More than 400 million words a year in 27 languages: that is the volume Dell committed to run through RWS Evolve in a multi-year deal. Evolve is RWS’s answer to the question every large localization program now faces, which is how much of the work AI can do on its own. Its pipeline translates with Language Weaver, scores every segment with a quality estimation model, rewrites the weak ones with a private language model and hands what is left to RWS language specialists.
RWS Evolve suits global enterprises with very large documentation, support and marketing programs that want to keep human experts in the loop without paying for them on every line. Access is the main thing to plan for: Evolve is an enterprise engagement priced by quote, with an eligibility check. We rate it 7.3/10 on our scale, a score that mostly reflects quote-only pricing and a sales-led setup. We have not used Evolve ourselves; this review is based on RWS announcements and customer press releases.
Dell’s 400 Million Words and the Human Last Step
- Dell runs more than 400 million words a year in 27 core languages through Evolve under a multi-year agreement.
- The pipeline is automatic up to the last step: neural MT, quality estimation and private LLM post-editing, then human specialists finalize flagged segments.
- AGCO uses Evolve for technical documentation in 32 languages and training content in 16 core languages, serving customers in 140 markets.
- There is no published price; RWS checks eligibility and signs enterprise, often multi-year, contracts.
- RWS launched Language Weaver Pro with Cohere in March 2026, which it says ranked first in 31 of 32 languages in its benchmarking.
Two Big Customer Deals Behind Evolve’s Record
Evolve’s public record is built on two large customer deals and RWS’s own figures about its people and technology.
| Fact | Detail | Source and Date |
|---|---|---|
| Launch | Released after a beta program that included Dell; available in several language pairs at launch | RWS announcement, January 25, 2024 |
| Dell deployment | Multi-year agreement covering more than 400 million words a year in 27 core languages | Silicon Canals, viewed October 2026 |
| AGCO deployment | Technical documentation in 32 languages; service and training content in 16 core languages, with new languages added each quarter; customers in 140 markets | Business Wire, July 15, 2025 |
| Human specialists | More than 1,750 in-house language specialists and domain experts | RWS announcement, January 25, 2024 |
| Wider network | 250,000 data specialists, cultural and language experts and domain professionals | RWS via Business Wire, March 25, 2026 |
| Patents | More than 45 AI patents | Business Wire, July 15, 2025 |
| Customer reach | More than 80 of the world’s top 100 brands; 60+ locations on five continents | Business Wire, July 15, 2025 |
| Underlying MT | Language Weaver Pro, built with Cohere on a model of more than 100 billion parameters, ranked first in 31 of 32 languages in RWS’s benchmarking | RWS via Business Wire, March 25, 2026 |
| Listing | Publicly listed on London’s AIM market under RWS.L | RWS announcement, January 25, 2024 |
The customer deals tell buyers more than the feature list. Dell’s 400 million words a year and AGCO’s 32-language documentation program are the kind of volumes where AI triage pays off: if quality estimation can confirm that most segments are fine, specialists only touch a fraction. Both customers chose multi-year terms, which suggests the economics are measured over years rather than in a quick pilot.
The Language Weaver Pro benchmark is an RWS claim, and it describes the translation engine family rather than Evolve itself. It matters because Evolve’s first step is Language Weaver output: better raw machine translation means fewer segments for the LLM and the specialists to fix.
RWS Technology and Linguists Sold as One Service
Evolve is not a standalone app. It is a combination of RWS technology and RWS people delivered as one service, with each component handling a stage of the job.
The Pipeline
- Translate: Language Weaver, RWS’s neural machine translation, produces translations in real time and adapts through feedback loops between quality assessment, generated output and expert-reviewed translations.
- Score: a machine translation quality estimation model, built on expert-labeled data, assesses and prioritizes each segment automatically.
- Post-edit: a private large language model rewrites the segments that need improvement, without a human in that step.
- Finalize: RWS language specialists review and finalize the output, focusing on the content that still needs attention.
- Manage: Trados Enterprise runs the end-to-end workflow across content types, repositories, translation methods and stakeholders.
Where It Has Been Used
At Dell, Evolve is a core part of the company’s AI-driven localization strategy, covering marketing materials, technical documentation, support resources and multimedia. Dell’s Director of Translation, Wayne Bourland, described its translation supply chain as highly complex. At AGCO, the agricultural equipment maker, Evolve localizes all technical documentation into 32 languages and service and employee training content into 16 core languages, with several new languages added each quarter, so customers in 140 markets get support content in their own language.
The Company Behind Evolve
RWS has been in the language business since 1958, operates from more than 60 locations on five continents and is publicly listed on London’s AIM market. It organizes its business into Generate, Transform and Protect segments. Evolve was introduced by Thomas Labarthe, then President of Language and Content Technology, and the AGCO deal was announced by Mark Lawyer, President of Regulated Industries and Linguistic AI. By March 2026, Ben Faes was CEO, and he described Language Weaver Pro as built to understand culture, context and compliance as well as words.
That history explains the shape of the product. RWS owns a translation management system, a machine translation engine and one of the largest in-house linguist teams in the industry, so Evolve stitches those assets into one pipeline instead of relying on partners. The upside is a single accountable vendor. The trade-off is that buyers who already use other tools for one of those layers have less room to mix and match.
Eligibility Checks and Onboarding Through Trados
RWS asks prospective customers to check eligibility, and at launch Evolve covered several language pairs with more planned. Onboarding runs through RWS rather than a self-serve signup, and the work sits in Trados Enterprise, so companies already on Trados have less to change.
RWS Language Specialists Finalize Every Job
Every job ends with RWS language specialists, who finalize the segments the system routes to them. Customers still own terminology, style decisions and sign-off for regulated content. RWS calls the approach a fusion of human and artificial intelligence, and the product depends on that human layer.
Evolve Pricing by Quote on Multi-Year Agreements
RWS does not publish Evolve pricing. Both public customer deals are multi-year agreements, and prospective customers start with an eligibility check.
| Plan | Price | Billing | What’s Included | Limits |
|---|---|---|---|---|
| Evolve enterprise program | Custom quote | Contract, often multi-year | Language Weaver MT, quality estimation, private LLM post-editing, RWS language specialists, Trados Enterprise workflow | Supported language pairs and volumes set by contract |
Translation Spend at a Dell-Sized Word Volume
With no list price, the useful arithmetic is about scale. These examples use Dell’s published volume; the per-word rates are illustrations for comparing quotes, not RWS prices.
- Monthly volume at Dell’s scale: 400 million words a year is about 33.3 million words a month, or about 14.8 million words per language a year across 27 languages.
- Sensitivity to the rate: at that volume, every tenth of a cent per word adds $400,000 a year, so a small difference between quotes becomes a large number.
- A smaller program: a company translating 10 million words a year would see each tenth of a cent per word add $10,000 a year.
Ask RWS to separate the price of segments that pass quality estimation, segments the LLM post-edits and segments a specialist finalizes. The share that reaches a specialist is the main cost driver in any AI-plus-human model, and it should fall as the system learns.
RWS Evolve Compared With LILT, Smartling and DeepL Agent
Evolve competes with other systems that mix AI translation and human review, and, at the edges, with AI tools from translation specialists.
| Agent | Starting Price | Free Plan | Main Strength | Main Trade-Off | Best For |
|---|---|---|---|---|---|
| RWS Evolve | Custom quote | No | Automated triage and post-editing backed by 1,750+ RWS specialists | Enterprise-only, no published price | Very large documentation and support programs |
| LILT | Custom quote | No | Assist agent that routes high-stakes content to experts | No published pricing | Enterprises and public sector agencies |
| Smartling AI Agents | $49,000 for 12 months on AWS Marketplace | No | LQA Agent and a long certification list | Quote-only for most buyers | Large brands with strict security reviews |
| DeepL Agent | Custom quote | No | General business agent built on DeepL language models | Not a translation production system | Multilingual teams automating business tasks |
LILT is the closest match: both route segments between AI and people. LILT puts more of that decision in an agent customers interact with, while Evolve packages it as a managed service. Smartling sells a platform with agents that customers run themselves. DeepL Agent sits in a different lane, handling multilingual business tasks rather than translation production. For more options, see our AI agents for translation and enterprise AI agents guides, or compare Phrase AI Agents if you want published prices.
To compare these fairly, run the same sample through each, ideally a few thousand words of your hardest content in your hardest language pairs. Record how many segments each system sent to a human, how long review took and how many errors your own reviewers still found. That turns quotes built on different models into numbers you can line up.
Global Manufacturers Translating at Huge Scale
- Global manufacturers and hardware companies: Dell and AGCO use Evolve for technical documentation, support content and training material in dozens of languages.
- Companies with hundreds of millions of words a year: AI triage makes the biggest difference at that scale.
- Teams that want human experts accountable for output: RWS specialists finalize the content, and RWS employs more than 1,750 of them.
- Existing Trados Enterprise customers: Evolve runs its workflow in Trados, so there is less to migrate.
- Buyers who prefer a managed service: Evolve combines software and language services in one engagement.
What these buyers share is volume and complexity. Evolve is built for programs where a modest improvement in the share of segments that skip human review is worth hundreds of thousands of dollars a year.
SMBs and Do-It-Yourself Localization Teams
- Small and mid-size businesses: Evolve is an enterprise program with an eligibility check.
- Teams that want to buy software and run it themselves: platforms such as Phrase publish prices and offer trials.
- Teams with rare language pairs: Evolve launched with several language pairs and expanded over time, so confirm coverage first.
- Companies that need translation as a self-serve API: a machine translation engine is simpler for that job.
Segment Rates, Commitments and Threshold Control
- Pricing by quote: a list price is not published, so ask for separate rates for automated, post-edited and human-finalized segments.
- Long commitments: both public deals are multi-year. Build in quality targets and exit terms.
- Vendor-reported performance: figures such as Language Weaver Pro’s benchmark results come from RWS. Ask for a pilot on your content.
- Threshold control: quality estimation decides what humans see. Agree who sets the thresholds and how often they are reviewed.
- Language coverage: confirm every language pair you need is supported before signing.
- Bundled services: software, AI processing and human review come from one vendor, so plan how translation memories and glossaries would move if you switch.
None of this is unusual for an enterprise language contract, but Evolve’s automation adds one new variable: the share of content that never reaches a person. Make that share visible in monthly reports, because it drives both cost and risk.
Language Weaver Pro Launch and the AGCO Deal
- On March 25, 2026, RWS launched Language Weaver Pro, built with Cohere and integrated into Trados; RWS said it ranked first in 31 of 32 languages in its benchmarking.
- On July 15, 2025, AGCO selected Evolve under a multi-year agreement to localize technical documentation into 32 languages and training content into 16 core languages.
Credible for Huge Programs, With a 4.6 Score
For a global enterprise translating hundreds of millions of words, Evolve is a credible option with real deployments behind it. It scores 7.3/10 on our scale, a number that reflects quote-only pricing and a sales-led setup rather than the technology. The pipeline automates translation, scoring and post-editing, and keeps RWS specialists on the segments that need them.
For smaller programs, other products may fit better. If you cannot commit to an enterprise engagement, a platform with published prices and a trial will be faster to test. If you can, ask RWS for a pilot that reports how many segments reach a human and how that share falls over time.
The Dell and AGCO deals are the strongest evidence in Evolve’s favor. Both companies run complex, multi-language programs, both chose multi-year terms, and Dell took part in the beta before signing. Ask RWS for references from customers with content and languages similar to yours.
What Localization Leads Want to Know About Evolve
How Much Does RWS Evolve Cost?
RWS does not publish Evolve pricing. Prospective customers check eligibility with RWS, and public customers such as Dell and AGCO have signed multi-year agreements.
Does RWS Evolve Use Human Translators?
Yes. After machine translation, quality estimation and LLM post-editing, RWS language specialists finalize the output. RWS says it has more than 1,750 in-house language specialists and domain experts.
Which Companies Use RWS Evolve?
Dell, which took part in the beta, signed a multi-year deal covering more than 400 million words a year in 27 languages. AGCO uses Evolve for technical documentation in 32 languages and training content in 16 core languages.
What Is the Difference Between Evolve and Language Weaver?
Language Weaver is RWS’s neural machine translation, and it is one component of Evolve. Evolve adds quality estimation, private LLM post-editing, human specialists and Trados Enterprise workflow on top.
How Does RWS Evolve Compare With LILT?
Both combine AI translation with human review for high-value content. LILT offers an agent, LILT Assist, that routes content and reports to customers directly, while Evolve is delivered as an RWS service built on Trados Enterprise and Language Weaver.

User Reviews
Nobody has rated it here yet. Be the first AI Agents Choice user to rate RWS Evolve.