Relay $36M Funding Turns Frontline Talk Into an AI Data Test

Relay has secured $36 million to turn frontline radio conversations into structured business data, making the Relay $36M funding more than another hardware expansion.

The Raleigh, North Carolina company wants its wearable radios to capture spoken reports, recognize workplace terminology, and initiate actions in connected software. That shifts Relay’s pitch from replacing walkie-talkies to building an AI interface for physical operations.

International Paper led the round after using Relay in its own industrial environments. That customer-to-investor path gives the financing unusual strategic weight. It also raises the standard Relay must meet as it expands beyond communications.

The company is challenging two established habits. Traditional radios carry messages without preserving their context. Enterprise software expects employees to stop working, find a screen, and document events manually.

Relay believes a voice-first system can bridge that gap. Its hardest test is whether continuous speech capture produces reliable operational intelligence without creating noise, mistrust, or intrusive worker monitoring.

Relay $36M Funding Backs a Larger Software Ambition

The financing supports Relay’s attempt to make every useful frontline conversation part of an operational workflow.

Relay announced the round on September 30, 2026. The company said the financing brought its total capital raised to more than $90 million.

International Paper led the investment. New participants included Cerity Partners clients, Harmonic Growth Partners, and Thayer Investment Partners. Existing backers G2 Venture Partners and Wind River Ventures also invested again.

The composition matters because International Paper was already a Relay customer. Thayer’s investment partners also represent hotel operators that have used Relay across thousands of properties, according to the company.

That history gives Relay more than financial validation. Its lead investor understands the conditions where the product must operate, including loud facilities, distributed teams, safety risks, and limited access to screens.

Relay’s funding announcement says the round will support safety, productivity, and operational intelligence. Those goals extend well beyond basic push-to-talk communications.

The company plans to invest in next-generation devices and accessories for specialized physical environments. It also intends to improve models used for transcription, translation, noise cancellation, location analysis, and operational inference.

Relay will expand product, customer success, and go-to-market teams, particularly across industrial operations, hospitality, and healthcare. Chief Executive Chris Chuang told Axios that the company expects to hire around 100 people during the next year.

Relay had approximately 300 employees when the funding was announced. It had grown enough to move into larger office space at Bandwidth’s campus near Raleigh’s Lenovo Center.

The company does not disclose current revenue or its valuation following the investment. Those omissions make it difficult to assess the financing against normal growth metrics.

Relay did report unusually strong momentum before this round. Chuang said revenue increased 120 percent during 2023, when Relay raised a separate $35 million Series B.

That earlier financing helped Relay compete with traditional radios and broaden its software capabilities. The latest round targets a different milestone: converting the installed communications layer into an AI system for physical work.

Relay says its platform now operates across nearly 10,000 sites. It also claims to process more than one billion operational data points each week and serve about 10 percent of Fortune 500 companies.

Those figures come from Relay and have not been independently audited. Still, they indicate the scale of the dataset behind its next product phase.

A radio replacement can win by offering clearer audio, broader coverage, simpler management, or useful safety functions. An operational intelligence platform faces a more demanding standard.

It must identify which conversations matter, represent them accurately, and deliver the result to the appropriate workflow. It must do so without distracting the worker or overwhelming the manager.

This is what makes the Relay $36M funding significant. The money is financing a transition from communications equipment toward an enterprise data layer built around spoken work.

That ambition creates the central tension. Relay must preserve the simplicity of a radio while adding the governance, integrations, and accuracy expected from business software.

Why Frontline Intelligence Starts With Missing Data

Relay is betting that the most valuable unstructured enterprise dataset is already being created through everyday speech.

Frontline workers regularly report damaged equipment, safety concerns, inventory problems, guest requests, and incomplete handoffs. Much of that information reaches a colleague but never reaches a database.

Relay calls this loss “signal evaporation.” The term describes useful context that disappears because nobody records it or because a digital alert receives no response.

Consider an operator saying that a palletizer has failed three times during one week. The statement contains a maintenance pattern, not merely a status update.

A traditional radio delivers the message to anyone listening. It does not create a maintenance record, connect previous failures, or confirm that somebody opened a work order.

Manual documentation can recover the information later. However, a worker may be several tasks removed from the incident when completing a report.

Details can disappear during that delay. Specialized terminology can be simplified, and apparently minor observations may never enter the formal record.

Relay says its system can capture the original voice exchange, transcribe it, and convert the content into structured, searchable information. A workflow can then route the event to another system.

The company has designed models to recognize domain vocabulary. In the palletizer example, the model must distinguish an industrial machine from a similar-sounding everyday word.

This capability matters because generic speech recognition performs poorly when background noise, accents, abbreviations, and specialized equipment names appear together. Physical workplaces frequently contain all four.

Relay also offers TeamTranslate, which delivers voice translation across more than 30 languages. A worker speaks normally, while teammates hear the message in their chosen language.

Translation can improve coordination among multilingual teams. It also introduces another accuracy layer between a worker’s observation and the action taken by a supervisor.

Relay’s broader system combines wearable radios, accessories, cloud software, mobile and desktop interfaces, AI models, and captured workflow data. The device becomes the collection point for that stack.

This design avoids asking employees to carry a conventional smartphone throughout every shift. Phones can be unsuitable around machinery, gloves, patients, guests, or environments where a screen creates a distraction.

Voice also reduces the number of steps needed to report an event. The worker does not need to select a form, identify the correct field, or type a description.

That advantage is meaningful only when the system correctly understands the message. Fast collection cannot compensate for an inaccurate transcript or a wrongly generated task.

The company says roughly 80 percent of activity on a physical operations floor never reaches an enterprise system. It also says 62 percent of surveyed manufacturers encounter delays, rework, or missed handoffs several times monthly.

Both figures come from Relay’s own research and should be treated as company claims. Relay has not published enough methodology to establish how broadly the results represent industrial operations.

Even so, the underlying problem is familiar. Enterprise systems contain what people took time to record, not necessarily everything workers observed.

An older manufacturing survey found that two-way radios remained a primary plant communication method. It also found strong interest in connecting communications with operational data.

That combination explains why Relay sees an opening. Radio conversations already sit close to the event, while most enterprise software sits closer to management.

Frontline intelligence is Relay’s name for closing that distance. The term refers to converting workers’ immediate spoken knowledge into data that managers and software can use.

The funding does not establish that Relay has solved the problem. It gives the company more resources to test whether voice can become a dependable system input.

The Real Opponent Is the Traditional Radio Model

Relay’s primary contest is between disposable radio traffic and communication that can trigger software-driven action.

Chuang has described Motorola Solutions as the entrenched force Relay first needed to challenge. He told Axios that competing with Motorola required considerable time and investment.

Motorola represents the traditional strength of professional radio systems. Those systems emphasize instant communication, durable equipment, dependable coverage, and established support networks.

Relay is not trying to win solely through better audio or a smaller device. It is arguing that communications should also create persistent, usable operational context.

The contrast is sharper than a normal hardware comparison. A conventional radio solves the immediate question of who needs to hear a message. Relay wants to solve what happens after the message is heard.

That difference turns the radio from an endpoint into a software interface. A spoken maintenance report can become a ticket, while a safety alert can carry location and incident context.

Managers can also communicate with workers beyond the range of traditional local radio networks because Relay uses cloud connectivity. Smartphone integrations extend the communication loop to employees carrying other devices.

Relay’s wearable form helps it compete in jobs where a phone is awkward or prohibited. Hospitality workers can communicate across floors without exposing personal phone numbers or repeatedly unlocking a screen.

A factory team can use a push-to-talk motion while wearing gloves. A healthcare or education employee can request assistance without navigating an application during an urgent event.

Relay’s smart radio reporting describes how the platform can recognize workplace terms and initiate follow-up actions. This mechanism is central to its differentiation.

However, Motorola is not standing still. Its portfolio has expanded beyond land mobile radios into software, video security, analytics, and command-center systems.

Motorola also acquired Theatro Labs in 2025. Theatro developed voice-controlled communications and workflow software for frontline teams, making the acquisition particularly relevant to Relay’s strategy.

Other challengers are pursuing similar ideas. Weavix sells connected smart radios with translation and data capabilities, while Zello offers push-to-talk communication through software and mobile devices.

These alternatives create several distinct buying paths. A company can modernize a traditional radio fleet, deploy dedicated connected hardware, or use applications on existing phones.

Relay’s argument is strongest where ordinary phones are unsuitable and traditional radios leave too much information behind. It is weaker where employees already use screens comfortably.

The company also needs integrations that fit each customer’s existing systems. A generated maintenance task has limited value if it lands outside the software technicians actually monitor.

Hardware distribution gives Relay a practical advantage. A device used throughout every shift produces recurring opportunities to capture signals and support workflows.

That same dependency creates execution risk. Relay must maintain hardware reliability, connectivity, device management, security, AI services, and enterprise integrations as one experience.

Traditional radio vendors can focus on communication performance. Software vendors can focus on data and workflow. Relay is taking responsibility for both.

International Paper’s investment suggests that at least one large industrial customer sees strategic value in this combined approach. Thayer’s participation points toward similar confidence from hospitality operators.

Those endorsements do not establish broad market fit. Customer-investors can have objectives, deployment conditions, and risk tolerances that differ from the average buyer.

Still, the relationship improves the quality of Relay’s feedback loop. An industrial customer can expose the product to acoustic conditions and terminology that laboratory testing cannot reproduce.

That is why the Relay $36M funding places pressure on established radio suppliers. Relay is reframing the purchasing decision around captured knowledge, not just reliable communication.

If buyers accept that framing, radio vendors must answer a new question. They must explain how their platforms preserve, govern, and activate the information traveling across their networks.

Capturing Speech Creates Accuracy and Trust Risks

The same system that preserves operational knowledge can become unreliable or intrusive when its boundaries remain unclear.

Relay’s opportunity begins with recording information that previously disappeared. That also means collecting communications workers may have assumed were temporary.

Enterprises will need clear rules covering what is captured, how long it remains available, and who can search it. Workers also need to understand when AI is interpreting their conversations.

A hotel request, a patient-related message, or a safety report may contain sensitive information. The applicable controls can vary by industry, jurisdiction, and deployment.

Relay’s public materials emphasize safety, productivity, and operational insight. They provide less detail about data retention, worker access, appeal processes, and limits on managerial analysis.

Those questions are not peripheral. A system can change workplace behavior when employees believe every informal observation may become a permanent performance record.

The product’s value may decline if workers avoid candid communication. Teams might shift sensitive discussions outside the system, recreating the missing-data problem Relay wants to solve.

Accuracy presents another challenge. Industrial environments contain loud machinery, radio compression, overlapping voices, uncommon equipment names, and multilingual conversations.

A transcript may appear plausible while changing a critical word. Translation adds another opportunity for meaning to shift before software initiates an action.

False negatives can allow a safety issue or maintenance pattern to disappear. False positives can generate duplicate tasks, unnecessary escalations, and alert fatigue.

The greatest risk emerges when an incorrect interpretation triggers automation. A searchable transcript can be reviewed, but an automatically routed action may affect operations immediately.

Relay must therefore show more than average transcription accuracy. Customers need performance measures broken down by noise level, language, accent, vocabulary, and workflow type.

They also need confidence thresholds. A low-confidence interpretation should request human confirmation instead of silently creating an authoritative record.

NIST’s AI risk framework offers a useful standard for evaluating systems like this. It emphasizes governance, measurement, monitoring, and documented privacy risks.

For Relay customers, practical governance should identify the accountable human for each automated output. It should also establish how employees correct transcripts and challenge inaccurate records.

Security matters because the system combines voice, location, identity, and operational events. A breach could reveal facility conditions, employee behavior, or unresolved equipment problems.

Cloud dependence introduces a separate operational concern. Buyers need to know how core communication and safety features behave during weak cellular or Wi-Fi coverage.

Relay’s hardware must also withstand the environments described in its sales pitch. A device that supports AI workflows but fails during a demanding shift does not replace a trusted radio.

Integration quality creates another uncertainty. Enterprise customers operate maintenance, safety, scheduling, and workforce platforms that differ widely across sites.

A successful pilot can depend on customized work. Scaling that deployment across hundreds of facilities requires repeatable integrations, administration, and training.

Relay has disclosed broad usage figures but not the performance details needed to judge this layer. It has not published retention rates, expansion revenue, or AI-generated action accuracy.

The company also has not separated how many sites use advanced intelligence features from those using communication or safety functions. Nearly 10,000 sites does not necessarily mean 10,000 AI deployments.

That distinction will shape whether Relay becomes an enterprise intelligence provider or remains a communications vendor with useful AI features.

Relay’s growth profile shows that it has already built a substantial organization. The new hiring plan increases the need for disciplined execution.

Adding sales capacity can accelerate device deployments. Building trusted operational intelligence requires slower work around data quality, governance, integrations, and customer process design.

The funding gives Relay room to pursue both. It does not remove the tradeoff between moving quickly and proving that AI-mediated workplace records deserve authority.

Three Signals Will Show Whether Relay’s Bet Is Working

Relay’s next phase should be judged through adoption depth, verified workflow outcomes, and worker trust rather than funding alone.

The first signal is whether existing customers expand from communication into AI-driven workflows. International Paper is the most visible test because it is both a user and lead investor.

A meaningful expansion would involve more facilities, more frontline roles, and repeated use of captured speech in maintenance or safety systems. That would strengthen Relay’s platform argument.

A rollout limited to radio replacement would tell a different story. It would show demand for connected communications without proving that frontline intelligence drives the purchase.

Relay should disclose how many deployed sites use transcription, translation, operational inference, and automatic workflow creation. Aggregate device or site counts cannot answer that question.

The second signal is independently verifiable operational performance. Customers should report whether the system reduces response time, repeated equipment failures, incomplete handoffs, or administrative work.

Relay’s palletizer example makes the desired outcome easy to understand. The important measurement is whether detecting that pattern leads to faster maintenance and less unplanned disruption.

Buyers should also compare the number of useful actions with the number of false or duplicate alerts. High activity alone can hide an expensive signal-to-noise problem.

Translation deserves its own performance measures. A system used for safety communication should be evaluated under real workplace noise, not only through clean recorded speech.

The company’s previous growth offers encouraging context. During its 2024 financing, Relay reported 120 percent revenue growth and said it still had unused capital from its earlier round.

The earlier funding also focused on sales expansion and product capabilities. Two years later, buyers can reasonably expect more specific evidence.

Published customer case studies should include starting conditions, deployment size, measured outcomes, and evaluation periods. They should separate communications benefits from AI-derived improvements.

If those results appear across manufacturing, hospitality, and healthcare, Relay’s intelligence thesis becomes stronger. If evidence remains anecdotal, the company’s narrative stays ahead of its validation.

The third signal is how Relay handles worker governance. Adoption depends on whether frontline employees view the device as a useful tool or a monitoring system imposed by management.

Customers should explain what conversations are processed, who can access them, and how workers correct errors. The system should make its capture state understandable during normal use.

Relay should also distinguish safety and workflow analysis from individual performance surveillance. Without that boundary, managers may use the data in ways that damage trust.

Worker participation can improve the system. Employees know which terminology, alerts, and exceptions matter in practice, while centralized project teams often miss local context.

A credible deployment process would include frontline feedback before expanding automation. It would also provide a path for workers to report recurring model errors.

These three signals connect directly. Broad adoption is fragile without measurable outcomes, while strong results may not persist without worker trust.

Competition will influence every signal. Motorola and other vendors can bundle more software into established customer relationships, while phone-based products can reduce dedicated hardware requirements.

Relay’s advantage is the possibility of owning the full loop. Its system can capture a spoken event, identify its meaning, route an action, and return an update to the worker.

That loop is valuable because frontline employees often contribute information without receiving visibility into what happened next. Closing it can make reporting feel useful instead of administrative.

The Relay $36M funding gives the company enough capital to pursue this larger role. Its customer-led round also provides access to environments where the approach can be tested at scale.

The next several months should reveal whether Relay publishes deeper deployment metrics, expands its intelligence features, and describes stronger data controls. Those disclosures will matter more than another list of supported capabilities.

Enterprise buyers should evaluate Relay as both communications infrastructure and an AI data system. That means testing audio reliability and workflow accuracy with equal rigor.

They should start with a bounded use case, such as recurring maintenance reports or urgent hotel requests. The deployment should define the desired action, human reviewer, and correction process.

Teams can then compare captured events with existing records. They should measure whether the system finds useful information that would otherwise disappear.

The result will determine what Relay has actually built. It could become a valuable interface between physical work and enterprise software, or remain a modern radio with optional analytics.

Relay’s financing makes that test possible, but customer evidence will decide the outcome. Watch what happens after workers speak: what gets captured, what action follows, and whether they trust the process.