- Starting price
- Free; Pro coming soon
- Free tier
- Yes
- Platforms
- Web browser
- Developer
- Frontier Research and Development
- Launched
- 2025
- Updated
- Aug 9, 2026
The verdict
Aquilla is an unusually thoughtful translation workspace: source and target cells, live collaboration, AI suggestions, back-translation, checks, terminology, audio and human validation form one coherent operating loop. Its free workspace and documented field projects make it worth a serious pilot. Organizations handling unpublished Scripture or sensitive language data should not move production work until Frontier supplies the privacy, security, subprocessor, retention, ownership and service commitments that the public site currently does not.
Try Aquilla ↗Opens aquilla.app
Aquilla is a browser-based translation-operations platform from Frontier Research and Development, a mission-oriented technology company led by founder and CEO Ryder Wishart. It is meant for teams translating text and audio in low-resource and community settings, not merely for an individual pasting paragraphs into a generic AI chatbot. Projects pair source and target content in structured cells; translators, reviewers and project leads work in the same workspace; AI can propose a draft or back-translation; and checks, terminology, comments, assignments and validation keep human judgment visible throughout the process.
The product succeeds an earlier Codex extension that lived in Visual Studio Code and stored work through GitLab. Aquilla moves that model to the web so a contributor does not need to install a developer editor or understand repositories. Frontier describes it as an agentic platform whose assistants can draft, check, remember and coordinate while experts steer. Text and audio are available now. Captions, subtitles, video, image and oral-story workflows appear on the roadmap, so buyers should distinguish shipping capabilities from future scope.
The published entry offer is striking: the whole workspace is described as free forever, with no card or trial clock, including text and audio translation, real-time guidance, back-translation, cloud sync, collaboration and on-device speech. Pro is marked “coming soon” with higher AI limits and pay-as-you-go usage, but there is no current amount, allowance or launch date. Enterprise help is available by contacting the team. A no-cost platform can still create material migration and governance costs, particularly when it becomes the source of truth for a multi-year language program.
The 4.5 score above is an editorial assessment based on the public product, help center, Frontier materials and vendor case studies. It is not a customer-review aggregate and it does not imply that we translated a corpus, entered a protected project, measured output quality or inspected the infrastructure. Aquilla has compelling design and meaningful reported deployments. The most consequential limitation is not a missing convenience feature: we could not locate substantive public privacy, terms or security documentation. Protected-looking routes that return a generic application shell are not a usable legal notice. That absence matters when source manuscripts, recordings, identities and community language knowledge may be confidential.
✓ The good
- Translation-specific workspace - paired cells, status, validation and review are more controlled than general chat prompts
- Human judgment stays central - translators can accept, edit or ignore AI suggestions and reviewers explicitly validate work
- Useful free offer - the current workspace includes real collaboration, text, audio, back-translation and cloud sync without a trial clock
- Low technical barrier - browser access replaces the developer-oriented setup required by the earlier Codex extension
- Quality context extends beyond one sentence - terminology, surrounding content and project translation philosophy can inform suggestions
- Live collaboration and roles - organizations can separate viewing, commenting, contribution, review, project leadership and ownership
- Audio is not an afterthought - recording, transcription and browser speech features support oral and low-literacy workflows
- Credible implementation stories - Frontier documents extensive Come and See and Biblica projects rather than relying only on mockups
✗ Watch out
- Public legal information is missing - we found no substantive privacy policy or terms defining the data relationship
- Security and AI vendors are undisclosed - public pages do not identify model providers, subprocessors, hosting regions or assurance reports
- Future price is unknown - Pro limits, pay-as-you-go rates, launch date and enterprise costs are not published
- Migration can be one-way - after a Codex import, Aquilla becomes the source of truth and GitLab no longer synchronizes
- Roadmap language is easy to overread - video, image, captions, subtitles and oral stories are not all generally available now
- Case studies are vendor-published - impressive project claims are evidence of use, not independent quality benchmarks
- AI output still requires expert review - fluent wording, back-translation and automated checks cannot establish faithful meaning by themselves
- Young platform risk remains - organizations need export, backup and continuity plans before centralizing irreplaceable work
Best for
- Bible and mission translation teams coordinating text or audio with distributed reviewers
- Community translation projects that need a browser interface for nontechnical contributors
- Organizations moving from Frontier’s Codex and GitLab workflow
- Teams that want AI assistance inside a structured human validation process
- Pilot projects able to require contractual privacy and security answers before sensitive production use
Avoid if
- Your procurement policy requires public legal terms, a DPA or named subprocessors before evaluation
- You need production-ready video, image or subtitle workflows today
- Your team wants fully automatic translation without qualified community review
- You cannot tolerate an unpublished future price for heavier AI usage
- You need a bidirectional ongoing sync with the legacy Codex GitLab repository
What Aquilla is
Aquilla is a collaborative production system for translation. A project holds aligned source and target cells, progress and quality state. Contributors draft or record work; an AI copilot can suggest language or produce a back-translation; built-in checks flag potential problems; reviewers inspect and validate; terminology and comments preserve decisions; and project health views help leads find stalled or decaying work. This is closer to a lightweight translation-management system than a one-click translator.
Organizations group projects and members under shared billing and control. Published roles span Viewer, Commenter, Reviewer, Contributor, Project lead, Maintainer and Owner, allowing access to follow responsibility. The web interface removes the earlier Codex dependency on Visual Studio Code and GitLab. Exports remain important: a cloud workspace should improve coordination without becoming the only usable copy of years of community knowledge.
Why translation teams may prefer Aquilla to a general AI translator
A general model sees a prompt and returns text. Aquilla sees a project: the source segment, neighboring material, terminology, prior decisions, review state and the team’s translation philosophy. Frontier’s Living Memory concept is intended to retrieve semantically relevant examples while considering context rather than mechanically repeating the nearest stored sentence. That can make suggestions more consistent and reduce the repeated explanations that consume a translator’s day.
The deeper advantage is accountability. A draft remains a proposal inside a cell, not an answer that silently becomes Scripture. Translators can edit or reject it, back-translation exposes what a target sentence may communicate, automated checks create review work, and a designated reviewer validates the result. None of those controls proves accuracy. Meaning, naturalness, cultural resonance, oral comprehension and community ownership require qualified people. Aquilla is strongest when it shortens their administrative work while making their decisions easier to trace.
Source and target cells: a shared production line instead of scattered files
Aquilla presents source and target content in paired cells. Each unit can hold text, work state, comments, checks and review decisions, while navigation preserves the wider project. Live presence lets contributors see one another rather than emailing documents or reconciling separate spreadsheets. Project health and queues help a lead identify untranslated cells, unresolved warnings and material awaiting validation. Structured units also make links, assignments and exports more dependable than a pile of differently named files.
The benefit is not merely speed. A team can establish who changed a unit, why a term was chosen and whether a reviewer accepted it. That history supports continuity when volunteers rotate or a language consultant joins late. The risks are fragmentation and false completion: segmenting discourse into cells can hide paragraph-level flow, and a green status can feel more authoritative than it is. Reviewers should regularly read and listen to longer passages, test with community members and treat workflow state as process evidence rather than a verdict on meaning.
AI Copilot, Living Memory and validation: assistance with visible brakes
The AI Copilot can propose drafts, revise text and produce back-translations that help a reviewer see how target wording maps toward a language they know. Validation and built-in checks surface issues for action. Frontier describes Living Memory as context-aware retrieval that can consult project examples and translation philosophy instead of preserving mistakes merely because they appeared in a conventional memory database. An agent layer is also positioned to coordinate repetitive work across a project.
These are decision aids, not substitutes for comprehension. A convincing back-translation can conceal ambiguity; an automatic check can miss a conceptual shift; retrieved examples can amplify an earlier compromise; and an agent can reproduce an instruction at scale before anyone notices it is wrong. Establish approval thresholds, evaluate output on representative difficult passages, record why exceptions are made and keep a clean rollback path. Ask Frontier exactly which content reaches external models, whether it trains any provider, how long prompts and outputs remain, and whether a project can require a particular model or disable AI completely.
Organizations, audio and Codex migration: broader access with a hard handoff
Organization membership centralizes teams and projects, while granular roles help limit changes. Audio tools can use browser text-to-speech and support recording or transcription, with documentation emphasizing on-device speech behavior where available. That makes oral checking possible without a separate studio workflow and can include contributors who communicate more naturally by listening and speaking. Cloud synchronization gives distributed teams one current workspace; integrations and automation can connect translation work to operational systems such as Monday.com.
Legacy Codex users receive a managed import from GitLab that includes project history, but Frontier documents a consequential boundary: migration is one-way and continuous during the handoff, then Aquilla becomes the source of truth and GitLab stops synchronizing. Before approval, freeze roles, inventory repositories and media, compare cell and history counts, export a recovery copy and run acceptance tests. For audio, test weak connections, noisy rooms, accents, permissions and ownership. For every role, verify whether exports, deletion, invitations and organization-wide settings follow least privilege.
Pricing
Everyone
Free forever
Aquilla says everyone can use the whole workspace without a card or trial clock. The published offer includes text and audio translation, real-time guidance and back-translation, cloud synchronization, team collaboration and on-device speech. The site does not publish a storage ceiling, project cap, member cap, support entitlement or exact baseline AI allowance, so confirm operational limits before planning a large program.
Pro
Coming soon
Pro is described as a per-project monthly option with higher AI limits and pay-as-you-go capacity. No dollar amount, included volume, overage unit, release date or migration rule is currently displayed. Teams should not build a future budget from the word “free” alone; model expected source words, target languages, back-translations, audio minutes and automated checks, then request a written scenario.
Enterprise support
Contact Frontier R&D
Organizations with deployment, migration or support needs can contact Frontier. Ask for implementation scope, support hours, response targets, uptime, backups, recovery objectives, data residency, security evidence, DPA terms, indemnity, ownership and a tested complete export. If grant funding is time-limited, request price protection and an exit schedule as well.
The current free plan is generous enough to evaluate the real product rather than a toy demo. A team can create a representative project, invite different roles, draft text, record audio, use back-translation, perform checks and test review. Use fictional or publishable content until data handling is documented.
“Free forever” describes the baseline workspace; it does not define future AI economics. Pro is expected to raise AI caps and add pay-as-you-go use on a per-project monthly basis. Translation programs should request a meter: what consumes a unit, whether source and target length both count, how audio is billed, whether retries and checks count, and what happens at the limit.
The people cost remains larger than the software line. Budget translators, community testing, consultants, review meetings, audio recording, training, terminology work and project leadership. AI may reduce first-draft and coordination time, but it does not remove responsibility for the published result.
Enterprise buyers should price governance. Legal review, security assessment, access recertification, backups, export drills and incident handling are part of operating a translation system. A free license is not a reason to accept undefined ownership or retention for an irreplaceable corpus.
If the team is leaving Codex, include migration acceptance and exit portability in the agreement. A successful import should preserve source, target, history, status, comments, users and media—or explicitly document every field that does not transfer. Keep an independently readable archive before GitLab ceases to be the current system.
Where Aquilla falls behind
The most serious gap is public legal transparency. The official marketing and help sites provide substantial product instruction but no readily usable privacy policy or terms of service naming the contracting entity, jurisdiction, customer roles, intellectual-property treatment, acceptable use, warranties or termination rights. An application route that returns a shell without legal text does not let a translator give informed consent or a procurement team assess risk.
Security information is similarly thin. Public materials do not state encryption practices, hosting provider and region, tenant isolation, access logging, administrator controls, security testing, vulnerability reporting, incident notification, backup frequency, recovery targets or independent assurance. Translation projects may expose unpublished Scripture, minority-language recordings, contributor identities and locations. Those are not details to infer from a polished interface.
AI data flow needs a diagram. Teams should know every model and speech provider, which fields leave Aquilla, whether content trains Frontier or a third party, how prompts are logged, whether human provider reviewers can see them, where processing occurs, when data is deleted and how an organization disables or constrains AI. “On-device” speech for a particular function should not be generalized to all audio and AI processing.
The product boundary is changing. Pro is not yet priced, and captions, subtitles, video, image and oral-story capabilities are presented as coming next. Roadmap momentum is encouraging, but a grant proposal or delivery contract should cite only features demonstrated in the production account. Ask for current availability, beta labels, known limits and target dates in writing.
The published case studies are valuable but are written by the vendor. Come and See reports expansion of The Chosen season one to 125 languages, including 39 additional low-resource languages and a large pipeline; Biblica describes accelerated resource and Bible-update workflows. These demonstrate that Frontier has encountered real complexity. They do not independently establish linguistic quality, productivity gain, total cost or suitability for another organization. Ask for references and define pilot success in advance.
Accessibility and low-connectivity behavior deserve more public detail. A browser removes installation barriers, yet large language files, audio, live collaboration and model calls can still be difficult on intermittent or expensive connections. Publishable conformance information, keyboard and screen-reader testing, data-use controls, offline expectations and supported-browser guidance would help organizations evaluate whether the people closest to the language can participate equitably.
Aquilla vs. Paratext vs. Scripture Forge and Lokalise
Aquilla emphasizes an approachable web workflow, AI assistance, audio and coordinated human review. Paratext is the established specialist environment for Bible translation, with deep checking, corpus and organizational workflows built across decades. Its ecosystem, training and project expectations can be heavier, but serious teams may need precisely that maturity. Aquilla’s browser experience can be easier for community contributors and fast pilots; it should not be assumed to reproduce every Paratext check, format, rights workflow or institutional connection.
Scripture Forge is a collaborative web environment associated with SIL technology that supports community checking and translation suggestions, including synchronization with Paratext in relevant workflows. It is a natural comparison when the requirement is browser-based participation tied to an established translation system. Aquilla’s integrated agent, Living Memory, audio direction and independent project model may appeal to teams seeking a unified new workspace. The choice depends on current source-of-truth systems, language support, checking methodology and data agreements.
Commercial localization platforms such as Lokalise and Transifex excel at software strings, terminology, approvals, automation and release integration. They publish broader enterprise procurement material and integrate deeply with product-development pipelines. They are not designed around Scripture discourse, oral community testing or translation-consultant practice. Aquilla is more domain-shaped; a localization platform may be better if the source is an app interface rather than a sacred or narrative corpus.
Run a blinded pilot rather than comparing slogans. Select representative narrative, poetry, discourse, names, recurring terminology and difficult audio. Have the same translators and independent reviewers use each workflow. Measure setup, draft usefulness, correction time, consistency, naturalness, comprehension, accessibility, connectivity cost, export fidelity and incident controls. The winner is the system that supports trustworthy human work over the life of the project, not the one that produces the fastest first sentence.
The bottom line
Aquilla is one of the more credible attempts to put generative assistance inside an accountable translation workflow. The product connects cells, context, terminology, back-translation, checks, roles, review and audio instead of offering a magical translate button. Its free baseline and Frontier’s reported field projects make a pilot easy to justify. The readiness question is governance. Before sensitive production use, require substantive terms, privacy documentation, a DPA, an AI and subprocessor map, retention and deletion commitments, security evidence, backups, ownership and complete export. With those protections and qualified human review, Aquilla could remove a great deal of translation friction without hiding who remains responsible for the words.
Alternatives to Aquilla
Frequently asked questions
What is Aquilla?
Aquilla is a browser-based translation-operations workspace built by Frontier Research and Development. It combines aligned source and target cells, live collaboration, roles, terminology, comments, AI draft suggestions, back-translation, checks, validation, project health and audio tools. It is designed for organized human translation teams, including mission and low-resource language projects, rather than as a general consumer translator.
Is Aquilla free?
The official site says the whole workspace is free forever without a credit card or trial clock, including text and audio translation, real-time guidance, back-translation, cloud sync, collaboration and on-device speech. A Pro plan with higher AI caps and pay-as-you-go use is coming soon, but no price or exact allowance is published. Enterprise support requires contact.
Can Aquilla translate a Bible automatically?
Aquilla can assist with drafts, back-translations, context and checks, but its documented workflow keeps translators and reviewers responsible for acceptance and validation. No AI system can establish faithful meaning, naturalness, cultural comprehension and community ownership on its own. Use qualified translators, read and listen in larger context, test with the language community and maintain an explicit human approval process.
What is the difference between Aquilla and Codex?
Codex was Frontier’s earlier Visual Studio Code extension with GitLab-based project storage. Aquilla moves the work into a browser and adds a more accessible collaborative interface. Frontier’s migration guide says its team can continuously import Codex history during transition; after cutover, Aquilla becomes the source of truth and GitLab no longer synchronizes. Preserve an archive and verify all migrated fields before that handoff.
Does Aquilla support audio?
Yes. Current public materials describe text and audio translation, recording and transcription-related workflows, browser speech features and on-device speech where supported. Test the exact browser, language, device and connectivity conditions your contributors use. Captions, subtitles, video, image and additional oral-story capabilities appear as future directions and should not be treated as generally available without confirmation.
Is Aquilla private and secure?
Public materials do not currently provide enough detail to answer that confidently. We could not locate substantive public privacy, terms or security documentation that names model providers, subprocessors, regions, training rules, retention, deletion, encryption, audits, incidents or backups. Ask Frontier for those documents, a DPA and a data-flow diagram before uploading unpublished Scripture, protected recordings or contributor information.
Who should choose Aquilla?
It is most compelling for distributed text or audio translation teams that want nontechnical browser access, AI assistance inside a human review loop and one place for roles, terminology, checks and progress. A team already deeply invested in Paratext or Scripture Forge should test interoperability and missing specialist functions. Any organization should pilot with low-risk content, verify exports and settle legal, security and future-pricing questions first.
Sources & further reading
More Greek & Hebrew Tools
Blue Letter BibleThe serious-study site for everyone who can't afford Logos - and a surprising number of people who can.
Bible HubThe most-cited free Bible study site on the internet - a thirty-translation parallel view, a Strong's-tagged interlinear, and a wall of classical commentaries on every verse, all one click apart and all completely free.
STEPBibleA free, open-source original-language Bible study site built by actual textual scholars at Tyndale House Cambridge - and the best-kept secret in serious Bible study.
BibleArcThe diagramming workbench seminarians swear by and most casual readers have never heard of - and the reason that gap exists is the same reason its fans will not give it up.Scripture4AllA quiet, donor-run project that gives the whole Bible away as a word-for-word interlinear - Hebrew and Greek under a literal English gloss, with grammatical parsing on every word, free and downloadable.
BiblingoBiblingo teaches biblical Hebrew, Greek and Latin by immersion rather than by parsing charts, and it charges for it. Fifty-two lessons per language, a Hebrew University doctorate behind the curriculum, and one uncomfortable fact - every tool it competes with is free.