Textkernel & Sovren alternative
Textkernel’s Tx Platform is the API many Sovren customers now run on. It goes deep: ISCO and O*NET normalization, 29 resume languages, Search & Match and a skills ontology, each billed as credit add-ons. HireLayer is smaller and flatter: normalized skills and a geocoded location are part of every one-credit parse, and matching runs per request without an index.
A Bullhorn-owned recruiting AI vendor whose Tx Platform (the API that absorbed Sovren) offers parsing, Search & Match and skills intelligence.
HireLayerMigration from Textkernel
ModerateKey differences
Textkernel acquired Sovren in 2021 and became part of Bullhorn in 2024. Its developer platform is mature and broad; the questions are what each add-on costs you and how its usage rules fit your traffic.
A Tx parse costs 1 credit, then skills normalization adds 0.1, job title normalization 0.2, geocoding 1 (0.1 with your own provider key) and OCR 0.05. HireLayer’s single credit already includes typed skills, a geocoded location and OCR, but it has no ISCO or O*NET codes.
Textkernel’s FAQ asks for batches under 100,000 documents to be parsed serially, its acceptable use policy says requests rejected with HTTP 429 still consume credits, and non-ad-hoc search and match calls are capped at 100 a day unless negotiated. HireLayer bills only successful calls.
Textkernel normalizes professions to ISCO and O*NET, publishes 29 resume languages and sells skills intelligence built on 4,500+ professions and 12,000+ skills. HireLayer’s taxonomy is narrower: typed skills, CEFR, EQF and French ROME codes.
Textkernel is “Textkernel by Bullhorn”. Its partner list includes Bullhorn, Salesforce, SAP SuccessFactors and Oracle. If you build an ATS, your parser supplier would be owned by an ATS vendor; whether that matters is your call.
Hosting, speed and integrations
HireLayer parses in France and keeps AI processing in the EU. Fast mode is in beta and connectors are built on request: the statuses below say exactly that, next to what the other vendor documents.
HireLayer·where a CV goesParsing
Paris data center, France
Files & photos
Stored in Paris, short-lived signed links
AI models
In-house AI pipeline, processed in the EU
JSON response
Back to your backend; do_not_store_data available
Textkernelas documented
HireLayer
Beta:Fast mode in beta, enabled per account on request; no published timing
We do not publish timings for the beta. Ask for access and measure it on your own CVs before relying on it.
Not available:No separate mode; the standard parser is stated at about 0.5 s median, the LLM Parser is “several seconds slower”
HireLayer has no marketplace apps yet. Connectors for Zapier, Make, Salesforce, SAP SuccessFactors, Workday, Tableau or your ATS are built by our team on request; in the meantime every API works from any HTTP step, with an optional webhook on each parse.
Textkernel statuses come from its own documentation and the tools’ marketplaces. “Via API” means no connector exists but the API can be called from that tool.
Ask for a connector| Tool | HireLayer | |
|---|---|---|
| ATS / HCM apps | On request: | Available:Bullhorn and Oracle Recruiting apps; built into 100+ partner products |
| Zapier | On request: | Not available:No Zapier app |
| Make | On request: | Not available:No Make app |
| Salesforce | On request: | Available:AppExchange listing |
| SAP SuccessFactors | On request: | Available:SAP Store listing |
| Workday | On request: | Not verified:Not on the official marketplace list |
| Tableau | On request: | Via API only:No connector; possible through the API |
HireLayer vs Textkernel
Textkernel cells describe Tx Platform v10 and the Professional plan unless noted. Credit costs come from the v10 transaction cost page. Highlighted rows show where one side has a documented edge.
Checked on 1 October 2026
HireLayer:Sign up and get an API key; Free plan with 50 credits / month
Textkernel:Free trial with 500 credits and instant access; Professional bought online[1]
HireLayer:€0.045–0.058 per credit on public plans and packs
Textkernel:“From $99 / month”, credits per tier not listed; $0.04 per credit on the first-purchase Accelerator[1]
HireLayer:1 credit per successful call, shared by all five APIs
Textkernel:1 credit per parse, plus add-ons: skills +0.1, job titles +0.2, geocoding +1[2]
HireLayer:Scans and photos cost the same 1 credit
Textkernel:OCR add-on, +0.05 credit[2]
HireLayer:Credits are deducted only after a successful response
Textkernel:Requests rejected with HTTP 429 consume the same credits[8]
HireLayer:Pack credits stay on the account
Textkernel:Expire monthly or yearly unless agreed in writing; not refunded[14]
HireLayer:No published benchmark with a reproducible method; test your own CVs in the demo
Textkernel:LLM Parser “reduces remaining errors by up to 30%”, without a method[6]
HireLayer:Contact, experience, education, languages with CEFR level, typed skills, certifications, availability, mobility, driving licences, ROME codes
Textkernel:Raw and normalized values, professions with ISCO and O*NET codes, skills with months of experience, management scores[4]
HireLayer:Fixed schema; no user-defined fields
Textkernel:FlexRequests parse user-defined fields, +1 credit[2]
HireLayer:face_url: the candidate photo as a temporary signed link, including from PDF CVs
Textkernel:Opt-in OutputCandidateImage returns the image as base64[4]
HireLayer:French ROME codes with a prediction score; no O*NET or ISCO
HireLayer:Hard, soft or software skill, with domain, subcategory and raw / normalized status
HireLayer:No anonymization or redaction endpoint
Textkernel:Resume HTML Anonymization add-on[1]
HireLayer:13 formats: PDF, DOC, DOCX, ODT, PPT, PPTX, ODP, XLS, RTF, TXT, JPG/JPEG, PNG, BMP
HireLayer:6 MiB encoded request · 100,000 extracted characters
Textkernel:About 16 MB; about 3 MB with the LLM Parser[3]
HireLayer:🇫🇷 Parsing runs in France, in a Paris data center
HireLayer:🇫🇷 Files and extracted photos stored in Paris; photos served through short-lived signed links
HireLayer:🇪🇺 HireLayer’s in-house AI pipeline, with processing kept in the EU
HireLayer:Parsed in France, processed by AI in the EU
Textkernel:DPA allows transfers to Bullhorn Inc. and sub-processors in the US[15]
HireLayer:Hosted only; no self-hosting or region choice
Textkernel:Possible “only in exceptional cases”[7]
HireLayer:ISO 27001 and SOC 2 in progress
Textkernel:ISO 27001 and SOC 2 stated on the company page[12]
HireLayer:On request; no public Zapier app
Textkernel:No Zapier app
HireLayer:On request; no public Make app
Textkernel:No Make app
HireLayer:On request; no AppExchange listing
Textkernel:AppExchange listing[21]
HireLayer:On request; no SAP Store listing
Textkernel:SAP Store listing[22]
HireLayer:On request; no Workday Marketplace listing
Textkernel:Not on the official marketplace list[18]
HireLayer:On request
Textkernel:No connector; possible through the API
HireLayer:Public reference with cURL, Python and Node.js examples, field formats and every error code
Textkernel:Developer portal and a Swagger UI per data center[23]
HireLayer:REST · multipart file upload · synchronous JSON response
Textkernel:Synchronous REST; file sent as DocumentAsBase64String in a JSON body[3]
HireLayer:One X-API-Key header for every API
HireLayer:Optional webhook_url on every parse; V2 callback-only contract still available
Textkernel:None documented[3]
HireLayer:One file per request; batch by running requests from your queue
Textkernel:No batch endpoint; batches under 100,000 documents must run one at a time[6]
HireLayer:No official SDKs; cURL, Python and Node.js examples in the docs
Textkernel:C# and Java[3]
HireLayer:Job text → weighted, CV-checkable criteria; labels are generated in French
HireLayer:Stateless: send job, candidate and criteria; get a status and explanation per criterion
HireLayer:Skills Matching API: up to 100 free-text skills per request against a catalog
Textkernel:Skills, Professions and Ontology APIs: 12,000+ skills and 4,500+ professions[10]
HireLayer:Rank 1–10 candidates against one job in a single call
Textkernel:Match queries on an index, 0.05 credit each[2]
HireLayer:No hosted candidate index or search engine
Textkernel:Indexes kept 18 months, then 1 credit per document every 6 months[8]
Cost per CV
Textkernel publishes its credit costs per transaction, but not how many credits each Professional tier includes. A cost per CV can therefore only be worked out with a quote, or from the first-purchase Accelerator offer.
CVs parsed per month
HireLayer
€89 / month
Scale plan · €0.045 per CV · ≈ $101 / month
Textkernel
Textkernel shows “Starting at $99 / month” for Professional but does not list the credits included in each tier, so the calculator cannot price a monthly volume.
Lowest published option that covers the volume, at list prices. Excludes enterprise quotes and first-purchase offers. HireLayer credits are shared by all APIs; this assumes they all go to parsing. Bars compare USD equivalents at the ECB reference rate of 2026-09-30 (€1 = $1.1355).
Reading the numbers
Compare by credits per document rather than plan price. A Tx parse with skills normalization and geocoding uses 2.1 credits (1 + 0.1 + 1); at the $0.04 Accelerator rate that is $0.084 per CV, for the first 5,000 credits only. A HireLayer parse with typed skills and a geocoded location is one credit, €0.045–0.058 on paid plans and packs. Ask Textkernel for the credit count of the tier you would buy before comparing at volume.
Textkernel published prices
Free trial$0 · 500 credits
Demo UI and API access, instant.
Accelerator$200 · 5,000 credits
Professional features, first purchase only: $0.04 per credit.
ProfessionalFrom $99 / month
Monthly plans from 500 to 25,000 credits; annual plans up to 100,000.
EnterpriseOn quote
Higher rate limits and skills / profession taxonomy download.
Billing rules
HireLayer published prices
1 successful API call = 1 credit across every API.
Enterprise volumes on quote. Prices from the HireLayer pricing page.
Analysis
For teams on Tx Platform v9 or v10, or still on a Sovren-era integration.
01
Textkernel announced the Sovren acquisition in November 2021 and unified the brand in 2023. sovren.com and its pricing page now redirect to Textkernel, and Sovren customers moved to the Tx Platform API. Tx v10 is recommended for new projects; v9 is still supported.
If you are re-evaluating at a contract renewal, the migration cost to Tx v10 and the migration cost to another vendor are similar in shape: new authentication, a new response schema and a field-mapping pass.
02
Tx Platform is synchronous, with no batch endpoint and no documented webhooks. Its FAQ asks for routine batches below 100,000 documents to be sent one at a time, and concurrency limits are set per account. Requests rejected with HTTP 429 still consume credits.
HireLayer is also one file per request, so you run your own queue. Credits are deducted only after a successful parse, and you can add a webhook_url to receive a callback alongside the synchronous response.
03
Textkernel’s output is deep. Positions carry normalized professions with ISCO and O*NET codes, skills carry months of experience and last-used dates, and FlexRequests extract fields you define yourself. That depth is what large job boards and analytics teams buy it for.
HireLayer normalizes fewer dimensions but returns them in every call: experience brackets, EQF education levels, CEFR language levels, contract types, geocoded locations and mobility, and ROME codes for French labour-market classification.
04
Search & Match V2, still marked Beta, works on documents you index (2 credits each) and keeps them 18 months before charging storage. Automated search and match calls are capped at 100 a day unless you negotiate more.
HireLayer Match is stateless: send the job text, the candidate text and the criteria, get a status and an explanation per criterion and a weighted score. Nothing is indexed, so there is no storage fee and no daily cap, but there is no search across a stored pool either.
05
Tx Platform has US, EU and AU data centers on AWS, and its terms place the EU endpoint within the European Union, without naming a country. The LLM Parser runs on AWS Bedrock in an undocumented region. Textkernel’s data processing agreement allows transfers to Bullhorn Inc. and to sub-processors in the United States.
HireLayer parses CVs in Paris, France and stores files and extracted photos there, serving photos through short-lived signed links. Its in-house AI pipeline processes CV content in the EU, so AI processing stays in Europe but is not pinned to France. The web app is hosted in Europe.
Which to choose
Both lists come from the comparison above. If most of your requirements sit in the right-hand column, stay where you are.
Stay with Textkernel when…Switching
The request is simpler, but the Tx schema is large and uses Raw / Normalized pairs, so plan a careful mapping pass.
Estimated effort
curl -X POST https://api.eu.textkernel.com/tx/v10/parser/resume \
-H "Content-Type: application/json" \
-H "Tx-AccountId: YOUR_ACCOUNT_ID" \
-H "Tx-ServiceKey: YOUR_SERVICE_KEY" \
-d '{
"DocumentAsBase64String": "<base64 of resume.docx>",
"DocumentLastModified": "2026-09-30"
}'curl -X POST https://onlineresumeparser.com/api/v3/parser \
-H "X-API-Key: YOUR_HIRELAYER_KEY" \
-F "[email protected]"Tx paths are relative to Value.ResumeData in the v10 API reference; HireLayer paths from the V3 API reference.
| Textkernel | HireLayer V3 |
|---|---|
| ContactInformation.CandidateName.GivenName / .FamilyName | info_candidate.first_name / .last_name |
| ContactInformation.EmailAddresses[0] | info_candidate.email |
| ContactInformation.Telephones[0].Normalized | info_candidate.phone_number |
| ContactInformation.Location.Municipality / .CountryCode | info_candidate.location.city / .country_code |
| EmploymentHistory.Positions[].Employer.Name.Normalized | work_experiences[].company_name |
| Positions[].JobTitle.Raw | work_experiences[].job_title |
| Positions[].StartDate.Date / .EndDate.Date | work_experiences[].start_date / .end_date |
| Positions[].IsCurrent | work_experiences[].currently_active |
| Positions[].NormalizedProfession | rome_jobs[]ROME, not ISCO / O*NET |
| Education.EducationDetails[].SchoolName.Normalized | educations[].school_name |
| EducationDetails[].Degree.Name.Normalized | educations[].degree_title |
| Skills.Normalized[].Name | skills[].skill_title |
| LanguageCompetencies[].Language | languages[].languageplus CEFR level |
STEP 01
Swap authentication
Replace Tx-AccountId and Tx-ServiceKey with one X-API-Key header, and stop base64-encoding: post the file as multipart form data.
STEP 02
Map Raw / Normalized pairs
Tx returns Raw and Normalized variants for many values. HireLayer returns one value per field, plus a status on skills. Decide which Tx variant your database stored and map it once.
STEP 03
Rethink Search & Match
If you used indexes, keep your own search for the candidate pool and call HireLayer Match or Rank on the shortlist it returns.
STEP 04
Compare on your own files
Run a permissioned sample through both APIs and compare the fields you store, especially if you relied on ISCO or O*NET codes.
HireLayer Match · Candidate & Job Matching API
Where Search & Match V2 needs both documents indexed first, HireLayer Match takes the texts and criteria in the request. Each criterion comes back with a status and an explanation; summaries and explanations are generated in French in the current version.
curl -X POST https://onlineresumeparser.com/api/v1/matching/job-candidate \
-H "Content-Type: application/json" \
-H "X-API-Key: YOUR_API_KEY" \
-d '{
"job_text": "Backend engineer. Java and Spring required. Kafka experience expected. German is a plus.",
"candidate_text": "Six years of Java and Spring Boot on payment APIs. Built event pipelines with RabbitMQ.",
"matching_criteria": [
{ "id": "crit_1", "label": "Java and Spring", "weight": 3, "is_mandatory": true, "rationale": "Java and Spring are required." },
{ "id": "crit_2", "label": "Kafka", "weight": 2, "is_mandatory": false, "rationale": "Kafka experience is expected." },
{ "id": "crit_3", "label": "German", "weight": 1, "is_mandatory": false, "rationale": "German is listed as a plus." }
]
}'{
"score": 0.7833333333333333,
"summary": "Le candidat maîtrise le socle Java et Spring ; l'expérience Kafka reste à confirmer en entretien.",
"evaluated_criteria": [
{
"id": "crit_1",
"label": "Java and Spring",
"weight": 3,
"match_status": "ideal",
"match_explanation": "Le CV mentionne six ans de Java et Spring Boot sur des API de paiement."
},
{
"id": "crit_2",
"label": "Kafka",
"weight": 2,
"match_status": "potential",
"match_explanation": "Le CV cite des pipelines événementiels avec RabbitMQ, mais pas Kafka."
},
{
"id": "crit_3",
"label": "German",
"weight": 1,
"match_status": "not_mentioned",
"match_explanation": "Le CV ne mentionne aucune langue étrangère."
}
]
}The only comparison that settles quality is yours. Take 20 to 50 permissioned CVs that look like your real intake, scans included, parse them with both APIs and compare the fields you store.
mkdir -p hirelayer
for f in sample/*; do
curl -s -X POST https://onlineresumeparser.com/api/v3/parser \
-H "X-API-Key: $HIRELAYER_API_KEY" \
-F "file=@$f" \
-o "hirelayer/$(basename "$f").json"
doneFAQ
HireLayer parses CVs in Paris, France, stores files and photos there and processes CV content with its in-house AI pipeline in the EU. Textkernel offers US, EU and AU data centers on AWS without naming the EU country, and its data processing agreement allows transfers to Bullhorn Inc. and US sub-processors.
Textkernel announced the acquisition of Sovren on 30 November 2021 and unified the brand in 2023. sovren.com now redirects to Textkernel, and Sovren customers use the Textkernel Tx Platform API. Textkernel itself became part of Bullhorn in 2024.
Textkernel offers a free trial with 500 credits, a first-purchase Accelerator plan at $200 for 5,000 credits and a Professional plan “starting at $99 / month”, with Enterprise on quote. A parse costs 1 credit plus add-ons such as skills normalization (+0.1), job title normalization (+0.2) and geocoding (+1).
No. HireLayer returns French ROME occupation codes with a prediction score. If your product depends on ISCO or O*NET, Textkernel covers that and HireLayer does not.
Not in a single call; like Tx Platform, HireLayer parses one file per request. You run the files through your own queue, and credits are only used for successful parses.
We cannot say: Textkernel publishes speed claims without a method, and HireLayer has no published benchmark. Run your own permissioned CVs through both APIs and score the fields you store.
Checked on 1 October 2026
Textkernel facts come from its own public pages and documentation, opened on the date above. Third-party sources are labelled. Accuracy and speed figures are reported as each vendor states them; none of them, including ours, is an independent benchmark.
Prices are list prices in each vendor’s currency. USD equivalents use the ECB reference rate of 2026-09-30.
Offers change. If a fact here is out of date, tell us through the contacts in our legal notice and we will re-check it.
Textkernel and the other product names and logos on this page are trademarks of their respective owners, shown only to identify the products compared.
Keep comparing
