Perspectives

Last month, [a Cursor agent deleted a production database at Pocket OS](https://thenewstack.io/ai-agents-credential-crisis/). A few weeks before that, [Replit's agent did the same thing to another company during a code freeze](https://www.reddit.com/r/replit/comments/1sxorvf/ghosted_by_replit_support_the_agent_wiped_my/). Every time one of these stories hits Hacker News, a wave of CTOs quietly ban AI agents from touching prod. We understand the reflex. But we think it's the wrong lesson. We've been running a Claude Code agent against our production FMS at Wove for months. It reads our production database, inspects live shipments, reconstructs why cross-validation failed on a specific customs entry, and pulls the actual PDFs our extraction pipeline was working from. Turnaround on customer-reported bugs has gone from hours to minutes. It has never deleted anything. It structurally cannot delete anything. This post is about the pattern we built to make that true and why we think "give the agent raw psql" and "ban the agent from prod entirely" are both wrong answers. ## The failure mode nobody talks about The Pocket OS story gained traction because the agent deleted a database. That's the dramatic version. The far more common failure mode, the one that nobody writes articles about, is quieter: > **Customer:** "My shipment is stuck. Can you look?" > **Engineer:** pings backend dev on Slack > **Backend dev:** sshes to a bastion, opens psql, runs three queries, tails a log file, pieces together what happened, writes a summary > **Engineer:** forwards summary to customer **Elapsed:** 45 minutes to 4 hours, depending on backend dev's calendar. Every SaaS company runs this loop dozens of times a day. It burns your most expensive engineers on repetitive lookups. It gates every support conversation on backend eng availability. And it's the reason "we'll get back to you" has become the default customer experience. The naive AI fix, "let the agent run psql for us," is what created the Pocket OS story. So teams bounce back to the manual loop and accept the slowness as the price of safety. There's a third option. It just requires architecture, not vibes. ## The pattern: agent-safe production access The core idea is a small admin-gated HTTP surface (we call ours `/_agent_query/*`) that exposes exactly the read operations an agent needs to debug production, and nothing else. Not psql. Not the shell. Not the deployment pipeline. Not write endpoints. Four properties make it agent-safe: **1. Read-only enforced at the database layer, not the application layer.** Every query the agent runs executes inside a Postgres transaction that begins with: ```sql SET LOCAL transaction_read_only = on; SET LOCAL statement_timeout = 5000; ``` If the query builder above it has a bug, if the agent gets prompt-injected, if a malicious payload somehow slips through, Postgres refuses the write. The safety property isn't "we hope the agent doesn't do something bad." It's "the database wil

How we let Claude debug our production FMS in minutes (without giving it a footgun)

On February 20, 2026, the Supreme Court struck down IEEPA tariffs in [Learning Resources v. Trump](https://www.supremecourt.gov/opinions/25pdf/24-1287_4gcj.pdf). By Sunday night, we had a free tool live at [refunds.wove.com](https://refunds.wove.com) that lets any importer upload their entry data and see exactly what they're owed — entry by entry, with the recovery mechanism (PSC, Protest, or CIT Litigation) for each one. A lot of tools came out this week claiming to calculate refund exposure. Most of them are applying a flat rate to a customs value and calling it a calculation. Ours is not. Here's what it actually took to build a real tariff refund calculator. ## The Problem Nobody Talks About: Tariff Rates Changed 30+ Times The first thing you need to understand is that IEEPA tariff rates were not static. Over the past year, rates changed by country, by HTS chapter, and by effective date — 35 times in total. China went from 20% to 145% and back down. Canada and Mexico had their own rate schedules. Most countries were subject to a baseline 10% with specific carveouts. Every other tool we've seen applies a single flat rate. That's not a refund calculator. That's a rough estimate that will be wrong for the majority of entries. Getting the calculation right means knowing exactly which rate was in effect on the exact date of each import entry. To do that, we had to ingest all 35 tariff schedules — parsing Federal Register notices, reconciling conflicting effective dates, and mapping every rate change to the correct HTS chapter, country of origin, and date window. That alone required a full day of engineering work. ## We Had to Build a PDF Parser from Scratch The HTS code system doesn't just live in spreadsheets. Chapter notes — the legal language that determines how goods are classified — are published as PDFs by CBP. To build a classification engine that's actually defensible, you need to understand those notes. We built a custom PDF parser to extract chapter notes from each HTS chapter PDF. This is the kind of infrastructure work that nobody sees but that makes the difference between a tool that produces a number and a tool that produces a defensible number. When CBP audits a refund claim, they don't just look at what rate you applied. They look at whether your HTS classification was correct in the first place. A misclassified entry doesn't just affect your refund — it creates audit exposure. Our parser is the foundation of the classification accuracy layer that sits underneath the refund calculator. ## Chapter 99 Overlay Codes Are Where Most Tools Fall Short IEEPA tariffs weren't applied directly to standard HTS codes. They were applied via Chapter 99 overlay codes — the 9903.88.xx series for China, with separate series for other countries. An entry might have a standard HTS code of 8471.30.0100 (laptop computers) plus a Chapter 99 overlay of 9903.88.03 representing the specific IEEPA tranche that applied. Getting the refund calculation right mea

We Built the First Real IEEPA Tariff Refund Calculator in 72 Hours. Here's What It Actually Took

Product Updates

If you've ever worked with HTS Chapter 99, you know it's a special kind of chaos. Unlike most HTS chapters that classify physical products, Chapter 99 is where the exceptions live: temporary entries, antidumping duties, Section 301 tariffs, and a constellation of special provisions that can apply to thousands of different products. Recently, we took on the challenge of parsing the notes from Chapter 99 documents and making them machine-readable. What we discovered: a single note can govern the treatment of thousands of HTS codes, with logic so complex that even experienced compliance professionals need to read it multiple times. ### The Problem with Chapter 99 Notes Take Note 16(n) as an example. Here's what it says: > "The rates of duty in heading 9903.81.91 apply to: (1) all imported derivative iron or steel products classified in the following subheadings of the HTSUS, including products admitted into a U.S. foreign trade zone under privileged foreign status before August 18, 2025, unless the derivative iron or steel product was processed abroad from steel melted and poured in the U.S." Straightforward? Not quite. This single note applies to thousands of HTS codes across dozens of chapters. Here's what that actually looks like: Each of those numbers represents a specific product classification. This note contains: - Temporal logic (the August 18, 2025 cutoff) - Conditional exceptions (foreign trade zones, privileged foreign status) - Material origin requirements (steel melted and poured in the U.S.) - Cross-references to hundreds of "specified HTSUS provisions" scattered throughout the tariff schedule For a customs broker trying to classify a steel product, this means: 1. Identifying which base HTS code applies 2. Determining if Note 16(n) creates an additional classification under heading 9903.81.91 3. Checking if any of the exceptions apply 4. Tracking down which specific provisions are covered 5. Documenting the entire chain of reasoning Miss one condition, and you've misclassified the shipment. ### Why This Matters for Automation Traditional tariff calculation systems struggle with Chapter 99 because the notes don't follow predictable patterns. They're written in legal language, reference other parts of the HTS indirectly, and change frequently through Federal Register notices. Our ingestion engine had to handle: - **Scope expansion**: Parsing "all imported derivative iron or steel products" and mapping it to actual HTS codes - **Exception logic**: Understanding nested conditions and carve-outs - **Temporal awareness**: Tracking effective dates and applicability windows - **Cross-referencing**: Resolving references to hundreds of specific provisions across the entire HTS The result? We can now automatically identify when Note 16(n)—or any other Chapter 99 provision—applies to a given shipment, without requiring manual lookup. ### What We Built Our approach combines document parsing with trade intelligence: 1. **Structured extraction**: We 

When One Note Covers Thousands of Tariff Codes: Decoding Chapter 99

We're making Wove's tariff calculation engine available as a standalone API. If you're building landed cost estimators, procurement tools, quoting systems, or trade compliance software, you can now embed duty rate lookups and tariff calculations directly into your application. ### What It Does The Tariff API handles the complexity of modern trade policy so your application doesn't have to. - **Duty rate lookup** - Query by HS/HTS code and get current rates for US, EU, Canada, and Australia. Returns MFN rates, preferential rates, and additional duties (Section 301, 232, IEEPA, etc.) with a single request. - **Stacked tariff calculation** - Automatically calculates combined duty impact when multiple tariff programs apply to the same product. If you're importing aluminum components from China, the API returns Section 232 + Section 301 + any applicable IEEPA rates, not just the base MFN rate. - **Semantic search** - Don't know the HS code? Search by product description using natural language. "laptop computer" returns ranked HTS codes with similarity scores so you can classify products programmatically. - **Historical rates** - Look up what tariff rates were on any past date. Critical for refund calculations, audit preparation, or analyzing how policy changes affected your costs. - **FTA eligibility** - Specify origin and destination countries to see available free trade agreement rates and calculate duty savings from preferential programs. ### Why We Built This 2025 saw 50 modifications to the U.S. tariff code—half from emergency economic powers (IEEPA) that are now under Supreme Court review. Companies building procurement systems, marketplace checkout flows, or ERP integrations kept hitting the same wall: tariff calculation is too complex to maintain in-house. Rate lookups aren't enough. You need to know which additional duties apply, how they stack, whether exclusions exist, and what trade agreements might reduce the rate. And you need this to stay current as policy changes. We built Wove's tariff engine to power our own landed cost calculations for freight forwarders and importers. We're now making it available as infrastructure that other platforms can build on. ### What Makes Us Different - **Policy-aware, not just data lookup** - The API understands tariff program interaction. It knows that Section 232 steel tariffs apply based on component percentages, that Chapter 99 exclusions override base rates, and that certain products are subject to multiple overlapping tariff actions. - **Current by default** - Rates update automatically as tariff schedules change. You don't maintain rate tables or parse Federal Register notices. - **Built for embedding** - Clean REST endpoints, predictable response formats, and customer-specific overrides mean you can integrate this into quoting flows, procurement dashboards, or customs filing systems without building your own tariff infrastructure. ### Use Cases We're Seeing - **Landed cost estimators** - E-commerc

Wove Tariff Calculator Now Available via API

The U.S. Supreme Court could issue a ruling any day that invalidates over $129 billion in tariffs collected under the Trump administration's use of emergency economic powers. For the 301,000 U.S. importers who've paid these levies across 34 million entries in 2025, this could mean substantial refunds—but only if you can prove what you paid and move quickly. The last time the federal government processed mass tariff refunds (the harbor maintenance fee case in the late 1990s), it took years. Companies that had their documentation ready got paid faster. Those that had to reconstruct records from fragmented systems waited. This time can be different, but you have to prepare now. ### What's at Stake The Supreme Court is reviewing the legality of the "reciprocal tariffs" and "fentanyl tariffs" imposed on most U.S. trading partners throughout 2025. A federal appeals court already found that the Trump administration overstepped its authority by invoking the International Emergency Economic Powers Act (IEEPA) to implement these tariffs. The scale is staggering. According to [data from the Cato Institute tracking IEEPA tariffs](https://www.cato.org/ieepa): - $129 billion collected as of December 10, 2025 - 34 million entries filed under IEEPA authority - 301,000 importers potentially eligible for refunds - 60% of all tariff revenue from trade enforcement actions in 2025 If the Supreme Court agrees with the appeals court, importers could be eligible for refunds on duties paid since the tariffs took effect. The reciprocal tariffs alone account for 61% of total IEEPA revenues meaning the average importer has significant exposure. The problem? According to customs experts, importers will need to "go through some sort of due-diligence process" to prove what they paid and what they're owed before filing claims with U.S. Customs and Border Protection. ### The Documentation Challenge No One's Ready For Here's what makes this particularly difficult: **2025 has been the most volatile tariff year in modern U.S. trade history**. [Fifty modifications to the U.S. tariff code were made in 2025]((https://www.cato.org/ieepa)—half of them from IEEPA actions. That means: - Tariff rates changed constantly throughout the year - The same HTS code could have different rates depending on when you imported - "Fentanyl tariffs" on China, Mexico, and Canada operated differently than "reciprocal tariffs" on other countries - Your entries from January look completely different than entries from October Now you need to untangle which of your 34 million collective entries were subject to which IEEPA tariff, at what rate, on what date. **For each refund claim, you need to link multiple data points:** - Commercial invoices for each shipment - Entry summaries showing tariff classifications - Proof of duty payment amounts - HTS codes and the specific tariff rates applied at the time - Country of origin documentation - Timeline showing when payments were made under which tariff program **The

Supreme Court Tariff Decision: Why Your Refund Documentation Needs to Be Ready Now

The team at Foundation Capital recently [articulated the idea of context graphs](https://foundationcapital.com/context-graphs-ais-trillion-dollar-opportunity/) — systems that capture the reasoning connecting data to action. Their work resonated deeply with what we see every day in global trade operations. Why was this rate approved? Why was this HS code used instead of another? Why did this shipment require an exception when the last one didn’t? In global trade operations, these questions come up constantly. And the answers almost never live in your systems. They live in inboxes, spreadsheets, Teams threads, and people’s heads — and once the action is taken, they disappear. For decades, enterprise software has focused on one thing: **state**. Systems of record tell us what’s true right now — the final rate, the booked shipment, the cleared entry, the paid invoice. They’re very good at that. But in global trade, what breaks operations isn’t missing state. It’s missing **context**. At Wove, we believe this missing layer is the biggest unsolved problem in global trade software. It’s what led us to think deeply about context graphs. ### The problem isn’t automation — it’s memory Most trade operations today look like this: • An email arrives with a rate request, document, or exception • A human interprets it, applies judgment, and makes a call • The final outcome gets written into a system • The reasoning behind that outcome is lost Multiply this by millions of shipments and years of operations, and you get an industry that constantly re-learns the same lessons. This is why automation alone doesn’t compound in trade. If every edge case resets the system back to human judgment, with no memory of how similar cases were handled, progress stalls. ### State vs. context: the two clocks of trade operations One way to think about this is that every organization runs on two clocks: • A **state clock**: what’s true right now • An **event clock**: what happened, in what order, and why Global trade systems are excellent at the first. They are almost blind to the second. A TMS shows that a shipment was booked. It doesn’t show the tradeoffs that led to that booking. A customs system shows an HS code. It doesn’t show the ambiguity that was resolved to select it. A pricing system shows a rate. It doesn’t show the exception, approval, or precedent behind it. That second clock — the event clock — is where real operational intelligence lives. ### What a context graph actually is (and isn’t) A context graph is often described as “a graph of decisions,” but that framing can be misleading. It’s not a static knowledge graph. It’s not a document index. It’s not a memory bolted onto an agent. A context graph is the accumulation of **execution-time traces** — records of how real trade actions were carried out, with context preserved. Each trace captures things like: • What inputs were considered (documents, emails, prior shipments) • What rules or policies applied • What excep

Context Graphs and the Missing Memory of Global Trade

Each year, Australia enforces strict **Brown Marmorated Stink Bug (BMSB)** seasonal measures to protect its agriculture and environment from this invasive pest. For freight forwarders, importers, and logistics providers, the 2025/26 season brings **new rules, new risks, and new operational challenges.** Here’s how Wove helps your team stay compliant while keeping cargo moving. ### What’s Changed for 2025/26? The Department of Agriculture, Fisheries and Forestry (DAFF) has announced several important updates to this season’s BMSB program: - New emerging risk countries: Republic of Korea and Japan are now on the list, joining China and the UK. - Airfreight inspections: Target high-risk goods shipped as airfreight from the USA and China will face random inspections. - New offshore treatment option: Ethyl Formate is now approved as a treatment method (offshore only). - Policy changes for in-transit and rolled cargo: A new application process applies, creating extra paperwork for affected shipments. For importers, these changes mean more vigilance at the RFQ stage, greater coordination with shippers, and the need to validate new treatment methods. ### Why BMSB Matters The Brown Marmorated Stink Bug (BMSB) is one of the world’s most invasive agricultural pests. Originating in Asia, it has spread across the US and Europe, causing billions in crop losses. BMSB is a notorious “hitchhiker,” hiding in: - Containers - Vehicles and machinery - Break bulk cargo - High-risk manufactured goods (iron, steel, vehicles, wood, ceramics, etc.) Its greatest risk period is the Northern Hemisphere winter (Sept–Apr) when the bug seeks shelter in shipments bound for Australia. Once established, it is nearly impossible to eradicate, making strict biosecurity enforcement critical. ### Target Countries for 2025/26 DAFF has identified 41 target risk countries for mandatory treatment (e.g., Italy, Germany, USA, Turkey). Additionally, **China, Japan, and the Republic of Korea** (along with the UK) are classified as **emerging risk countries**. Goods from these nations may be subject to random inspection, and vessels calling these ports face **heightened surveillance**. ### Target Goods - **Target High-Risk Goods (Mandatory Treatment)**: Vehicles (HS 87), machinery (HS 84), electrical equipment (HS 85), iron and steel (HS 72–73), wood (HS 44), ceramics (HS 69), glass (HS 70), and more. - **Target Risk Goods (Random Inspection)**: Plastics (HS 39), paper (HS 48–49), rubber (HS 40), chemicals (HS 28–29, 38), etc. Non-compliant shipments risk **re-export, costly onshore treatment, or refusal of discharge**. ### Wove for RFQs: Spotting BMSB Risk Upfront BMSB compliance starts before the shipment is even booked. With Wove’s Email Copilot, forwarders can identify risks immediately at the RFQ stage. - When an RFQ includes an HS Code (e.g., 8701 – agricultural tractors), Wove automatically flags if it falls under BMSB seasonal rules. - The system highlights treatment requirements, deadl

How Wove Helps Australian Logistics Companies Stay Ahead of BMSB Measures

Logistics professionals are buried under a mountain of unstructured documents. These documents include rate sheets and tariffs, bills of lading, customs forms, invoices, packing lists, and RFQs just to name a few. Historically, companies have turned to OCR (Optical Character Recognition) and RPA (Robotic Process Automation) platforms like UiPath to tame the chaos. But those tools weren’t built for freight, and the "automation" many promised often ended up creating more headaches than they solved. However, Wove is different. Wove is built entirely on Large Language Models (LLMs) and designed to specifically automate complex logistics workflows. Here's why that matters. ### 1.🧾 Document Complexity: Wove Doesn’t Break on Freight-Specific Formats |Feature| Wove | OCR & RPA | |----------|----------|----------| | Handles scanned, skewed PDFs and emails? | ✅ Yes, trained on noisy logistics inputs | ❌ Often fails unless heavily pre-processed| | Works on complex tables, rate matrices, and tariffs? | ✅ Yes, near-100% accuracy on nested rate sheets | ⚠️ Requires manual tuning or fails silently | | Handles shipping packet “bundles” like pre-alerts? | ✅ Yes, extracts multiple docs from zip or email threads | ❌ Needs custom scripting or third-party tools | Wove’s AI has been trained specifically on logistics documents — not just invoices, but multi-document pre-alert packets with inconsistent formats and ambiguous labels. ### 2. 🔁 Workflow Automation: Purpose-Built for Freight Forwarding OCR and RPA tools like UiPath are general-purpose. That means: - You need dedicated developers to build and maintain scripts - The tools don’t “understand” CargoWise, master/house shipments, or customs declarations - Creating custom fields that map into your TMS might take months rather than seconds Wove, on the other hand: - Comes with prebuilt integrations (CargoWise, email, shared folders) that can be customized to your organization's specific wants and needs instantly - Supports house/master shipment mapping - Recognizes logistics-specific field labels and units (e.g. CBM, freight terms, HS codes, port codes) – Enables users to create custom fields and map them into a TMS in just seconds No configuration needed. Just drop your files and go. ### 3. ⚙️ Maintenance & Scalability: Wove Learns While You Work | Feature | Wove | OCR & RPA | |----------|----------|----------| | Learns from corrections to improve future accuracy? | ✅ Built-in feedback loop + auto-training | ❌ Manual retraining and rule editing required | | Requires technical resources to maintain?| ❌ No — business users can operate| ✅ Yes — DevOps and RPA specialists needed | | Time to first value| ⚡ Hours| 🐢 Weeks to months (plus consultant fees) | With Wove, freight forwarders, customs brokers, BCOs, and LSPs can train the system passively — no prompt engineering or ML ops required. Our system gets smarter every time you use it. ### 4. 📈 Accuracy & Cost Most OCR/RPA pipelines are brittle and require human cle

Wove vs. OCR + RPA Tools: What’s the Best Choice for Logistics Document Automation?

If you are a freight forwarder, odds are your team lives inside three core systems: email, Excel, and a TMS. Since launching Wove, we have made a concerted effort to integrate with all three. Today, however, we want to share why we have chosen to integrate Wove so deeply with one TMS in particular: CargoWise. It is the backbone of your quoting, booking, invoicing, and customer service operations. But for all its power, CargoWise was built in a different era—one where manual data entry was the norm and automation was the exception. At Wove, we believe that era is ending. That is why we are building one of the deepest, most native integrations with CargoWise on the market. This is not just a checkbox on a features list. It is a strategic investment to help our customers eliminate manual work and unlock speed, accuracy, and scale. CargoWise powers over 16,000 logistics providers worldwide and has built one of the most robust ecosystems in the industry through its Global Partner Network. That scale—and the diversity of workflows it supports—is exactly why we are going deep. We are also actively collaborating with several consulting firms within the CargoWise Partner Network to make implementation even smoother and to ensure our solutions adapt seamlessly to the specific needs of each forwarder. ## **Why Go Deep?** ### 1. It Is Where the Work Happens Rate uploads, invoice reconciliation, quote generation—this work flows through CargoWise. If we want to automate it, we need to integrate where the work starts and ends. We are not sending exports or generating CSVs. We are: • Reading rate sheets and shipping documents directly from emails • Parsing them with near-perfect accuracy • Pushing the data straight into your CargoWise instance via our eAdaptor integration. Whether you are uploading contract rates or reconciling invoices, we ensure the data lands exactly where it needs to—no copy-paste required. ### 2. Freight Forwarders Deserve Native Automation Most "integrations" are not really integrated. They send alerts or generate files you still have to upload manually. Wove is different. We are building a true operational layer that connects to CargoWise like an internal system, not a third-party workaround. If your team should not be rekeying it, we will handle it. ### 3. Speed Is a Competitive Advantage Your competitor is quoting in minutes. If your team is buried in Excel and email threads, you lose deals. By integrating deeply with CargoWise, Wove helps: • Cut quote generation time by 80 to 90 percent • Eliminate backlogs caused by timezone mismatches or weekend delays • Reduce costly errors caused by manual entry. We have also launched full API connectivity, making it even easier to embed Wove directly into your existing systems and workflows. Implementation takes just minutes, not months—so you can realize value almost immediately. ### 4. This Is Just the Beginning Wove also allows forwarders to sync one-off quotes directly into CargoWise with a si

Why Wove Is Going Deep on CargoWise

Case Studies

### Customer Background - **Company**: Explorate - **Industry**: Digital Freight Forwarding - **Headquarters**: Brisbane, Australia - **Size**: 20,000 containers shipped annually [Explorate](https://explorate.co) is a leading Australian digital freight forwarder focused on delivering streamlined, technology-driven shipping solutions. They manage thousands of container shipments per year, leveraging their proprietary RMS to serve customers across the globe. ### The Challenge Traditionally, Explorate received carrier rates in a variety of formats—PDFs, spreadsheets, emails—which required its team to manually reformat each file to be compatible with their internal rate management system. This issue became more acute when shipping lines shifted to releasing updated rates every two weeks, turning what used to be an occasional task into a high-frequency operational burden. Each rate file, sometimes over 200 lines long, had to be processed manually line-by-line, delaying pricing updates and occasionally resulting in missed quoting opportunities. > We were constantly at risk of quoting outdated rates or delaying responses to customers, which meant potentially losing business. — Alex, Co-Founder & CEO at Explorate ### The Solution Explorate turned to Wove’s AI-powered rate data ingestion engine to automate the process. Wove’s solution integrated directly with Explorate’s home-grown RMS, enabling the team to upload and ingest shipping line tariffs at speed. What once required manual formatting and hours of effort can now be completed in minutes. > The Wove product is very simple and easy fix to what is traditionally a very manual process. The speed we can now ingest shipping line rates and update our RMS is extremely efficient. — Brad, Commercial Manager at Explorate ### Results & Impact **Time Savings** While hard to precisely quantify, Explorate’s Procurement and Pricing teams report significantly reduced time spent on manual data entry—freeing up resources to focus on more strategic, revenue-generating tasks. **Cost Efficiency** Automating rate ingestion has reduced the need for repetitive admin work and helped prevent costly quoting errors. **Accuracy & Performance** With cleaner, faster and more accurate data ingestion, Explorate can now respond to RFQs with up-to-date pricing, reducing delays and increasing the likelihood of securing bookings. **Employee Experience** Team members no longer need to deal with repetitive formatting tasks and can instead concentrate on higher-value activities, improving morale and productivity. ### Reach Out Are you a freight forwarder, shipper, or technology vendor struggling with manual rate ingestion workflows? Let's get in touch!

Case Study: How Explorate Automated Rate Ingestion and Unlocked Operational Efficiency with Wove

In a [recent MIT analysis of artificial intelligence in logistics](https://mitsloan.mit.edu/ideas-made-to-matter/how-artificial-intelligence-transforming-logistics), industry experts highlight how AI is revolutionizing supply chain management while noting that many organizations are still navigating its implementation. The insights from MIT's Chris Caplice and Uber Freight's Lior Ron reveal both the opportunities and challenges in this rapidly evolving space, offering valuable perspectives on how AI is reshaping the industry's future. ### Understanding the AI Evolution in Logistics The landscape of analytics in logistics has evolved significantly, with three distinct but complementary approaches. Traditional AI focuses on completing specific, predefined tasks and relies on structured data analysis, working well for routine, repetitive processes. Generative AI leverages large language models for context understanding and adapts to varying input formats, while Operations Research uses scientific methods and network models for decision-making systems. ### Major Logistics Challenges Being Addressed The industry faces several critical challenges that AI is uniquely positioned to address. Supply chain fragmentation remains a significant hurdle, with disconnected networks leading to inefficiencies and lack of visibility across multiple stakeholders. Market volatility, including rapid price fluctuations and service disruptions, continues to challenge organizations. Safety and security concerns, from driver safety to digital fraud prevention, require innovative solutions. Additionally, the environmental impact of logistics operations, particularly carbon emissions from inefficient routing and fuel waste, demands immediate attention. ### The AI Advantage: Breaking New Ground Research from MIT demonstrates several groundbreaking advantages of AI implementation. Most notably, AI models consistently outperform their training data, showing adaptive learning that improves outcomes over time. These systems have proven particularly effective at handling complex, multi-variable scenarios that traditional methods struggle with. The technology's ability to automatically adapt to new policies and eliminate the need for specialized algorithms represents a significant leap forward in operational efficiency. ### Wove's AI-Powered Solutions in Action In this evolving landscape, Wove has developed two key solutions that leverage AI to address critical industry challenges. Our Data Ingestion Engine employs advanced OCR and document understanding capabilities to process multiple document formats, from PDFs to scanned documents and, of course, Excel spreadsheets. This technology has achieved a 90% reduction in processing time while maintaining near-perfect accuracy rates, delivering significant cost savings and improved data quality. The Email Copilot represents another breakthrough in logistics automation. Using natural language processing and context-aware information extr

AI in Logistics: The Evolution from Traditional Methods to Intelligent Automation

Wove Email Copilot is a revolutionary AI-powered solution that helps logistics professionals process emails 10x faster. By leveraging advanced machine learning algorithms, it can automatically extract key information from emails, identify action items, and suggest appropriate responses.

Introducing Wove Email Copilot

Global supply chains continue to face unprecedented challenges in 2025. From ongoing geopolitical tensions affecting shipping routes to the accelerating impact of climate change on transportation networks, the complexity and vulnerability of international logistics are testing organizations like never before. A recent [McKinsey Global Supply Chain Leader Survey](https://www.mckinsey.com/capabilities/operations/our-insights/supply-chain-risk-survey) reveals a concerning trend: while disruptions persist, many organizations are scaling back their resilience initiatives. This pullback comes at a critical time when 90% of companies report facing supply chain challenges. ## The State of Supply Chain Management in 2025 As we progress through 2025, the survey highlights several worrying gaps in organizational readiness. Perhaps most concerning is the disconnect at the executive level - only a quarter of companies have formal processes to discuss supply chain issues at the board level. In an era where supply chain disruptions can instantly impact global operations, this lack of strategic oversight leaves organizations exposed to significant risks. While progress has been made in supply chain visibility, with 60% of companies reporting comprehensive visibility into their tier-one suppliers, there's a troubling decline in deeper supply chain transparency. This limitation is particularly problematic as major disruptions often originate deep within the supply chain network. Key Challenges Identified: - Only 25% of companies have formal processes to discuss supply chain issues at the board level. - Deeper supply chain visibility is declining year over year - 90% of organizations lack sufficient talent to meet their digitization goals - Two-thirds of companies are implementing advanced planning systems, but only 10% have completed deployment - It takes companies an average of two weeks to plan and execute responses to disruptions ### The Digital Transformation Imperative In 2025, while organizations recognize the critical need for digitization, implementation remains a significant challenge. The rise of artificial intelligence and machine learning has raised the stakes for digital transformation, yet many companies struggle to effectively implement these technologies. The survey reveals that while two-thirds of companies are investing in Advanced Planning and Scheduling (APS) systems, only 10% have completed their deployments. ### The Role of Data Automation As companies grapple with these challenges, the importance of efficient data management and automation becomes increasingly clear. This is where modern solutions like Wove are transforming supply chain operations. By leveraging AI and automation, organizations can address many of the key challenges identified in the McKinsey survey. How Wove's Solutions Address Current Challenges: ### Data Ingestion Engine: - Automatically processes and standardizes rate data from various formats (spreadsheets, PDFs) - Elim

Notes on trade, tariffs and automation.