Rev
Rev is an Austin, Texas-based speech technology company founded in 2010 that provides transcription, captioning and subtitling services, and has more recently repositioned itself as an investigative platform for legal professionals, law enforcement, journalists and researchers working with speech data. It began as a freelance marketplace built by early employees of oDesk, offering human transcription at one dollar per minute, and over fifteen years layered its own automatic speech recognition on top of that human foundation, launching the automated service Temi in 2017 and later the Rev.ai developer platform. The company says that since 2010 it has processed more than eighteen billion words from seven million hours of human speech, that its artificial intelligence transcription reaches 96 percent accuracy with human review pushing accuracy above 99 percent, and that it is trusted by more than one million users, including major law firms, court reporting agencies and media publications worldwide.
Founding by Early oDesk Employees
Rev was founded in 2010 by Jason Chicola and Josh Breinlinger, who had been, respectively, the third and fourth employees at oDesk, the online freelance marketplace that later became part of Upwork. Their idea was to apply marketplace thinking to a narrower problem. At oDesk, customers had to find, vet and manage remote workers themselves, which Chicola identified as the hardest part of working with distributed labor. Rev would instead sell a finished service, transcription or translation, while abstracting away the individual freelancers who performed it, so a customer bought an outcome rather than a hiring process.
The company operated quietly for its first years under other names, FoxTranscribe for transcription and FoxTranslate for translation, beginning with translation in 2010 and adding transcription in late 2011. It was headquartered in San Francisco at the time of its public launch, and later established its headquarters in Austin, Texas, where the company is based today. Breinlinger eventually stepped back from day-to-day involvement while remaining connected to the board, leaving Chicola as co-founder and chief executive and the public face of the company.
Public Launch and Early Pricing
Rev publicly launched in March 2013, announcing 4.5 million dollars in Series A funding led by Globespan Capital Partners, in a round that had closed in August 2012, with participation from investors including Craig Sherman, the former chief operating officer of Ancestry.com, and Austin Ligon, the founder of CarMax, both of whom joined the board. At launch, human audio transcription cost one dollar per minute with turnaround usually within a day, business document translation cost twelve cents per word, and immigration document translation was priced per page. Even then, the company's stated ambition reached beyond transcription: its founders believed internet-enabled work from home was at a tiny fraction of its potential, and that building the best environment for remote workers was a competitive advantage in attracting talent.
Early customers validated the model across sectors. During its testing period Rev built a base of around one thousand regular business customers including Pfizer, U.S. Bank, New York University, Princeton University, Land O'Lakes and government agencies such as the National Park Service and the Small Business Administration. About twenty percent of its transcription customers were in media, using the service to transcribe interviews, a use case that would become central to the company's identity.
The Freelance Workforce Model
Rev's supply side is a managed network of freelance transcriptionists, captioners and translators who work from home. Unlike open marketplaces, Rev designed tools specifically for the work: a browser-based transcription environment with controlled rewind, speaker labeling and shortcuts for frequently used or complex terms, so workers did not need to download and upload files separately. The company argued that on most freelance platforms workers spend half their time looking for work, and that by keeping its freelancers utilized with a steady flow of jobs, they could earn more per week even at moderate per-job rates.
Contemporary reporting put typical earnings at roughly twenty to thirty dollars an hour for translators and seven to fifteen dollars an hour for transcribers, with regular translators earning around one thousand dollars a month part-time. The model drew both praise for flexibility and, over the years, wider debate about pay in the gig transcription economy, but structurally it gave Rev something its later technology strategy depended on: an enormous, continuously growing corpus of accurately human-transcribed speech, which became training data for its speech recognition systems.
From Human Transcription to Artificial Intelligence
Rev's second act was automation. Because the company transcribed thousands of files every day, it accumulated the labeled audio needed to train a competitive automatic speech recognition engine, which it built from the ground up rather than licensing from a larger provider, training on tens of thousands of hours of accurately transcribed audio. In tests the company published comparing word error rates on a collection of public podcasts, Rev reported the lowest error rate among the major providers it benchmarked, including Google, Amazon and Microsoft. In 2019 Rev was named to the inaugural Forbes AI 50 list, recognizing its speech recognition work.
The first mass-market automated product was Temi, created in May 2017 and formally launched that summer, with a mobile app following in August 2017. Temi offered machine transcription in minutes at ten cents per minute, with turnaround of one to five minutes for a typical short file, an online editor for correcting and highlighting transcripts, and the ability to upload files or record directly in the app. Chicola pitched it especially to journalists, noting that reporters could spend six hours a week transcribing interviews, and Temi was reported in use by news organizations including the New York Times, Bloomberg, CNN and the Huffington Post. Alongside Temi, the Rev.ai platform exposed the speech recognition engine through an application programming interface so other businesses could build transcription into their own products and services.
Products and Services
Rev's service menu grew to cover the main ways speech becomes text. Human transcription, produced by its freelance network, targets customers who need maximum accuracy. Automated transcription serves speed and volume. Captioning and subtitling address video accessibility and localization, and the platform later added support for languages beyond English in its asynchronous speech recognition, including French, German, Portuguese and Spanish, using speaker identification technology to attribute words correctly. The company also developed recording, real-time transcription, artificial intelligence notetaking and summarization tools as part of an enterprise subscription platform.
In its current positioning, Rev emphasizes verifiability. Its legal and investigative platform is built around what it calls citations-first artificial intelligence, in which outputs are tied back to the original audio, video or document evidence so that a user can check any claim against its source. The company states that it applies strict confidentiality protocols and does not train third-party models on customer data, and it cites SOC 2 Type 2, HIPAA and GDPR compliance as evidence of its security posture. That combination of speech data, citations and compliance reflects a deliberate move upmarket, from commodity transcription toward work where an error has legal consequences.
The Legal Market and SmartDepo
The clearest sign of that strategy came in 2025, when Rev acquired SmartDepo, described as the legal industry's trusted solution for artificial intelligence-powered deposition summaries with fully accurate page-line citations. Rev announced the acquisition in March 2025, presenting it as powering a new generation of artificial intelligence for testimony analysis. At the same time, the company announced that its enterprise subscription platform offered HIPAA-compliant solutions for legal customers, including secure audio and video recording, real-time transcription, artificial intelligence notetaking and advanced summaries, and it exhibited the combined offering at legal industry events including Legalweek in New York and the ABA Tech Show in Chicago.
The legal focus fits Rev's assets. Depositions, hearings and interviews are long, terminology-heavy and unforgiving of transcription error, and Rev says its models are particularly strong on complex legal terminology and challenging audio. Its customer references in this phase include AM Law 100 firms and leading court reporting agencies, alongside the media publications that had used it since its early years.
Journalism and Research Use
Journalism remains one of Rev's defining markets. Newsrooms used the human service to turn interviews into text quickly and cheaply, then adopted Temi and automated transcription for first drafts, meeting notes and searchable archives. The company's own pitch to reporters was time: automating transcription returns hours each week that would otherwise be spent typing, allowing journalists to be present in interviews rather than focused on recording every word. Researchers and podcasters adopted the services for similar reasons, and the company long cited transcription of interviews as one of its most common professional uses.
Scale and Company Today
By its own accounting in 2025, Rev had processed over eighteen billion words drawn from seven million hours of speech since 2010, served more than one million users, and employed between fifty-one and two hundred people as a privately held company headquartered in Austin. Its artificial intelligence transcription is rated at 96 percent accuracy, with optional human review reaching above 99 percent. Those figures describe a company that has crossed from a services marketplace into a data and software business, while retaining the human service that made its name and trained its models.
Assessment
Rev's history captures a broader shift in media services. It started by organizing human labor more efficiently than its rivals, used that labor to generate the data that automation requires, and then moved toward the professional markets, law above all, where trust, citations and confidentiality command premium prices. Throughout, its product has been the same in essence: turning the spoken word into text that other people can search, quote, publish and act on. For a directory of media and information agencies, Rev stands out as a transcription provider that evolved into an artificial intelligence company without abandoning the human accuracy standard on which it was built.
