Is Kofax the same company as Tungsten Automation?
Yes. Kofax rebranded to Tungsten Automation on 16 January 2024, and it was a name change rather than a sale: ownership did not change, and existing contracts, licences and support entitlements were unaffected (Tungsten Automation press release, 2024). Products took a Tungsten prefix, so Kofax TotalAgility is now Tungsten TotalAgility and Kofax Capture is Tungsten Capture, while review sites, analyst records and procurement systems still index the company under Kofax.
A second name causes more confusion than the first. There was a separate, UK-listed Tungsten Corporation plc operating the Tungsten Network e-invoicing platform. Kofax acquired it in June 2022 for around £70.6 million and adopted its name for the whole group eighteen months later. Tungsten Network is a product line inside Tungsten Automation today, not a competing company.
Ownership is US private equity. Founded in 1985 in Irvine, California, the company passed through a public listing, DICOM Group, Lexmark in 2015 for roughly $1 billion and Thoma Bravo in 2017, and has been held by Clearlake Capital and TA Associates since July 2022 (TA Associates portfolio, 2026). It reports more than 25,000 customers and roughly 2,200 employees, and was named a Leader in Gartner's inaugural 2025 Magic Quadrant for Intelligent Document Processing. anyformat is a European company backed by a €3.3 million seed round; on installed base and platform breadth this is not a close comparison, and the rest of this page is about the axes where it is.
Extraction approach and training sets
Kofax's core extraction pipeline is template and rules based with supervised machine learning that you train, while anyformat extracts zero-shot against a schema you define. In TotalAgility, Transformation Designer projects are built around training documents, a Classification Set and an Extraction Set (Tungsten TotalAgility 2026.1 documentation), so adding a document type means collecting representative samples, labelling them, training, testing and iterating. That approach produces a precisely tuned classifier for stable, high-volume document populations, which is exactly the mailroom-scale AP and claims work the platform was designed for.
Tungsten does ship a generative alternative. Copilot for Extraction in DocAI Studio lets you describe fields in natural language and extract without training data. It is a newer capability inside one design surface rather than the default path across the older ReadSoft and AP Essentials lines, so the labelled-set workflow is still what most existing deployments run.
anyformat inverts the order. Define the fields in a schema, upload the document and receive validated JSON, with no labelled set and no training cycle. It accepts 100+ formats without templates, combines large language models with deterministic rules, and applies schema changes immediately without a code deploy. On a parsing benchmark of 1,000+ real documents spanning 30+ document types, anyformat scored 78.1% (anyformat benchmark, 2026), and on line-item tables growing from 1 to 50 pages and roughly 2,400 rows it held around 99% row recovery (anyformat benchmark, 2026).
Time to a first production workflow
A TotalAgility rollout is an implementation project measured in months and usually staffed with a systems integrator, whereas anyformat is self-serve from a free account. The platform is deliberately configurable, which is also the source of the friction: reviewers say it is not truly low code yet, and report that custom development is still needed every few projects (PeerSpot, 2026). Reviewers on Capterra also describe sandbox environments that did not match production, which lengthens the test-and-release cycle further (Capterra, Kofax TotalAgility reviews, 2026).
anyformat ships the design surface to the team that owns the documents. Building a production workflow with the Annie assistant takes around 5 minutes against roughly 4 hours of manual configuration (anyformat internal benchmark, 2026), and the Pay As You Go plan starts free with a one-time grant of 50,000 credits, so a first extraction happens before any contract exists. There is no integrator in the loop and no professional services engagement to schedule a schema change around.
Deployment: where Kofax wins
Kofax is stronger than anyformat here, and it is not close. TotalAgility deploys on-premises, in private cloud, in Docker containers and as Azure-hosted SaaS with dedicated tenancy, on an architecture that has run disconnected from any vendor cloud for four decades. Around it sits a capture ecosystem no AI-native vendor matches: scanner and MFP device support, mailroom and fax ingestion, and the drivers and device certifications that come with 40 years of selling into physical document operations.
anyformat also offers cloud, private cloud or VPC, and on-premise deployment including air-gapped environments on the Enterprise plan, in a modern containerised install that does not require the infrastructure footprint of legacy capture software. The honest difference is maturity and surface area, not the checkbox: if your requirement includes scanner fleets, MFP panels or a physical mailroom, Kofax has an ecosystem for it and anyformat does not.
Compliance and public sector: where Kofax wins
For US federal and state government, Kofax is ahead of anyformat by a wide margin. Tungsten TotalAgility Cloud holds a FedRAMP High Authority to Operate and has been listed as FedRAMP Authorized since 19 March 2026, the platform is IRAP assessed at PROTECTED level in Australia, and the company states it is pursuing StateRAMP/GovRAMP authorization so SLED buyers get a comparable baseline (Tungsten Automation, 2026). anyformat holds none of those authorizations, and no amount of product differentiation substitutes for an ATO when the buyer is a federal agency.
On the general certifications the two are closer. Tungsten states on its trust center that it is certified to ISO/IEC 27001 and maintains SOC 2 Type II; the trust center renders client-side and we were not able to retrieve the underlying attestations directly, so treat that as a vendor statement rather than a document we verified. anyformat is ISO 27001 certified with scope covering the end-to-end document pipeline.
The gap runs the other way on European requirements. We found no EU-region hosting commitment and no zero-retention statement in Tungsten's public material, and its GDPR posture reads as generic programme language rather than architecture. anyformat is EU-native, and zero-retention processing is a single toggle on the Enterprise plan: documents processed, data returned, source files gone.
Total cost and implementation
Kofax publishes no pricing; every route is contact sales, historically a perpetual licence plus annual maintenance and now with subscription options, with implementation and professional services as a separate and substantial line item. Third-party directories carry self-reported starting figures that we could not corroborate and do not repeat here. What is documented is sentiment: on Capterra, the two lowest sub-scores for Kofax TotalAgility are Value for Money and Ease of Use, both 3.8 out of 5, and reviewers on PeerSpot describe the platform as expensive and parts of its console as dated (2026).
anyformat publishes its rate card. Billing is credit-based, charged per page and per operator: 10 credits cost €0.01, so Parse is 25 credits per page (€0.025, published as $25 per 1,000 pages), Extract 35 credits per page, Classify 10, Split 25, and Validate 25 credits per rule (anyformat pricing, 2026). Pay As You Go is free to start with a one-time grant of 50,000 credits, Business is €499 per month (€399 billed annually) for 500,000 credits, 5 users and 10 workflows, and Enterprise is custom-priced with unlimited users and workflows, VPC or on-premise deployment, SSO, RBAC and SAML, and zero data retention. The comparison that matters is not licence against credits, it is licence plus integrator against credits plus an afternoon.
Evaluation and accuracy monitoring
Kofax measures extraction quality at design time; anyformat measures it continuously against ground truth in production. TotalAgility gives you Classification Benchmark runs and test results inside Transformation Designer, which tell you how a trained project performs on a labelled set before release, and Tungsten Insight adds process analytics over throughput and cycle time. We found no continuous ground-truth accuracy monitoring and no drift alerting in their documentation, which means the question of whether last quarter's project is still right on this quarter's documents is answered by exception volume rather than by measurement.
That is the axis anyformat is built around. Every field carries calibrated confidence, measured at 99.1% calibration accuracy with an Adaptive ECE of 0.009 (anyformat benchmark, 2026), so a reported 90% means correct about 90% of the time and automation thresholds behave predictably instead of optimistically. Every extracted value carries a visual citation linking it to its exact position on the page for audit and review. On complex mixed-format invoices, 94% were extracted perfectly, every field, every line item, totals reconciled (anyformat benchmark, 2026).
An evaluation suite that runs after go-live, and a design surface owned by operations rather than by an integrator.
Evaluations, available today. Every anyformat Extract and Classify workflow has a Health tab. You build a dataset with verified ground truth, either by promoting a document the workflow already processed and confirming its values in the review interface, or by uploading labelled documents with their expected JSON. Files are taggable into sub-datasets, so accuracy reads per provider, per document type or per difficulty slice instead of as one flattering average. An evaluation re-extracts every in-scope document against a chosen workflow version and scores it field by field, with a result-versus-expected view on each failure; runs are numbered and immutable, so the effect of a change is measured rather than assumed. Health Overview then places the dataset benchmark next to live production accuracy, confidence and through rate, and reports the gap between them (Evals). Kofax's benchmarking is design-time and project-scoped: it scores a trained project against a labelled set before you release it, with no continuous production comparison, no sub-dataset slicing and no immutable run history.
Optimizer, announced and not yet available. anyformat has announced Optimizer, a workflow that tunes itself against its own dataset using your evaluations as the target. It is on the roadmap, not in the product today (Evals).
The pipeline you do not have to build. This one needs stating carefully, because Kofax genuinely ships the pipeline: Process Designer is a mature visual BPM and case-management canvas that predates every AI-native competitor, with RPA and capture in the same platform. The difference is who assembles it and how long that takes. A TotalAgility deployment is configured by IT or an integrator across Transformation Designer projects, Classification and Extraction Sets and a process canvas, over months. anyformat's Studio is a no-code workflow builder with branching, conditions, splitting, routing, extraction operators and built-in human-in-the-loop review, changed by the operations team without a code deploy. If your requirement is orchestration breadth across a whole enterprise, Kofax has more of it. If your requirement is a document workflow that the people running it can change on a Tuesday, that is the trade.
When to choose Kofax (Tungsten Automation)
Choose Kofax when on-premise or disconnected deployment is a hard requirement and you want a vendor with four decades of it, when you are a US federal or SLED buyer who needs FedRAMP High today, or when Australian PROTECTED-level data is in scope. Choose it when your scope is broader than extraction and you want BPM, RPA and capture from one vendor under one contract, when scanner fleets, MFP devices or a physical mailroom are part of the flow, and when SAP-centric AP through ReadSoft is already in your landscape. Choose it when your document types are stable and high volume, so the cost of building labelled Classification and Extraction Sets amortises properly, and when you have the budget and the integrator capacity for a months-long implementation. It is a Leader in Gartner's 2025 IDP Magic Quadrant for reasons that are real.
Choose anyformat when your document types change faster than you can label them, so zero-shot schema extraction beats a training cycle; when you need to know whether accuracy is still holding this month, not whether the project passed a benchmark at release; and when European data residency, GDPR by architecture and zero-retention processing are regulatory obligations rather than questionnaire answers. Choose it when you want the operations team, not IT or an integrator, to own and change the workflow, when you want published per-page pricing instead of a quote, and when calibrated confidence with visual citations is what makes straight-through processing defensible in an audit.
Teams evaluating a Kofax replacement most often cite the labelled-training cycle, the implementation timeline and the absence of continuous accuracy measurement as the deciding factors. Teams that stay cite on-premise maturity, FedRAMP and the breadth of the platform, and both sets of reasons are legitimate.