anyformat vs Parseur (2026)

Choose Parseur for SMB email and document parsing that feeds a Zapier or Make automation; choose anyformat when confidence-routed straight-through processing, workflow orchestration and EU jurisdiction are requirements. Parseur's public document API returns no confidence field at all, while anyformat calibrates confidence per field at 99.1% calibration accuracy (anyformat benchmark, 2026).

What is Parseur and who is it for?

Parseur is a cloud parsing service for emails, PDFs and documents, founded in 2016 and operated by Parseur Pte Ltd, a bootstrapped and profitable company incorporated in Singapore whose terms specify Singapore law and exclusive Singapore jurisdiction (Parseur terms, 2026). The team is small and the product is built for SMB and mid-market operations: a mailbox address that receives documents, a point-and-click template editor, and delivery into a spreadsheet, a webhook or a no-code automation. Users rate it well for that job, 4.8/5 across 63 Capterra reviews, with support among its highest sub-scores (Capterra, 2026). Parseur also publishes an unusually large content surface for a company of its size, more than 1,000 URLs in roughly 11 languages (Parseur sitemap, 2026), so it surfaces often in search and AI-generated shortlists. That reach measures content coverage rather than product scope, and the two are worth reading separately.

anyformat is built for a different buyer: teams running 500+ documents a month across varied layouts, where extraction has to be orchestrated, measured, reviewed and defended in an audit. Most of what follows is a segment difference rather than a quality judgment.

Template engines, the AI engine and the maintenance loop

Parseur runs three extraction engines in a fixed priority order, text template first, then OCR template, then AI, and a template wins whenever one matches (Parseur help docs, 2026). The AI engine, added in October 2023, is genuinely template-less: you name the fields you want and no sample document is required. The priority order means the template path stays the default for the recurring senders most accounts are built around, and templates are the part that needs upkeep. Public reviews describe the same loop repeatedly: a sender changes a layout, a template that worked yesterday stops matching, and someone rebuilds it. Parseur's own help articles also state different page ceilings for the AI engine in different places, so the working limit on long documents is something to test against your own files rather than read off a page.

anyformat has no template layer. You define a schema, upload the document and receive validated JSON zero-shot across 100+ formats, combining LLMs with deterministic rules, and schema changes apply immediately without a code deploy. On a 2026 parsing benchmark across 1,000+ real documents and 30+ document types, anyformat scored 78.1% on a combined parse score (anyformat benchmark, 2026); that study covered frontier models and dedicated document-AI engines, and Parseur was not among them. Long documents are where the two architectures diverge most: anyformat held ~99% row recovery on line-item tables running to 50 pages and roughly 2,400 rows, and extracted 94% of complex invoices with every field, line item and total correct (anyformat benchmark, 2026).

Confidence scoring and review routing

Parseur does not return a confidence score. Its public "Get a document" API schema returns processing status, credits consumed, engine metadata and OCR page data, and contains no confidence, accuracy or probability field (Parseur developer docs). Third-party articles describing confidence thresholds inside Parseur trace back to generic educational blog content rather than to the product. The practical consequence is that there is no per-field signal to threshold on: a document parses or it does not, and deciding which results a human should look at falls to rules you write downstream against the values themselves.

anyformat returns calibrated confidence per field, measured at 99.1% calibration accuracy with an Adaptive ECE of 0.009 (anyformat benchmark, 2026), so a field reported at 90% is correct roughly 90% of the time. Every value also carries a visual citation linking it to its exact position on the page. Calibration is what makes straight-through processing auditable: high-confidence fields post automatically, uncertain ones route to a reviewer, and the threshold is a number you can defend to a controller or an auditor.

Workflow orchestration

Parseur states its own position here. Its blog says that Zapier, Make and n8n handle workflow automation while Parseur focuses on data extraction (Parseur blog). Classification, conditional branching, splitting, routing, retries and human approval therefore live in a second tool, with its own subscription, its own error handling and its own audit trail. The frequently quoted "10,000+ apps" figure describes the reach of those ecosystems rather than native Parseur integrations.

anyformat includes Studio, a no-code visual workflow builder with branching, conditions, splitting, routing, extraction operators and built-in human-in-the-loop review, alongside a REST API, webhooks and pre-built connectors for Salesforce, SharePoint, Google Drive, Dropbox and OneDrive. Operations teams change a workflow on the canvas, and the change needs neither a code deploy nor a second platform to debug.

Compliance, jurisdiction and deployment

Parseur stores EU customer data in the Netherlands but is a Singapore company, and for regulated buyers that distinction is the whole question. Its ISO 27001 statement, in its own wording, describes the data centre where its servers sit rather than Parseur itself (Parseur GDPR page, 2026). SOC 2 is not yet issued: Parseur states it has finished its preparations, is under internal review with its compliance partner and expects a report by Q3 2026 (Parseur security page, 2026). It states plainly that it is not HIPAA compliant and will not sign Business Associate Agreements until that changes. It appoints an Irish Article 27 representative precisely because it is a non-EU processor, and its terms name Singapore law and exclusive Singapore jurisdiction. Retention is configurable down to one day, with an opt-in process-then-delete mode and backups purged within 45 days, which is controllable but not zero by default. Deployment is cloud-only: no self-hosted, VPC or on-premise option is documented at any tier, Enterprise included.

anyformat is EU-native and ISO 27001 certified across the document pipeline itself rather than through a hosting provider, offers zero-retention processing as a single toggle, and deploys in cloud, private cloud or air-gapped on-premise environments. For organizations under GDPR or DORA, the operative question is which law governs the contract, not only where the disks sit.

Pricing and credit rollover

Parseur charges one credit per page: a free tier of 20 pages per month with 90-day retention, a Base plan in the low tens of euros per month up to 3,000 pages, a Scale slider running to 1M pages and up to 100 users, and a custom Enterprise tier up to 10M pages. Credits do not roll over (Parseur pricing, 2026). For volumes that arrive in peaks, month-end billing runs or seasonal claims, that structure charges for the plan you sized for the peak in every month you do not reach it. Value for money is Parseur's lowest Capterra sub-score at 4.5/5, and reviewers point at per-page cost at volume as the reason.

anyformat bills in credits 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 and zero data retention. Because operators are priced separately, a workflow that only parses costs less per page than one that also classifies, validates and splits, so the bill follows the work rather than a flat page count.

What does anyformat ship that Parseur does not?

An evaluation suite, and the production layer around extraction. Getting a document parsed is one variable; keeping the output correct after go-live is where document projects spend most of their budget.

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 sender, per document type or per difficulty slice instead of as one 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). Parseur publishes no evaluation or accuracy-measurement surface; treat that as not documented rather than as a stated absence.

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. Parseur returns parsed fields and forwards them onward. Production also needs classification and routing, validation rules, retry handling, a human-review interface, accuracy monitoring, and re-tuning whenever a sender changes a layout or a model changes underneath you. In a Parseur deployment most of that is assembled in Zapier, Make or n8n plus whatever you write yourself, which is exactly the split their own blog describes. anyformat ships that layer as product surface: no-code Studio workflows, per-field calibrated confidence with review routing, visual citations for audit, and the evaluation loop above. Setting up a production workflow with the Annie assistant takes around 5 minutes against roughly 4 hours of manual configuration (anyformat internal benchmark, 2026).

When to choose Parseur

Choose Parseur when a small operations team needs a narrow set of recurring documents or emails turned into rows, and the automation around them already lives in Zapier, Make or n8n. Its email ingestion is mature, the free tier is a real 20 pages a month rather than a time-limited trial, the template editor is usable without engineering help, and its support is rated at the top of its category by its own users. If EU jurisdiction, SOC 2, HIPAA, on-premise deployment, confidence-based routing and native orchestration are not on your requirements list, the segment fit is good and the entry price at low volume is hard to beat.

When to choose anyformat

Choose anyformat when documents are varied rather than repetitive, when you need calibrated per-field confidence to decide what a human looks at, when classification, branching, validation and review have to sit in one auditable system instead of a chain of tools, or when EU jurisdiction, ISO 27001 over the pipeline itself, zero retention and on-premise deployment are procurement requirements rather than preferences.

The two products are sized for different buyers: Parseur for a team that wants recurring documents into a spreadsheet this afternoon, anyformat for teams that have to measure, route and defend the output at scale.