Joshua Gould on thebigword, WordSynk, and AI Operations

Joshua Gould on Turning Language Services Into an AI Operating Company

Before WordSynk, automated routing, and AI-assisted language workflows, translation work could arrive by fax. Someone at thebigword would count the words by hand.

Joshua Gould uses that memory to explain the distance the company has traveled. He began in sales, worked a night shift calling the United States from England, moved to New York, and opened an office serving financial clients. When the financial crisis damaged that market, the company moved deeper into government work. Stricter procurement and delivery demands made workflow automation a competitive requirement.

Episode 109 is not a story about a model replacing a translation agency. It is a story about a service business repeatedly redesigning itself around new technical capabilities.

From selling language work to redesigning its delivery

Gould says he learned sales in the drinks industry before joining thebigword as a telemarketer. The move to New York placed him near banking clients, but the recession changed the opportunity. He describes seeing customers leave Wall Street offices carrying boxes and realizing the business needed a different market.

Government contracting changed more than the buyer. It increased the importance of repeatability, security, auditability, and price. Those conditions pushed the company to examine the full operating path behind an interpretation request.

The company's current public identity is consistent with the broad arc. thebigword's official company page identifies Josh Gould as chief executive and describes translation, interpreting, and localization services. The UK registry lists THEBIGWORD GROUP LIMITED as an active private company whose stated activity is translation and interpretation.

Those sources establish the current role and legal record. They do not independently establish every customer, revenue, automation, or savings figure mentioned in the interview.

WordSynk is an operating layer

Gould compares WordSynk with a ride-hailing application. A customer requests a language service. The system identifies the need, routes work, coordinates a qualified linguist or machine path, tracks delivery, and supports administration around the job.

The current WordSynk product page describes a unified language platform spanning translation, transcription, and interpreting. A separate WordSynk Network page describes an AI-aided system that matches linguists to work.

The durable idea is not the analogy with a taxi app. It is the separation of the request from the capacity that fulfills it.

flowchart LR
    A["Customer language need"] --> B["Structured intake"]
    B --> C["Classify language, channel, context, and risk"]
    C --> D{"Route"}
    D --> E["Automated or machine-assisted path"]
    D --> F["Qualified human linguist"]
    E --> G{"Confidence and consequence acceptable?"}
    G -->|No| F
    G -->|Yes| H["Deliver and record evidence"]
    F --> H

In the interview, Gould says many requests can move through company applications without employee intervention. He also describes medical situations in which a customer may begin with an automated path and switch to a human interpreter when the conversation becomes too technical.

That 80-percent automation claim is useful as his account of the operating ambition. It is not used here as a current audited metric because the transcript does not define the period, denominator, service mix, or meaning of “touched.”

Translation technology did not remove the market

Gould reaches back to translation memory in the 1990s. Reusing previously translated content reduced the volume of words that needed fresh translation. That could have been treated only as lost billable work.

He says the company told a major customer and found that lower unit cost allowed more content to be translated. Later machine translation created another threat and another opportunity. The company could add orchestration, editing, subtitling, dubbing, and learning-content workflows around the underlying engine.

This is the episode's strongest answer to simple automation narratives. Lower cost can remove tasks while expanding demand. The expansion is not guaranteed, and the people affected by the removed work do not automatically receive the new work. But the market boundary can move.

Gould's version remains a company and industry observation. Public claims about market growth, the share of words translated by AI, or company savings need a defined source and period. The principle survives without the unsupported precision.

“AI is an engine, not the car”

Near the end of the conversation, Gould offers the episode's most useful metaphor. Buying a model is like buying an engine. The organization still needs fuel, cooling, timing, controls, a driver, maintenance, and a vehicle designed around the engine.

For a business, those surrounding systems are the workflow, data, roles, interfaces, permissions, evaluation, human escalation, incident response, and economics. A strong model connected to weak data can scale the wrong answer. A fast workflow without an exception path can make a rare failure expensive.

NIST's current AI Risk Management Framework core treats governance as a cross-cutting function and assigns executive leadership responsibility for decisions about AI risk. That supports Gould's broader point: integration is an operating decision, not a software install.

It does not support every architectural prescription in the conversation. Gould says older stacks and monolithic systems can make AI integration difficult and praises microservices. That can be true in a specific environment, but a company does not need to rewrite every system into microservices before it can use AI responsibly. Architecture should follow the workflow, constraints, and evidence.

Resistance is not only technical

Gould describes managers who understand the potential of AI but recognize that adoption may reduce the size or status of their teams. Their objections may be partly about quality and partly about identity, power, or job security.

Calling that resistance irrational misses the operating problem. People need to know how their work changes, what decisions remain human, how performance will be measured, and what happens when the system fails. A deployment that treats the workforce as an obstacle will receive hidden workarounds and poor incident reporting.

The current Census Bureau Business Trends and Outlook Survey shows that firm adoption is real but uneven. In its May 2026 analysis, overall reported business AI use remained around 17 to 20 percent, with higher use among larger firms and in information, finance, and insurance.

That is a more grounded picture than “everyone has adopted” or “no one is buying.” Adoption depends on company size, function, industry, and the survey definition.

Technical leadership is decision competence

Gould does not claim to be the person writing the code. He says he needs to understand the technology well enough to build the team and enter consequential decisions.

His cybersecurity example makes the distinction concrete. Before the interview, a security leader told him an outside exercise provider did not expect the CEO to join a simulated ransomware exercise. Gould objected. He might be required to make a shutdown decision or explain the event publicly.

A chief executive does not need to operate every defensive tool. They do need to understand the business consequence, authority, timing, uncertainty, and evidence behind a decision that could stop a service or affect customers.

The SEC's current cybersecurity disclosure rule requires covered public companies to describe management's role in assessing and managing material cyber risk and the board's oversight. NIST's Cybersecurity Framework 2.0 adds an explicit Govern function that connects cybersecurity with enterprise risk.

Technical fluency therefore means knowing enough to be accountable, not performing engineering as theater.

Learning through responsibility

Gould did not attend college. His technical education came through sales, operations, contracting, repeated disruption, and responsibility for outcomes.

He argues that many midcareer workers will learn AI on the job rather than returning to school. That is a plausible pathway, but the episode sometimes states it too absolutely. Adults do return to formal education. Apprenticeships, employer training, short courses, self-directed projects, and degrees offer different combinations of structure, theory, practice, credentials, and income.

The useful leadership move is not forcing an employee to “figure it out” without support. It is giving them a real problem, bounded authority, training time, feedback, and an observable outcome.

About Joshua Gould

Joshua Gould is chief executive of thebigword, a language technology and services company. The company's current profile says he leads AI-related innovation and global expansion across translation, interpreting, and localization.

The interview traces his path from telemarketing and sales through U.S. expansion, government services, product development, and group leadership. Read the full [[Joshua Gould Guest Profile|Joshua Gould profile]] for verified current links and the evidence boundary around company scale claims.

Continue the conversation

The full episode adds Gould's views on capital, employment, risk, competition, and executive behavior. Those viewpoints are worth hearing with their qualifications intact.

Continue with [[Why AI Integration Is an Operating Model Change, Not a Plug-In]] for the practical framework. The [[thebigword Company Profile|thebigword company profile]] and [[WordSynk Product Profile|WordSynk product profile]] own the current entity facts.

Sources and editorial notes

This article was checked on July 27, 2026 against the preserved E109 transcript, current thebigword and WordSynk pages, Companies House, Census business-adoption evidence, NIST AI and cybersecurity frameworks, and the SEC cybersecurity governance rule.

Career history and company transformation details are attributed to Joshua Gould. Unverified financial, customer, automation, investment, market, and workforce figures are not presented as independent facts. AI assisted with research organization and drafting; Dalton Anderson remains responsible for the editorial decisions and publication authority.