Oxolo logo
Audio Transcription · Voice Assistants

Oxolo

Oxolo is a voice-driven documentation assistant for construction teams. Workers dictate on site and the AI produces transcripts, extracts tasks, fills the company's own report templates, then sends everything out by WhatsApp or email.

Active GDPR compliant Free trial Subscription No public API Verified by Guidaio
Overview

What is Oxolo?

Oxolo is a voice-driven documentation assistant built specifically for the construction trades. Instead of typing up site notes in the evening, a worker speaks naturally while still on site, and the AI turns that speech into a finished, structured document. The chain runs from voice to dispatched report with no re-keying step: the recording is transcribed, decisions are detected, tasks are extracted with a priority, a deadline and an assignee, and change orders, delays and other notable mentions are flagged.

What separates it from a general transcription tool is that it fills the company's own paperwork. Firms upload their blank construction logs, site meeting forms or acceptance reports as Word files or PDF forms, and the AI populates them. Built-in templates — Daily Site Report, Site Inspection, Progress Report — cover teams that have none of their own, and finished documents export to PDF or Word under the company's branding.

Multilingual crews are a core design assumption. The FAQ claims support for more than thirty languages with automatic detection of the spoken language, and the product literature names German, English, Polish, Turkish and Romanian as typical site languages. Content stays available in the original and in the target language, and search works across languages, so a statement or photo logged in one language can be found from another. When a task is shared, the sender picks the recipient's language and sends it by WhatsApp, email or link.

The tool is built for field conditions. Recording works offline in deep basements and syncs once the connection returns, noise suppression and speaker recognition cope with site noise, and a multilingual construction dictionary handles trade jargon. Photos and videos are captured with automatic location, timestamps and EXIF or GPS metadata, which is what makes them usable as evidence. A secure project chat answers questions from the documented record.

The product splits into two tiers: a desktop hub where administrators upload templates, set roles and organise project folders, and a streamlined mobile app for capture in the field, synchronised in real time. iOS and Android apps are published alongside the web application.

Oxolo GmbH sells strictly to businesses — consumers are excluded by the terms. Recordings are capped at sixty minutes, and the vendor states plainly that AI output may be incomplete or wrong and has to be reviewed.

What it does

  • Transcribe spoken site notes into clean, structured text, with automatic detection of the language spoken
  • Generate a finished construction report with summary, decisions and action items from those recordings
  • Fill the company's own blank templates — Word documents and PDF forms — automatically and without gaps
  • Extract tasks with a priority, a deadline and a named person responsible
  • Translate reports and tasks and dispatch them to the crew by WhatsApp, email or a share link
  • Capture photos and videos as evidence, with automatic location, timestamps and EXIF or GPS metadata
  • Answer questions about a project from the documented record through a secure project chat
Audience

When to use Oxolo / When not to

A quick filter to help you decide if Oxolo is the right fit.

When to use Oxolo

  • Tradespeople who document work with their hands full — carpenters, drywallers, painters, electricians, roofers, tilers, bricklayers, metal workers, kitchen installers, HVAC and plumbing crews
  • Site supervisors, project engineers and construction project managers who need daily reports, inspections and progress logs written the same day
  • Architects and planning offices that record site conditions, verify design compliance and log change orders during walkthroughs
  • Firms running multilingual crews, where information is spoken in German, English, Polish, Turkish or Romanian and misunderstandings cost rework
  • Contractors of five to twenty seats who want to keep their existing construction logs, meeting forms and acceptance reports rather than adopt a vendor's format

When not to use Oxolo

  • Private individuals and consumers — the terms restrict the service to businesses, freelancers and public-law entities, and exclude purely personal or household use
  • Teams looking for a free tool: there is no permanent free plan, and the terms state there is no entitlement to a free trial
  • Developers who need to embed the service in another system — no public API and no developer documentation are published
  • Very small outfits or single operators, since the entry plan is priced for up to five licences rather than per individual user
  • Anyone who needs machine output they can sign off unread: the vendor disclaims accuracy, completeness and fitness for purpose of its AI results
Get started

How to use Oxolo

A typical end-to-end flow, from setup to results.

  1. Have an administrator open the organisation account at app.oxolo.com and invite the crew, or register the business directly
  2. Sign in with an email address, a password and a phone number, confirming with the one-time code sent by SMS
  3. From the desktop hub, upload your blank templates — construction logs, timesheets, defect reports, acceptance forms — as Word files or PDF forms
  4. Set roles and permissions, create the project folders and add both internal members and external subcontractors
  5. On site, open the mobile app and start a voice recording, speaking naturally; it works offline and the location is captured automatically
  6. Add photos or videos as you go, so the evidence carries timestamps and GPS data
  7. Let the AI transcribe the recording, detect decisions, create the tasks and flag change orders and delays
  8. Choose the template to fill and pick which recordings feed it, then let the system populate the document
  9. Review the result, edit it if needed, and export it as a branded PDF or Word file
  10. Share tasks and reports with the crew, selecting each recipient's language and sending by WhatsApp, email or link
Quick read

Pros & Cons

Pros

  • Genuinely vertical rather than a generic transcriber: trade jargon, fourteen named crafts and native site templates
  • The tool adapts to your paperwork instead of the reverse — you upload your own blank forms and keep your document formats
  • An unusually transparent compliance file for a company this size: a published Article 28 DPA, thirteen named subprocessors with countries and transfer mechanisms
  • An explicit commitment not to train on customer content, covering both its own models and the third-party LLMs it deploys, whatever the product-improvement setting
  • Core infrastructure inside the EU — Ireland for storage and databases, France for speaker diarisation, Germany for error monitoring
  • Designed around real site conditions: offline basements, background noise, gloved hands, mixed-language crews
  • Published, readable pricing with cancellation at any time, and native iOS and Android apps rather than a web page in disguise

Cons

  • No permanent free plan, and the terms are explicit that a free trial is granted only within a sales process and is never an entitlement
  • Businesses only: consumers and private individuals are excluded by the terms, as is purely personal use
  • No public API and no developer documentation, so there is no documented path into an existing IT system
  • Licences are billed the moment an administrator creates them, not when the user activates them — an account opened by mistake is still charged
  • Automatic speaker recognition is on by default as an opt-out, generating biometric voiceprints, and the duty to inform and obtain consent from recorded third parties falls on the customer
  • Batch speaker diarisation is processed in the United States even though the rest of the infrastructure sits in the EU
  • No published support email address — support runs through a form — and the public review base was very thin at the time of review
Pricing

Pricing & Plans

There is no permanent free plan. A free trial exists but is not guaranteed: the terms state it is granted only within a sales process, an order form or a clearly designated trial, with no entitlement to one. The cheapest paid entry point is EUR 49.00 per month for the Team S plan, billed annually at EUR 588.00 per year, covering up to five licences. Annual commitment is presented as saving up to 29 percent against monthly billing. Billing is per licence, or seat, and each licence is charged from the moment an administrator creates it rather than when the user activates it.

Plan 1
  • Team S — up to 5 licences — EUR 49.00 per month
  • billed annually at EUR 588.00 per year
  • a stated saving of EUR 240.00 per year
Plan 3
  • Team L — up to 20 licences — EUR 199.00 per month
  • billed annually at EUR 2
  • 388.00 per year
  • a stated saving of EUR 600.00 per year
Plan 4
  • Need more? — beyond 20 licences — unlimited users with custom terms and pricing on request
Plan 5
  • All three named plans include unlimited usage
  • the full feature set
  • a legally secure archive
  • and the option to switch or cancel at any time
Special offers — Annual billing instead of monthly, presented as saving up to 29 percent · Stated annual savings per plan: EUR 240.00 on Team S, EUR 360.00 on Team M and EUR 600.00 on Team L · Custom package with unlimited users and negotiated terms above twenty licences · Free fifteen-minute demonstration run on your own company templates · No student, jobseeker, non-profit or launch discount is published
Prices and plans listed above may evolve. Always check the official pricing page before subscribing.
Trust & Privacy

Data, GDPR & hosting

A consolidated view of how Oxolo handles your data.

GDPR overview

Implementation is concrete and documented rather than asserted. Oxolo GmbH names itself controller for its own processing at its Hamburg address, publishes a data protection officer reachable at gdpr@oxolo.com, and names the competent supervisory authority, the Hamburg Commissioner for Data Protection and Freedom of Information. An Article 28 data processing agreement is published as a full page, not offered on request, and it names every subprocessor with its country and transfer mechanism. The eight data subject rights are listed explicitly, with requests handled manually since no self-service export exists. Transfers outside the EEA rely on Standard Contractual Clauses and, for certified US recipients, the EU-US Data Privacy Framework. Voiceprints are treated as Article 9 biometric data. The Trust Center shows GDPR and UK GDPR as compliant.

Who owns the data?

The customer keeps ownership. Under the terms, the customer retains its rights in the input it records or uploads, and any rights arising in the output belong to the customer in its relationship with Oxolo. Uploaded report templates likewise remain the customer's, and the reports Oxolo generates from that input are treated as customer content. Oxolo receives only a limited, non-exclusive licence to host, store, reproduce, transmit, convert, evaluate, display and process that material in order to run, secure, maintain, troubleshoot, bill and support the service. Oxolo retains rights in its own software, models, workflows, interfaces and trademarks. Under the GDPR, the customer is the controller for that content and Oxolo acts as its processor.

Reuse rights

Customers may use their input and the generated reports freely for their own business purposes without asking Oxolo's permission — the output belongs to them and the reports are treated as their content. Practically, that means exporting documents as PDF or Word, editing them after export, sharing them with subcontractors and clients, and keeping them as project records. Two limits apply. The customer warrants it holds the rights to whatever it uploads and that the material infringes no third party, which matters when recordings capture employees, subcontractors or visitors. And the licence granted to Oxolo is one-way and narrow: Oxolo may process the material only to deliver, secure, maintain, bill and support the service, plus product improvement if — and only if — the customer has expressly switched that setting on.

Data retention & training

Retention summary
Data is kept only as long as the purpose requires, with the exact period depending on the contract, your own settings and instructions, statutory duties, security needs and backup cycles. Your content is processed for the contract term and on your instructions, and administrators can delete or export it themselves at any point. After the contract ends, an export option stays open for at least thirty days, and the data processing agreement commits to deletion or return within ninety days, unless a legal retention duty, an evidentiary interest or a backup cycle stands in the way. Voiceprints can be deleted on request or in-app, and go automatically when an account closes or a user leaves an organisation; deleting a voiceprint also removes the source audio. Billing, contract and tax records are kept longer, as the law requires.
Trains on customer data
No
Training opt-out available
Yes
Subprocessors disclosed
Yes
DPA available
Yes
GDPR contact

Hosting summary

Core infrastructure runs inside the European Union. Application data, audio recordings and reports are stored on AWS in the eu-west-1 region in Ireland, with the database and authentication on Supabase in the same region and the transcript-embedding database on AWS RDS there too. Speaker diarisation and voiceprint identification run with pyannoteAI in France, eu-west-3. LLM inference is handled by Anthropic Ireland, with OpenAI Ireland where that function is enabled. Error monitoring runs on Sentry's German instance, product analytics on PostHog's EU endpoint, and payments through Stripe Payments Europe in Ireland. Two flows leave the EU: transactional email via Postmark in the United States, and the batch speaker-diarisation endpoint at AssemblyAI in the United States, while AssemblyAI streaming stays on the EU endpoint. Those transfers rely on EU Standard Contractual Clauses and, where the recipient is certified, the EU-US Data Privacy Framework. Traffic is encrypted in transit over HTTPS and TLS, and authenticated endpoints use token-based access.

Hosting countries
🇮🇪 Ireland🇫🇷 France🇩🇩 Germany🇺🇸 United States
Hosting regions
EU
Watch-outs

Things to keep in mind

Risks and trade-offs to weigh before adopting Oxolo.

  • Automatic speaker recognition is switched on by default and creates biometric voiceprints — special-category data under Article 9 of the GDPR. Decide deliberately whether to leave it on before rolling the tool out to a crew
  • The duty to inform recorded people and, where needed, obtain their consent sits with you, not with Oxolo, and has to be handled outside the product. That covers employees, subcontractors, clients and visitors caught on a recording
  • Recording speech that is not publicly spoken is legally restricted in a number of countries, a point the terms raise explicitly. Check your local rules before making dictation routine
  • The vendor disclaims accuracy, completeness and fitness for purpose of AI output. Treating an unread generated report as a contractual document — for billing, acceptance or a claim — shifts that risk onto you
  • Convenience erodes vigilance: when reports write themselves, the habit of actually checking the site against the paperwork weakens, and errors get archived as evidence rather than caught
  • Licences are billed from the moment an administrator creates them, so loose seat management costs money quietly
  • Batch speaker diarisation is processed in the United States while the rest of the stack is in the EU — worth confirming against your own data policy rather than relying on the Servers in Europe claim alone
Setup

Setup & Integrations

Technical difficulty

Low. Nothing to install beyond the iOS or Android app, with everything else running in the browser. An administrator opens the organisation, invites users, and signs in with email, password and an SMS one-time code. The real setup work is organisational rather than technical: uploading your blank templates, defining roles and permissions, arranging project folders, and deciding whether to leave automatic speaker recognition enabled. No development skills are needed and there is no API to integrate. Field users get a deliberately stripped-down mobile app, so crew onboarding is close to immediate.

Deployment

Web appMobile appIOS appAndroid app

Apps stores

Integrations

WhatsApp Microsoft Word

Supported languages

EnglishGermanPolishTurkishRomanian
Company

Behind Oxolo

Company name
Oxolo GmbH
Founded
28/05/2020
Country of origin
🇩🇩 Germany
Headquarters
Bohnenstr. 2, 1 OG, 20457 Hamburg, Germany
UBO
INFORMATION_NOT_FOUND
UBO country
INFORMATION_NOT_FOUND
Domain registrar country
INFORMATION_NOT_FOUND
Legal contact

Fundraising

Series A of EUR 13 million announced on 24 October 2023, led by DN Capital
Individual investors named in that round: Christian Vollmann, Johannes Plehn, Thomas Hagemann, Stefan Wiskemann, Alex Taubert and Mischa Ruerup
Stated use of proceeds: research and development, hiring, and accelerating commercialisation
Note for readers: that round financed the company's earlier AI video generation platform for e-commerce, not the construction documentation product described here — the figures come from press sources rather than from the site itself

Social

Official links

Resources

All the official URLs gathered for verification and reference.

FAQ

Frequently asked questions

Can I cancel at any time?
Yes. Your subscription stays active until the end of the billing period you have already paid for, and the published plans all include the option to switch or cancel whenever you want.
Can I edit the reports the AI produces?
Yes. You can define the structure yourself using protocols, or upload your own report template for the AI to fill. Once the document is exported as PDF or Word, you can edit it freely.
How does billing actually work?
Billing is per licence, or seat. An administrator creates each licence and it is charged from the moment it is created, not when the user first signs in — so a seat opened and left unused is still counted.
How long can a single recording be, and how fast is transcription?
Recordings are capped at sixty minutes each. Transcription takes anywhere from a few seconds to a couple of minutes, depending on how long the recording is.
Which languages does it handle?
The vendor claims more than thirty languages, with the system detecting the spoken language and processing it automatically. German, English, Polish, Turkish and Romanian are named as typical site languages.
Does it work without a mobile signal?
Yes. Recording keeps working offline, including in deep basements, and the data synchronises automatically once the device reconnects.
Is my site data used to train AI models?
No. The terms state that input and output are not used to train Oxolo's own models, nor the third-party large language models it deploys, regardless of any other setting. Using content for product improvement requires the customer to switch that option on deliberately, and it can be switched off again at any time.
Where is the data hosted?
Core infrastructure runs in the EU — AWS and Supabase in Ireland, speaker diarisation in France, error monitoring in Germany. Two flows leave the EU: transactional email and batch speaker diarisation are processed in the United States under Standard Contractual Clauses and, where certified, the EU-US Data Privacy Framework.
Can I use it as a private individual?
No. The terms restrict the service to businesses, freelancers and public-law entities, and the data processing agreement excludes use by individuals for purely personal or household purposes.
Is there a free plan?
No permanent free plan is offered. A free trial may be granted within a sales process or an order form, but the terms are explicit that there is no entitlement to one, and a trial never converts into a paid subscription by itself.
Conclusion

Should you pick Oxolo?

Oxolo is a narrow tool that knows exactly what it is for, and that focus is its strongest argument. Its value does not come from transcription, which is now a commodity, but from what happens afterwards: the company's own construction log, defect report or acceptance form is filled in automatically, in the vocabulary of the trade, and pushed to the crew in a language each recipient reads. For a contractor whose evenings currently go into typing up the day, that is a concrete saving rather than a vague productivity claim.

The compliance file is stronger than the size of the company would suggest. A published Article 28 data processing agreement, thirteen named subprocessors with their countries, EU infrastructure and a written commitment not to train on customer content are not routine at this level of the market, and they matter when recordings capture named employees and subcontractors.

The trade-offs are real. It is business-only, there is no permanent free plan, the entry price assumes five seats, and there is no API for wiring it into an existing system. Licences are billed from creation rather than activation, which punishes sloppy administration. The point deserving most attention before a wide rollout is biometric: speaker recognition is enabled by default and generates voiceprints, while the duty to inform recorded third parties and collect consent sits with the customer, outside the product.

Verdict: a credible, well-documented fit for construction and trade firms of roughly five to twenty users that want to keep their own paperwork. Read the speaker-recognition settings before deploying, and expect to check the AI's output — the vendor itself disclaims any guarantee of accuracy. The public review base was still thin at the time of review, so ask for the fifteen-minute demo on your own templates.