Eytan Sheerit

Their position is that integrations, workflows, customers and accumulated operational knowledge form the foundation, with AI adding new capabilities on top. Replacing the platform would throw away what already works, so early generative-AI failures argue for grafting on capability rather than starting over.

A post describing almost a decade of experiments with early Google and Microsoft machine-learning services argues that you do not transform a mature software company by throwing away the past.

• Instead, accumulated knowledge, integrations, customers, mistakes and workflows gain entirely new capabilities.
• The same post says early AI-built applications looked impressive, then destroyed code they had written themselves, causing features that had worked minutes earlier to stop without explanation.

Across the sequence, the argument stays the same while the emphasis shifts: earlier posts dwell on failed experiments and caution, while the order fulfillment post presents a rebuilt 1999 concept as a shipped AI-era product.

Read original · Original 2 · Original 3

Current AI capabilities build on integration work that predates the generative AI boom; earlier tools such as screen scraping and floor-side mobile devices addressed the same problem of systems that do not talk to each other.

A history post traces an engine built in the 1980s that used screen scraping and HLLAPI/EHLLAPI to drive legacy mainframes, simulating more than 100 concurrent users to create and update house bills, consolidations and milestones.

• In 1999, a warehouse mobile application translated floor work into mainframe updates instead of re-keying it.
• Early-2000s industrial Wi-Fi and touch devices from Symbol Technologies brought that capability to warehouse operators, and it later became the CMX Logistics Nerve Center.
• The order fulfillment launch ties its concept back to 1999 aerospace purchase-order matching.

Read original · Original 2 · Original 3

The post presents that experiment as the platform's foundation, not a side project. The current order fulfillment flow reads documents before matching, compliance checks or document generation run.

The order fulfillment post describes AI reading carrier labels, invoices, purchase orders and export documents before guiding warehouse operators through receiving, pieces, weights and dimensions. A separate post lists document understanding alongside warehouse vision, workflow generation and conversational voice as current capabilities.

Warehouse vision appears in the same posts as a current capability, but Eytan Sheerit's posts do not walk through a specific vision use case, failure mode or measured result, so it remains a stated capability rather than a demonstrated one.

Read original · Original 2 · Original 3

The post quotes a CIO describing the standard route for a large enterprise entering a new industry, using coal mining as an example of a business the company already runs end to end. Sheerit's point: describing the operation and generating workflows shifts the conversation from efficiency to how quickly a company can enter a vertical.

During the same conversation, Sheerit opened the CargoMatrix AI Workflow Generator, described what was needed and says workflows appeared in about five minutes.

• Another five to ten minutes of refinement could make them publishable to mobile and desktop users within the hour.
• The post contrasts a months-long process of business analysis, requirements, IT development, QA, UAT and production with a describe, generate, refine and publish path.
• The output is executable workflows covering tasks, forms, business rules, validations, photos, documents, approvals, exceptions and milestones—not a flowchart or slide.

Read original · Original 2

The agent starts with a company's approved website content, products and business knowledge.

• Connections through MCP to CargoMessenger and CargoWise let it retrieve operational context, validate information, create or update tasks, trigger workflows and keep the right team informed.
• The listed uses are operational, not informational.

Use cases named in the post include driver and visitor check-in, shipment-status and customer-service conversations, customs and document intake, appointment and dock coordination, product and service guidance, and employee or partner support. A driver self-check-in kiosk is cited as a production example.

The post says a video shows the same studio configured for different businesses; that video is not shown in the posts here, so the examples in it should be treated as described rather than seen.

Read original · Original 2

The company traces this approach to its aerospace origin, where purchase-order matching, traceability and compliance requirements are described as unforgiving. The same flow is presented as configurable for automotive, electronics, pharmaceutical, perishables and medical devices.

In the order fulfillment flow, every invoice line is automatically matched against the purchase order, and overages, wrong part numbers, missing certificates and other exceptions are identified before cargo moves.

• Export compliance checks including ECCN, ITAR, licensing and USPPI are built into the workflow.
• HAWB, BOL, proforma invoice and master shipment documents are generated from validated data, shipments synchronize with CargoWise One, and an EDI 856 ASN can be sent back to the customer.
• The stated principle: cargo does not move just because somebody clicked Next; it moves when the order, documents, physical cargo and required controls agree.

Read original · Original 2

The post traces three decades of integration work beneath the recent AI claims.

Sheerit traces logistics automation to the 1980s, when freight forwarders ran proprietary mainframe systems that were difficult to integrate with.

• They describe building a multithreaded engine that used screen scraping and HLLAPI/EHLLAPI to drive those mainframes directly, simulating more than 100 concurrent users to create and update house bills, consolidations and milestones.
• In 1999, the team built a mobile application for warehouse operations, translating floor work into mainframe updates through software agents instead of requiring it to be re-keyed.
• Early-2000s industrial Wi-Fi and touch devices from Symbol Technologies brought the capability to warehouse operators; it later became the CMX Logistics Nerve Center.
• The post frames AI as the next step in the same mission: remove manual work, connect fragmented operations and help logistics staff make decisions faster.

For brands planning a campaign

For a thought-leadership brief on AI-driven supply chain automation, a history-led angle—why automation projects survive contact with legacy systems—draws on material few competitors can match. It also gives an advisory collaboration a concrete theme: integration debt as a reason AI pilots stall.

Useful context for the workflow-generator post.

This post distinguishes a website voice agent that answers questions from an operational agent that helps get work done.

• Voice Agent Studio starts with a company’s approved website content, products and business knowledge to build a branded, multilingual voice agent.
• MCP connections to CargoMessenger and CargoWise through eAdaptor Next let the agent draw on operational context, validate information, create or update tasks, trigger workflows and notify the right team.
• Listed uses include driver and visitor check-in, shipment-status and customer-service conversations, customs and document intake, appointment and dock coordination, product guidance, and employee and partner support.
• A driver self-check-in kiosk is named as a production example, and a video shows the same studio configured for warehouse operations, customs brokerage, ecommerce and product guidance.

For brands planning a campaign

A demo-led content series has a natural anchor: the driver check-in kiosk is a specific, filmable use case, and the closing question—“which conversation in your business should become a workflow?”—is a ready-made engagement prompt.

Sheerit says the company did not decide one morning to become an AI company.

• It had spent almost a decade on early AI and machine-learning services from Google and Microsoft, mainly document intelligence: testing whether a machine could read an airway bill, commercial invoice, packing list or shipping label and turn unstructured documents into structured operational data.
• When generative AI arrived, they say, early builds looked impressive but broke themselves: asking the same AI to make another change destroyed code it had written, and features that worked minutes earlier stopped working.
• Their lesson: a good demo is not production readiness, and transforming a mature platform means adding capability rather than discarding integrations, workflows, customers and accumulated mistakes.
• Current capabilities listed include warehouse vision, conversational voice agents, AI-generated operational workflows and agents that interact with logistics systems.

For brands planning a campaign

A “what we got wrong with AI” narrative could anchor an advisory or co-authored article for a logistics software audience. The failed-demo detail reads as firsthand experience, not positioning, and contrasts naturally with vendors that show only polished pilots.

A launch post for a rebuilt Order Fulfillment product, first developed by the company in 1999 for aerospace logistics.

• The flow: AI reads carrier labels, invoices, purchase orders and export documents; warehouse operators receive guidance through receiving, pieces, weights and dimensions; every invoice line is automatically matched against the PO; overages, wrong part numbers, missing certificates and other exceptions are identified before cargo moves; export compliance checks, including ECCN, ITAR, licensing and USPPI, are built into the workflow; HAWB, BOL, proforma invoice and master shipment documents are generated from validated data; pick, pack and ULD-build tasks go to operators; shipments synchronize with CargoWise One, and an EDI 856 ASN can be sent back to the customer.
• The stated principle: cargo moves when the order, documents, physical cargo and required controls agree.
• Aerospace is the starting vertical, with automotive, electronics, pharmaceutical, perishables and medical devices named as configurable extensions.

For brands planning a campaign

Compliance-heavy verticals are a natural collaboration topic. The detail on ECCN, ITAR and PO matching suits a joint webinar, annotated workflow diagram or explainer for aerospace and medical-device shippers.

It also demonstrates a repeatable content format—a real customer conversation used to illustrate a product capability.

Sheerit recounts asking a CIO what it takes for a large enterprise to start servicing a completely new industry or vertical.

• The answer was the familiar sequence—business and process analysis, then requirements, development, QA, UAT and production—which can take months.
• During the same conversation, they opened the CargoMatrix AI Workflow Generator, described the operation needed and say workflows appeared in about five minutes.
• With five to ten minutes of refinement, the workflows could potentially be published to mobile and desktop users within the hour.
• The post stresses that the output is not a flowchart or PowerPoint but executable operational workflows covering tasks, forms, business rules, validations, photos, documents, approvals, exceptions and operational milestones.
• It reframes the question from how AI makes an existing operation more efficient to how quickly a company can enter a new vertical, contrasting a months-long traditional path with a describe-generate-refine-publish path measured in minutes to hours.

For brands planning a campaign

The months-to-minutes contrast could anchor a thought-leadership series on enterprise AI adoption, while the CIO-conversation format could be reused for a video or interview series. An advisory angle could explore what governance and validation generated workflows need before reaching production.

That experiment became the base layer, not a side project.

• The current order fulfillment product reads carrier labels, invoices, purchase orders, and export documents before downstream matching and compliance steps run.
• A separate post lists document understanding alongside computer vision and workflow generation as current capabilities.

For brands planning a campaign

This is the strongest entry point for a thought-leadership collaboration.

• Start with a messy document — a packing list with inconsistent fields, an invoice that does not match its purchase order — to show supply chain audiences a visible problem and how AI parsing feeds the rest of the operation.
• Framing document intelligence as a decade-old problem that generative AI finally made practical fits a build-in-public narrative better than a general AI-in-logistics pitch.

Examples: Source 1 · Source 2 · Source 3

Sheerit says they opened the AI Workflow Generator during that conversation, described the operation, and produced executable workflows in about five minutes, with refinement and publishing possible within the hour. The output is tasks, forms, business rules, validations, photos, documents, approvals, exceptions, and milestones running on the platform — not a flowchart or slide.

Anecdote anchored to a specific conversation, followed by a two-column timeline comparing the traditional development path with the platform path.

For brands planning a campaign

For a vendor building demand around enterprise AI, this anecdote is the strongest format in the corpus: a named role, a stated process, a timed contrast, and a runnable result. A collaboration could commission a series — one story per vertical — while keeping the unglamorous details, including what had to be described and what still needed human refinement, alongside the headline time saving.

Examples: Source 1 · Source 2

Its use cases are operational: driver and visitor check-in, shipment status and customer service conversations, customs and document intake, appointment and dock coordination, product and service guidance, and employee or partner support.

• The agent starts with a company's approved website content, then connects through MCP to CargoMessenger and CargoWise to retrieve operational context, validate information, create or update tasks, and trigger workflows.
• A driver self-check-in kiosk is cited as a production example.

The post references a video demonstrating configuration for different businesses, but the video is not shown in the posts here, so the specific visual examples should not be described as if inspected.

For brands planning a campaign

Voice is where this creator can most clearly stand apart from generic AI commentary: the argument is about connecting conversation to systems of record, not model quality.

• A campaign could ask readers which recurring conversation in their operation should become a workflow, then show the integration path.
• Treat the production kiosk as a claim to illustrate with new material, not as documented proof.

Examples: Source 1 · Source 2

One post rejects the idea that the company woke up and decided to become an AI company, recounting years of experiments with early Google and Microsoft AI services, followed by generative AI tools that produced impressive demos and then broke the code they had generated. Another traces the automation lineage from mainframe screen scraping, through an engine that simulated more than 100 concurrent users to create and update house bills and consolidations, to a 1999 warehouse mobile application that pushed floor activity back into mainframe updates without human re-entry.

For brands planning a campaign

Lead with this distinction for enterprise audiences who have heard general AI claims and worry about disrupting existing systems.

• Asking how to add AI to a working operation without losing integrations, customers, and accumulated workflow knowledge is more useful than a replacement story.
• Keep the failure material — AI tools destroying their own code and features disappearing — rather than smoothing it out; it is unusually credible.

Examples: Source 1 · Source 2 · Source 3

The principle: cargo moves when the order, documents, physical cargo, and required controls agree. The company traces the concept to 1999 aerospace PO matching and now positions it for automotive, electronics, pharmaceutical, perishables, and medical devices.

For brands planning a campaign

This is the corpus's richest source of specific, checkable detail and the strongest basis for a technical content partnership.

• A campaign could follow one control — ITAR or missing certificates, for example — through the full exception path, where audience trust is built and generic logistics content falls short.
• The aerospace origin and configuration story for other regulated verticals offer a natural series structure.

Examples: Source 1 · Source 2

The posts name the capability but offer no example, result, or deployment detail.

For brands planning a campaign

Vision is the clearest opening for original material: firsthand demonstration would add the most to a collaboration. Develop it with the creator rather than presenting it as a proven capability.

Examples: Source 1 · Source 2

The company's early AI experiences are described as terrible: generated applications looked impressive, then broke themselves.

• The mainframe-to-mobile-to-AI history post follows a straight chronology.
• Hashtags, product links, and numbered checklists appear in most posts; the register stays plain and first-person plural or singular, not analyst-style.

Contrast pairs (traditional process versus platform process), chronological company history, and explicit retellings of failed early experiments.

For brands planning a campaign

Match the voice: specific numbers, admitted setbacks, and a clear before-and-after structure, with product links in a single closing call to action.

• Ghostwritten pieces that open with market-size framing will sound like a different author.
• The failure stories are a reusable asset and a reason readers would keep following this creator.

Examples: Source 1 · Source 2 · Source 3 · Source 4

The claims are consistent with one another but appear only in CargoMatrix's own posts; ask for independent corroboration before using them as verified company history.

Dates and firsts are based on a single source and have not been independently verified.

For brands planning a campaign

Treat the chronology as the creator's stated narrative, not verified company history. Before using the origin story as a headline claim, ask for a dated artifact or independent reference.

Examples: Source 1 · Source 2

Written in the first person ("we built," "our existing logistics technology"), they trace the company's history to 1999 but don't state whether Sheerit owns or works for CargoMatrix. The Order Fulfillment launch post walks through AI reading carrier labels and invoices, matching invoice lines against purchase orders, running export compliance checks, and synchronizing shipments with CargoWise One.

View example

One post says shipments can synchronize with CargoWise One and send EDI 856 advance ship notices to the customer; another says CargoMatrix agents can connect to CargoWise through eAdaptor Next to retrieve operational context and create or update tasks.

View example

The post presents it as a technical connection, not a commercial relationship with Sheerit.

View example

The post doesn't clarify whether CargoMessenger is a CargoMatrix product or a separate platform, or disclose a commercial relationship with Sheerit.

View example

Their posts on inventory syncing, webhooks, rate limits and billing engines overlap with Sheerit's emphasis on connecting fragmented operations and adding intelligence without discarding what already runs.

Fisher's posts take an engineering vendor's view of e-commerce and shipping integrations, without discussing AI document parsing, voice agents or workflow generation. Sheerit's posts focus more explicitly on adding AI capabilities to a mature logistics platform.

Read original · Original 2

Like Sheerit, they treat non-core but critical shipment steps—documents, classification, exceptions and compliance—as opportunities for software to reduce manual work.

Fujimoto's posts cover a free-tool launch and an analysis of AI's effect on freight intermediaries. Sheerit focuses more on product capabilities and company history within one logistics platform, with less investor and strategy commentary.

Read original · Original 2

The overlap with Sheerit is the freight and logistics software setting, the focus on systems of record, and the question of where AI creates operational value rather than just cost savings.

Borri's posts offer strategic forecasts and industry provocations, including the argument that AI may kill the standalone TMS category. Sheerit writes as a vendor building and shipping features, drawing on company history and concrete workflow examples.

Read original · Original 2

In a post on who captures AI-driven productivity gains and an argument about the future of standalone TMS software, Borri frames AI as a question of value capture and market structure, not just automation.

That strategic lens suits executive discussions, but offers less implementation detail than Eytan Sheerit's examples of AI-generated operational workflows and order fulfillment controls.

Examples: Source 1 · Source 2

A post on a free import-duty calculator and API-first trade tools connects a practical tool to a broader product direction. In a separate post on AI and freight brokerage, Fujimoto argues that exception handling—not simply matching freight to capacity—is where logistics value can persist.

These examples focus on trade data and freight-market dynamics. For a brief centered on warehouse execution or operational workflows, Eytan Sheerit's order-fulfillment example is more directly aligned.

Examples: Source 1 · Source 2

Fisher's integration post highlights the less-visible engineering work: syncing inventory, handling webhooks and rate limits, and preventing duplicate processing. A second post emphasizes the infrastructure behind high-load logistics platforms.

This is a focused infrastructure perspective, rather than the AI-enabled operational workflows and logistics use cases at the center of Eytan Sheerit's posts. For a campaign that needs concrete AI workflow examples, see their workflow generator post.

Examples: Source 1 · Source 2