Make each logistics handoff measurable

Evos raised $9 million in pre-seed funding to build autonomous software operators for small and midsized industrial businesses, including freight brokerages. The company said it surpassed $1 million in annual recurring revenue four weeks after launch, while not disclosing valuation or customer count.

Its operators learn how experienced employees make decisions and execute assigned functions inside existing systems. In the disclosed MyFreightWorld example, an operator audited LTL carrier invoices against a TMS, made nearly 3,500 decisions over two months at a reported 97% accuracy, and identified more than $600,000 in billing discrepancies during a five-week period; Evos did not establish recovery or independently verify the accuracy figure.

The model targets organizations where process knowledge sits with long-tenured staff rather than documented procedures. For brokers, the opportunity is lower-touch audit and exception handling without a core-system replacement, but the control question is how the operator behaves when accessorials, classifications, or customer rules fall outside its learned pattern.

Gratis, a drugstore chain with more than 800 stores in Eastern and Southeastern Europe, went live with Solvoyo’s Transportation Control Tower. The deployment puts transportation planning, execution, and operational KPIs on one platform.

Solvoyo describes Solvi as an AI agent supporting the cloud-native control tower, which also uses optimization technology across inventory, supply, production, fulfillment, and transportation decisions. The company reports recommendation acceptance above 95% and planner-productivity gains of up to 12x across its platform, but those figures are not presented as a Gratis-specific result.

For a store network, the value is a shared operating view that can connect plans to execution and KPI review rather than leaving transportation decisions in separate tools. The live deployment creates a measurable test around planner adoption, exception closure, delivery reliability, and transport cost per store replenishment.

Global 3PL Logistics Plus selected Logiwa IO, a cloud-native warehouse management system, to scale enterprise fulfillment. The announcement positions the move around a 3PL environment that must support receiving, storage, order handling, and outbound work across customer operations.

Logiwa describes native AI-based processes intended to drive continuous improvement across those workflows. The selected architecture is cloud-based and designed to connect fulfillment execution with customer and operational data rather than relying only on facility-local controls.

The source does not disclose a customer KPI or implementation result. The operational implication is a platform-standardization decision: Logistics Plus can test whether shared WMS data and embedded intelligence improve onboarding speed, inventory accuracy, labor planning, and order-cycle consistency across accounts.

ShipStation Global was named to FreightWaves’ 2027 FreightTech 100 after combining its parcel software and freight-brokerage network. The company recently added LTL booking and management inside ShipStation, allowing shippers to quote, book, dispatch, track, and invoice pallet shipments alongside parcel work.

ShipStation Intelligence applies AI to carrier and service selection, delivery-risk prediction, fulfillment automation, and international customs and duties. Its freight agents handle dispatch confirmation, pickup and departure updates, delivery milestones, and proof-of-delivery collection; the platform says its network moves more than 3 billion shipments annually.

The combined workflow is aimed at small and midsized shippers that historically split parcel and palletized freight across systems, spreadsheets, and calls. The operational test is whether one decision surface improves carrier choice and visibility without hiding the different service, damage, and claims profiles of parcel and LTL.

Haoxin Holdings described LogiAgent, a 12-module AI logistics platform currently under strategic evaluation. The company said the platform is intended to run alongside existing TMS, ERP, and telematics systems rather than require a full replacement.

The modules cover quoting, document processing, customs and trade compliance, cold-chain intelligence, dispatch optimization, live fleet and reefer data, predictive exceptions, cargo integrity, customer portals, operations assistance, margin intelligence, and door-to-door coordination. The design also includes an owned GPS ingest gateway and an offline-first iOS field application for photos, signatures, and events.

The platform is presented as a functioning, demonstrable product with role-based access, audit logging, tenant controls, and approximately 220 automated tests, but Haoxin has not reached a definitive acquisition agreement. The relevant decision is whether a modular operating layer can reduce integration burden across cross-border workflows without creating a second source of operational truth.

Inbound Logistics’ 21st annual 3PL market survey found growth among providers and customers while operational costs, capacity, compliance, and labor remained central concerns. The report says AI and visibility-enhancing technology have moved firmly into the strategic conversation.

Shippers identified deploying AI as their most important challenge, with 73% citing it, while 3PL respondents ranked AI as the top disruptive innovation and only 50% called it a challenge. The survey also found that 78% of shippers measure 3PL performance with service more important than price.

Those results put AI adoption in a commercial context: technology must improve customer service and continuity rather than merely add another platform. For 3PL executives, the adoption question is whether AI can lower operating friction while preserving the service trust that determines partnership renewal.

Logistics Viewpoints distinguishes visibility from resilience: knowing that a supplier, lane, or facility is disrupted does not establish that the business can execute a different plan. The required alternatives include qualified suppliers, substitute materials, alternate routes, buffer inventory, flexible capacity, and secondary facilities.

The operating model links risk sensing, network models, inventory, transportation, supplier information, and workflow to decision rights. The article emphasizes pre-set thresholds for premium freight, inventory reallocation, supplier substitution, customer prioritization, and temporary process changes, followed by exercises that test whether the alternatives are actually available.

For logistics leaders, resilience is an operating capability with measurable recovery time and option quality, not a dashboard feature. The analysis does not report a deployment KPI, so the practical test is whether a scenario exercise reveals stale capacity, unavailable contacts, or approval delays before a real disruption.

DHL announced a North American expansion of data-center logistics infrastructure to support hyperscale and colocation operators building AI capacity. The plan includes 10 dedicated warehouse sites totaling more than seven million square feet scheduled to go live in 2026.

The facilities are designed for white-glove handling, rack configuration, secure warehousing, and specialized warehouse-to-site transportation. DHL says configuration and testing can move out of active construction zones, while controlled handling protects servers, power modules, networking equipment, and other sensitive components from damage.

The network is built around compressed construction schedules, intricate installation sequences, congested metro areas, and remote greenfield locations. The announcement does not publish a customer KPI, but it establishes a new logistics design point where warehouse configuration and final-mile timing can determine whether a data-center project remains on schedule.

NVIDIA’s internal IT and Operations team built AI Planner to coordinate a rack-scale AI supply chain involving GPUs, switches, NICs, cables, and server platforms. The company reports reducing what-if planning cycles from about one day to under ten minutes.

AI Planner combines NVIDIA cuOpt decision optimization with Nemotron-based AI agents and NVIDIA NIM. The system generates many scenarios for changing demand and disruption conditions, while the optimization layer handles constraints and the agent layer helps users interact with the planning process.

NVIDIA reports a 6x planner-productivity gain and maintained on-time delivery while demand expanded, but the case is an internal deployment with company-reported results. Its significance is the frequency of scenario evaluation: planners can compare alternatives before supply constraints become physical shortages.

ChainIT partnered with excelerated to combine digital verification and compliance workflows with procurement, SAP, source-to-pay, and supply-chain transformation expertise. The partnership targets supplier onboarding, contractor qualification, compliance monitoring, and final-mile operational visibility.

ChainIT’s platform collects and verifies identities, business authority, credentials, certifications, insurance records, sanctions data, and other compliance evidence. The workflow is designed to replace fragmented emails, spreadsheets, portals, and document repositories with automated monitoring and auditable records.

For logistics operators, onboarding quality affects who can move freight, enter sites, and remain eligible as requirements change. The announcement does not establish a logistics KPI result, so the practical value is faster qualification with fewer expired documents and clearer accountability for exceptions.

TrusTrace introduced a data and intelligence platform for brands managing sustainability and due-diligence requirements across complex supply chains. The rollout began with Supply Chain Intelligence after pilots with several global brands.

The platform combines supplier compliance, supply-chain intelligence, and product compliance on a shared data foundation. It supports lot-level traceability, AI-assisted chain-of-custody verification, natural-language queries, automated evidence collection, supplier onboarding, monitoring, and corrective-action workflows; suppliers can submit data once in different languages and formats.

The operating implication is less repeated data collection and faster response when a regulation or sourcing vulnerability changes. TrusTrace does not publish a quantified customer outcome in the accessible report, so buyers should test validation quality, supplier participation, and corrective-action closure.

UK 3PL Miniclipper Logistics completed a fourth go-live with Thomax at its Dunstable EAV warehouse. Four of the company’s six sites now run on the platform, with the remaining rollout due in the first quarter of 2027.

Miniclipper selected a configurable WMS with e-commerce and kitting support, ERP integration, and site-by-site implementation. The Burton-on-Trent fulfillment operation processes an order roughly every five seconds at peak and has up to a month of forward order visibility, while the first site went live in under four months.

The phased approach is an onboarding pattern for 3PLs that need new customer capability without risking service continuity. Miniclipper reports faster and more effective picking, but the source does not provide a controlled before-and-after KPI comparison across all sites.

United States Cold Storage introduced automated appointment scheduling through FourKites’ AI agent Alan across a network of 38 facilities. The use case addresses inbound delivery and outbound dispatch coordination in a regulated, temperature-sensitive environment.

During an eight-week pilot covering more than 600 shipments, Alan achieved an 87% success rate in booking appointments and 96% accuracy in obtaining requested delivery dates, according to the operator. It processed more than 150 scheduling requests simultaneously and maintained audit trails while working through email, web interfaces, and phone interactions.

The reported productivity equivalent was 36 to 40 hours during the pilot, but the results are operator-reported and not a promise for every facility. The deployment demonstrates a bounded starting point: automate appointment work first, then use the control-tower context for broader exception decisions.

Qued received FreightWaves’ inaugural AI Excellence in Supply Chain award for automating load appointment scheduling for brokers, 3PLs, and carriers. The workflow is intended to reduce late deliveries, detention, and OTIF penalties caused by appointment friction.

The platform integrates with existing TMS data, uses generative AI and machine learning to interpret facility-specific email feedback, and adapts to carrier preferences and location behavior. Qued says about 95% of appointments are confirmed autonomously and that its system scores submissions in real time for likely confirmation or rejection.

A disclosed deployment managed thousands of appointments across hundreds of facilities, with a reported confirmation rate above 95% and exception rates below 5%. The metrics are company-reported, so the implementation question is whether performance holds across facility types and appointment rules.

The AI infrastructure boom is driving trade in semiconductors, networking equipment, servers, cooling systems, and power hardware from Asian manufacturing centers to data centers. FreightWaves reports that AI-related goods accounted for 42% of global goods-trade growth in 2025 and about 76% in the first quarter of 2026, citing DHL-commissioned research.

The logistics challenge is not simply volume: equipment is oversized, high-value, sensitive, and often moves through multiple manufacturing, testing, assembly, and installation stages. Servers that once cost roughly $100,000 to $250,000 can now reach $1 million to $3 million, increasing the need for controlled handling, customs expertise, and visibility.

DHL estimates the data-center logistics market at about $23 billion in 2025 and at least $35 billion by 2030. For inbound operators, the growth case is paired with higher exposure to cargo damage, installation timing, multimodal handoffs, and security requirements.

Mujin is demonstrating an intelligent palletizing cell for bottled and canned products at PACK EXPO 2026. The company says its Mantis Gripper handles difficult multipacks, open-top cases, and heavy two-liter cases at speeds up to 1,000 cases per hour.

MujinOS maintains a live digital twin of robots, products, pallets, and connected equipment. MujinPack evaluates product geometry, weight, fragility, support surfaces, tipping risk, and gripper access before generating pallet patterns; new SKUs can be ingested in minutes without an engineer, according to the company.

The system is aimed at beverage manufacturers and warehouse distributors facing SKU proliferation and retailer demands for store-ready pallets. The source describes a demonstration and prior deployments in several convenience-format grocery chains, not a universal throughput guarantee.

Wurth Australia appointed Dematic to automate order fulfillment at distribution centers in Yatala and Keysborough. The network serves more than 22,000 products and over 56,000 active customers across industrial and trade sectors.

The design combines AutoStore goods-to-person storage with manual picking zones and downstream carton sealing, labeling, and dispatch sortation. Dematic warehouse-control software will coordinate activity between Wurth’s WMS and the mixed automated/manual equipment.

The choice acknowledges that a broad product range cannot be handled by one storage mode. Wurth expects faster, more accurate orders and lower reliance on manual work, but the announcement does not publish baseline or target KPI values.

Numina Group launched Batchbot 2.0, a warehouse picking system combining high-capacity autonomous mobile robots, voice-directed picking, and RDS intelligent orchestration. The company says carts can carry more than 35 orders per trip and support over 300 picks per operator hour.

The system uses AMRs and warehouse-execution software to increase batch density and reduce operator walking. NorthShore Care Supply reported nearly 50% staff reduction and another 25% to 30% efficiency gain after a software and AMR upgrade, but those figures are customer-reported and specific to its operation.

The deployment pattern is a coordinated labor-and-robot workflow rather than a lights-out facility. For a 3PL or distributor, the important measurement is whether denser batching improves picks per labor hour without increasing mispicks, congestion, or replenishment imbalance.

Norwegian fashion e-commerce retailer VILLOID selected StrongPoint to design and install a Pio facility built on AutoStore technology. The approximately NOK 17 million project is scheduled for completion in the second quarter of 2027.

Pio combines high-density automated storage with goods-to-person fulfillment for growing e-commerce businesses. The design is intended to increase capacity and operational efficiency while moving inventory to operators rather than sending operators through the storage grid.

The decision is an infrastructure commitment rather than a measured live result. VILLOID’s operating test will be whether dense storage and goods-to-person sequencing support peak demand without sacrificing order accuracy, replenishment speed, or resilience to apparel-SKU variation.

Ingram Micro described a three-year strategy that puts AI and automation into the operating infrastructure of its distribution business. The company positioned its Xvantage platform as an intelligent operating system for sales, quoting, pricing, service, and supplier-customer connections.

Xvantage combines more than 400 AI and machine-learning models with hardware, cloud subscriptions, recommendations, pricing, order tracking, billing, and other functions. Ingram Micro also said the platform contains more than 42 million lines of code and connects logistics, supply-chain, financial, marketing, and IT-asset-disposition services.

The distribution implication is cross-functional orchestration: decisions about price, inventory, fulfillment, and lifecycle services can be made with shared context. The source describes strategy and platform scale, not a single causal KPI outcome, so adoption should be evaluated by process-level conversion and margin measures.

Vecna Robotics raised $31 million to expand its autonomous case and pallet movement systems. The company said demand for CaseFlow more than doubled year over year since its 2025 launch and that it is expanding deployment teams and capabilities.

CaseFlow coordinates autonomous forklifts and tuggers through Vecna’s Pivotal orchestration platform. Modern Materials Handling reported a GEODIS deployment that doubled picking throughput; GEODIS also described new pickers reaching 200% of the prior performance level while learning the robot workflow instead of first learning to drive equipment.

The reported result illustrates a fulfillment model in which robots carry the movement burden while people perform picking decisions. Because the numbers come from a specific deployment, operators should test whether the same gain survives different slotting, product, travel, and safety conditions.

Waabi and Warrior Logistics entered a commercial partnership to move recurring customer freight between Dallas and Houston using trucks equipped with Waabi Driver autonomous-driving technology. Warrior, an asset-based carrier headquartered in Coppell, Texas, integrated the trucks into its network.

The companies say the arrangement is intended to add freight capacity, improve fleet utilization, and support service reliability as demand grows. Waabi has described driverless operations as a future target; the accessible report does not establish that the Dallas-Houston service is already driverless, so the present state should be treated as technology-assisted commercial freight with a future autonomy path.

The corridor provides a bounded operating environment for evaluating handoffs, safety performance, dispatch control, and customer acceptance. For outbound networks, the question is whether autonomy can add dependable capacity without creating new exception work around weather, terminal access, or human intervention.

Predictiv AI launched Flockentis, an AI-powered fleet decision engine, with a proof of concept underway at a commercial fleet operator in the United Kingdom. The system is designed to make pre-dispatch decisions using cost, delivery commitments, regulatory constraints, commercial pricing, and estimated carbon impact.

Flockentis evaluates more than ten considerations, including distance, driver workload, legal hours, vehicle health, capacity, cargo compatibility, hazardous-goods requirements, safety, and emissions. It can process up to 5,000 jobs per planning cycle and return three alternative plans prioritizing cost, service, or environmental impact, with reasons for jobs that cannot be completed.

The company explicitly says the product is not being represented as a deployed autonomous swarm-agent system without further validation. The current value proposition is explainable multi-objective planning that exposes tradeoffs before dispatch, not self-driving execution.

Freight Technologies launched AI POD Validation in its Fleet Rocket TMS for warehouses, carriers, and customers handling inconsistent proof-of-delivery formats. The module reads and verifies POD documents and flags those that do not meet validation requirements.

The workflow uses computer vision, machine learning, OCR, signature and stamp detection, spatial document understanding, and bilingual English-Spanish interpretation. It matches pickup numbers, bills of lading, carrier names, and invoice data against the shipment record, then returns Approved, Review, or Rejected with a confidence score and supporting evidence.

POD validation closes the shipment record and supports carrier payment, so the operational outcome is less manual reading and fewer mismatched approvals. Freight Technologies says the capability is available to Fleet Rocket Enterprise customers and may be extended to invoices and bills of lading.

Upstream is an inventory-intake and processing solution for returns, overstock, and challenged inventory. It identifies and triages each item, then routes it to a recovery channel such as eBay, Amazon, Shopify, Vendidit’s marketplace, or bulk liquidation.

The workflow classifies condition into paths such as direct-to-consumer resale or lower-quality liquidation, while dashboards show item status, channel, and sale timing. Upstream can layer onto an existing WMS with minimal integration, according to the product page.

The design shifts reverse logistics from a storage problem to a recovery-value decision. The page does not provide a customer KPI or measured recovery uplift, so a 3PL should validate grading accuracy, days-to-list, resale yield, and the cost of each touch.

ReturnStack describes a managed service in which merchants send returns to the company for inspection, resale, recovery, and reporting. Its commercial model combines a flat fee with a resale commission.

The service connects the store to a physical returns operation rather than asking the merchant to build inspection and disposition capability in-house. The accessible page emphasizes outsourced handling and reporting but does not disclose automation architecture, volume, or recovery metrics.

For smaller merchants and 3PL customers, the decision is whether outsourcing reduces reverse-logistics complexity without losing visibility into item condition, channel selection, and recovered margin. The dated page is being used as a sparse-category fallback because current returns coverage is limited.

Online Store News reports that mid-year 2026 online return rates averaged 22.4%, with apparel and footwear above 35%, increasing pressure on brands and 3PLs to move returned goods back into saleable inventory. The article frames reverse logistics as a margin and customer-retention problem rather than a back-room cost.

The described workflow spans receipt, inspection, grading, restocking, liquidation, accounting, returns-management software, warehouse automation integrations, and carrier-rate optimization. It cites Loop Returns’ $65 million Series C extension and Happy Returns’ expanded drop-off network, while the outlet’s figures and company announcements require independent validation before use as a forecast.

The operational implication is cycle time: a returned item that waits eight days for grading and relisting may miss the customer’s repurchase window, according to January Digital’s Sarah Engel. The article does not establish a controlled automation outcome, so the useful KPI is time from receipt to disposition and relisting by product category.

The Hackett Group introduced AI World Class Supply Chain benchmarks that model improvements across planning, inventory, manufacturing, warehousing, transportation, and returns. The announcement says process costs across plan-to-deliver could decline 12% to 17% and staffing requirements 15% to 24% in the modeled scenarios.

The benchmark also reports potential reductions of up to 59% in expedited orders, 52% in forecast error, and 41% in inventory carrying costs. These are benchmark-model outputs based on peer medians and AI World Class scenarios, not a disclosed customer deployment.

The value of the work is diagnostic: it gives supply-chain leaders a way to test whether an AI program changes connected process outcomes rather than isolated task efficiency. The risks are equally clear if modeled improvements are mistaken for guaranteed savings.

Logistics Viewpoints describes a shift from episodic prompting toward persistent decision workflows that maintain business context, changing state, decision logic, approvals, and outcomes. The argument is that supply chains cannot operate as a sequence of disconnected conversations.

The proposed workflow connects detection, context assembly, analysis, option generation, recommendation, approval, execution, and feedback. ERP, planning, TMS, WMS, visibility, risk, and network platforms remain systems of record, while a decision layer assembles context and controls actions through permissions, thresholds, and escalation.

The operating improvement is learnable decision history: recommendations, overrides, actions, and results can be reviewed to improve rules, models, prompts, and autonomy thresholds. This is an architecture pattern rather than a quantified deployment result.

Logistics Viewpoints argues that model capability is only one component of operational AI. Sustainable value requires authoritative context, governed tools, permissions, observability, and a reliable path from recommendation to action.

The analysis distinguishes prediction, optimization, generative AI, vision, and agents as different patterns with different controls and measures. It identifies orders, inventory, rates, carrier status, warehouse state, supplier records, and policy documents as sources that must be resolved into current operational context.

Once an AI system can create a shipment, change an order, release inventory, contact a carrier, or approve an exception, the architecture needs financial limits, approval gates, idempotency, retries, rollback, and recovery measures. The recommended metrics are decision latency, manual context gathering, exception closure, overrides, tool failure, and business outcome.