approval still varies by workflow and product generation. (1, 6, 7)
Conceptual sharpness on supply chain — 6/10. Coupling packing and delivery addresses a genuine interaction. The wider links between freight, inventory, production, and demand remain less developed in the public formulation. (5)
Freedom from obsolete doctrinal centerpieces — 6/10. Freight transactions and operational constraints organize the product, rather than ABC inventory classes or an S&OP ritual. Scenario selection and carrier-performance reviews still preserve substantial manual decision work. (1, 7, 8)
Robustness against KPI theater — 3/10. Absolute reliability and broad productivity claims lack the denominators needed to assess them. Task-level timings are more precise, but cannot support the same conclusion about total operating labor. (6, 9, 19)
Decision and optimization substance
| Criterion | Score |
|---|---|
| Probabilistic modeling depth | 3/10 |
| Distinctive optimization or ML substance | 7/10 |
| Real-world constraint handling | 7/10 |
| Decision automation and labor productivity | 5/10 |
| Resilience under real operational complexity | 4/10 |
Average: 5.2/10.
Probabilistic modeling depth — 3/10. Arrival predictions and risk information are advertised. Calibrated joint demand and supply uncertainty, propagated into economic action values, is not demonstrated. (1, 7, 16)
Distinctive optimization or ML substance — 7/10. The disclosed optimization is technically specific and goes beyond a generic AI label. Missing reproducible comparisons and deployment measurements limit assessment of its practical advantage. (5)
Real-world constraint handling — 7/10. The physical problem receives substantive treatment. A mechanism for routinely revising constraints and their economic meaning across the broader operation is less visible. (1, 5)
Decision automation and labor productivity — 5/10. Published cases identify completed administrative tasks and shorter workflows. Remaining oversight, exceptions, and maintenance are not measured across a whole customer operation, and Pi’s corporate continuation must be distinguished from the current TMS. (2, 4, 9, 19)
Resilience under real operational complexity — 4/10. Operational histories and confidence-guided adoption provide feedback channels. The public record does not show a costly failure producing a revised formulation, replay against past decisions, and validated production correction. (8, 20)
Product and architecture integrity
| Criterion | Score |
|---|---|
| Architectural coherence | 6/10 |
| System-boundary clarity | 5/10 |
| Security seriousness | 5/10 |
| Software parsimony and ownership cost | 4/10 |
| Compatibility with programmatic and agent-assisted operations | 5/10 |
Average: 5.0/10.
Architectural coherence — 6/10. Procurement, execution, and audit share relevant freight entities, with a described path from exploratory analysis to consistent reports. A shared economic model across those functions and the acquired fleet platform is not established. (1, 3, 8)
System-boundary clarity — 5/10. Corporate announcements identify the split and acquisition. The continuing overlap of TMS and Pi material leaves present product availability, operating responsibility, and integration boundaries less clear. (2, 3, 4, 6)
Security seriousness — 5/10. Administrative-access and recoverable-storage mechanisms are described. Application-level tenant enforcement and the containment of agents processing untrusted documents remain insufficiently exposed. (6, 13)
Software parsimony and ownership cost — 4/10. Parts of the application and cloud dependencies are identifiable. Neither their operating cost nor the continuing burden of customer-specific definitions and integrations is quantified. (8, 12, 13)
Compatibility with programmatic and agent-assisted operations — 5/10. Connected execution and document agents support automation. The published operating method still requires significant setup and human validation, with limited evidence of controlled autonomous improvement. (6, 20)
Technical transparency
| Criterion | Score |
|---|---|
| Public technical documentation | 6/10 |
| Inspectability without vendor mediation | 4/10 |
| Portability and lock-in visibility | 4/10 |
| Implementation-method transparency | 7/10 |
| Evidence density behind technical claims | 6/10 |
Average: 5.4/10.
Public technical documentation — 6/10. The patent, engineering article, hiring record, and security questionnaire expose meaningful mechanisms. Current model evaluations and operational failure-handling documentation remain sparse. (5, 8, 12, 13)
Inspectability without vendor mediation — 4/10. Readers can examine the illustrated algorithm and implementation outline. They cannot reproduce the reported customer gains or test the current decision engine from the supplied artifacts. (5, 9, 11)
Portability and lock-in visibility — 4/10. Integration with existing enterprise systems is central to the offer. Exportability of the accumulated graph, decision definitions, validation history, and carrier-specific automation is not explained. (1, 6, 8)
Implementation-method transparency — 7/10. The adoption sequence identifies data preparation, use cases, oversight, and gradual expansion. Recurring maintenance effort and quantified release criteria remain the principal gaps. (6, 20)
Evidence density behind technical claims — 6/10. Technical artifacts and named deployments support a real product. Broad outcome claims and cross-generation autonomy claims exceed the detail of their supporting evaluations. (5, 9, 10, 15)
Vendor seriousness
| Criterion | Score |
|---|---|
| Technical seriousness of public communication | 4/10 |
| Technology roadmap integrity | 5/10 |
| Conceptual sharpness | 6/10 |
| Economic accountability | 4/10 |
| Defensibility in an agentic-software world | 5/10 |
Average: 4.8/10.
Technical seriousness of public communication — 4/10. Reproducible analytics receives a concrete explanation. The blockchain regulatory error and absolute service claim undermine the precision expected of a logistics technology vendor. (6, 8, 17, 18)
Technology roadmap integrity — 5/10. The record contains identifiable engineering advances, alongside distinct business and product transitions. The continuation of older autonomy language across those transitions is clearer than the current ownership and delivery of each capability. (2, 3, 4, 5, 16)
Conceptual sharpness — 6/10. Pando understands freight fragmentation and the need for repeatable operational analysis. Its broader autonomy story does not yet articulate a comparably complete treatment of economic uncertainty. (7, 8, 14)
Economic accountability — 4/10. Value measurement appears in ongoing customer responsibilities, and customer savings are reported. Public material does not expose a counterfactual evaluation or a defined remedy for missing economic returns. (11, 12)
Defensibility in an agentic-software world — 5/10. Freight integrations, execution history, and physical loading logic provide assets beyond a conversational interface. Their durable value depends on reducing total operating and improvement effort as agents make document handling and report generation cheaper. (1, 5, 8, 20)
Sources
[1] Domestic Transportation Management System | AI-Driven Multi-Modal TMS by Pando
Domestic Transportation Management System | AI-Driven Multi-Modal TMS by Pando. current; accessed November 4, 2026. Current domestic TMS functions and constraints.
[2] Pando announces strategic restructuring of India and US Business Units to accelerate independent growth
Pando announces strategic restructuring of India and US Business Units to accelerate independent growth. 2025-04-03; accessed November 4, 2026. Separate TMS and US AI business roadmaps.
[3] From Fleet to Freight: Pando.ai and Fleetx.ai Join Forces to Connect the Transportation Lifecycle
From Fleet to Freight: Pando.ai and Fleetx.ai Join Forces to Connect the Transportation Lifecycle. 2026-08-20; accessed November 4, 2026. Acquisition announcement and retained brand; release dated August 20.
[4] Freehand Launches AI Teams for Supply Chain Ops | Freehand
Freehand Launches AI Teams for Supply Chain Ops | Freehand. 2026-02-09; accessed November 4, 2026. Pi lineage and the separate Freehand launch.
[5] US11537977B1 — Method and system for optimizing delivery of consignments
US11537977B1 — Method and system for optimizing delivery of consignments. 2022-12-27; accessed November 4, 2026. Patent specification; see the cost equation and constrained action mechanism.
[6] AI Agent for Freight Procurement | Automated RFQ Management
AI Agent for Freight Procurement | Automated RFQ Management. current; accessed November 4, 2026. Procurement workflow, absolute reliability claim, and onboarding sequence.
[7] Redefine freight negotiation strategies with next-gen intelligence
Redefine freight negotiation strategies with next-gen intelligence. 2025-01-08; accessed November 4, 2026. Tender preparation, scenario choices, and approval flow.
[8] AI Analytics in Logistics: Why Trust Matters More Than Conversation
AI Analytics in Logistics: Why Trust Matters More Than Conversation. 2026-08-12; accessed November 4, 2026. Engineering explanation of reusable analyses and decision context.
[9] Multi-Industry Parts Manufacturer Cuts $200K+ in Freight Costs with AI Agents
Multi-Industry Parts Manufacturer Cuts $200K+ in Freight Costs with AI Agents. undated; accessed November 4, 2026. Undated, anonymous Pi deployment; task and broader outcome claims.
[10] Accuride chooses Pando to drive agility & visibility across their inbound fulfillment operations
Accuride chooses Pando to drive agility & visibility across their inbound fulfillment operations. 2023-06-01; accessed November 4, 2026. Named customer selection and initial deployment scope.
[11] Minimizing cost leakages in logistics: Top 10 freight audit and payment best practices
Minimizing cost leakages in logistics: Top 10 freight audit and payment best practices. undated; accessed November 4, 2026. Reported Accuride savings and payback.
Jobs at Pando. 2026-08-10 and 2026-08-07; accessed November 4, 2026. Public embedded job records: Fullstack Developer, August 10, and Customer Success Manager, August 7, 2026.
[13] Pando CAIQ v4.0.2 security self-assessment
Pando CAIQ v4.0.2 security self-assessment. 2026-01-07 registry renewal; accessed November 4, 2026. Downloadable vendor questionnaire; references include CEK-03.1, IAM-10.1, IVS-06.1, and IVS-09.1.
[14] Excited to announce new (and old) partners!
Excited to announce new (and old) partners!. 2018-04-16; accessed November 4, 2026. Seed-funding announcement and original product rationale.
[15] Zydus Wellness chooses Pando to accelerate digital transformation of their supply chain & logistics operations
Zydus Wellness chooses Pando to accelerate digital transformation of their supply chain & logistics operations. 2021-11-22; accessed November 4, 2026. Deployment announcement; benefits are prospective.
[16] Pando witnessed the market evolve and a surge in logistics AI technology adoption by manufacturers and retailers in 2023
Pando witnessed the market evolve and a surge in logistics AI technology adoption by manufacturers and retailers in 2023. 2023-12-27; accessed November 4, 2026. Autonomy positioning and explicit link to the loading patent.
[17] How technology and innovation are reshaping the world of multimodal logistics
How technology and innovation are reshaping the world of multimodal logistics. 2024-09-10; accessed November 4, 2026. Blockchain promotion and DSCSA claim.
[18] DSCSA Standards for the Interoperable Exchange of Information
DSCSA Standards for the Interoperable Exchange of Information. 2023-09; accessed November 4, 2026. Final guidance, sections VIII–IX: data standard and technological approaches.
[19] Pando Launches Pi – AI Teams for Logistics, Enabling Autonomous Freight Procurement, Planning, and Payments for Global Brands
Pando Launches Pi – AI Teams for Logistics, Enabling Autonomous Freight Procurement, Planning, and Payments for Global Brands. 2025-02-10; accessed November 4, 2026. Launch announcement and the Accuride document-processing example.
[20] How Pando Is Using Agentic AI to Revolutionize Logistics Technology
How Pando Is Using Agentic AI to Revolutionize Logistics Technology. 2025-07; accessed November 4, 2026. Adoption method and illustrated use cases, pages 4–6.