Zero Trackers, No Cookie Banner, 27 Languages: Gewerkton’s Architecture Is the Privacy Policy

The quickest way to read a software company’s real attitude toward privacy is not its policy page. It is its marketing site. Open the average vendor homepage — security vendors included — and your browser quietly announces itself to analytics networks, tag managers, font services and ad pixels before you have read a single sentence. Gewerkton, a voice-first construction documentation and defect management platform built for global markets, does the opposite. Its marketing site ships in 27 languages with zero trackers, no cookie banner and a fully egress-free architecture. Whatever you do on that page stays between you and the page.

That is not a compliance trick. It is an engineering decision with a pleasant legal side effect, and it is the same decision that runs through the product itself — a product whose entire job is evidence. Gewerkton’s own line is blunt: “On site, what counts is what’s proven.” A platform that exists to prove things has to be able to show where the data lives, too. The website is simply the first, publicly visible layer of that stance.

FIFINE K438 4K Resolution Webcam with Built in Microphone for Streaming

[4K Resolution] Step into the spotlight with the FIFINE K438. Its 4K resolution and 13MP sensor bring your…

AmazonView Latest Price

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

A marketing site with nothing to confess

Cookie banners are usually treated as a legal necessity. In practice they are an architectural confession: a banner exists because the page leaks — to analytics providers, A/B-testing tools, retargeting pixels, embedded video hosts. Gewerkton’s site needs no banner because it has nothing to ask forgiveness for. Zero trackers means zero third parties receiving news of your visit, and a fully egress-free architecture means the site makes no outbound calls to anyone else’s infrastructure while you browse. There is no consent to manage because there is nothing to consent to.

Worth pausing on what egress-free actually demands, because it is a discipline, not a plugin. It means no external analytics beacons, no third-party embeds pulling their own scripts onto the page, no quiet calls home from components the marketing team did not even know were chatty. Every asset and every behaviour has to be owned. Most companies decide that trade-off is not worth it for a brochure site. Gewerkton decided it was — which tells you something about how it will treat decisions that actually matter.

The 27 content languages deserve a second look as well. Multilingual sites are commonly assembled by piping pages through third-party translation or localization services — yet another place where visitor data leaves the building. Gewerkton publishes its 27 languages directly. Even the media is in-house: the site draws on a media bank of more than 51 self-produced clips and posters rather than embedding external platforms that would drag their own trackers along. Global reach, it turns out, does not require a global data leak.

None of this is fear-mongering about analytics; plenty of companies run trackers and sleep fine. The point is different. When a vendor selling documentation and defect management for construction evidence builds even its storefront this way, it is showing you its instincts. Architecture is where a company’s real priorities are visible, and Gewerkton’s priority order is legible from the network tab alone.

Evidence is only as good as its custody

Why does this matter so much for this particular product category? Because of what the product holds. Gewerkton is built for construction sites, and what flows through it is evidence: defects documented with photo and deadline, daywork reports dictated on the move, signatures captured on device at handover, site instructions backed by the original audio recording. This is the material that later decides disagreements about who built what, when, and to which standard.

Evidence has a property ordinary business data does not: its value depends on the chain of custody being boring. If you cannot say plainly where a recording is stored, under whose jurisdiction, and who could have touched it, the evidence is weaker — not because anyone tampered with it, but because doubt is the enemy of proof. A documentation and defect management platform is, in effect, a custody business. Data custody cannot be a settings-page afterthought; it has to be a first-class feature, designed the way Gewerkton designed its own website: with nowhere for data to quietly wander off to.

Think about the projects where this bites hardest. Infrastructure and tunnel jobs run for years and accumulate change orders; when an instruction given two winters ago is suddenly contested, what helps is that the instruction is backed by the original audio — captured once, kept intact, unambiguous about what was actually said. That only works if the platform never had a reason, or a mechanism, to let that original drift through third parties on its way to storage.

The cross-border case is the real stress test. Gewerkton is built for projects where EU, US and APAC teams work on the same job, each in their own language. In that setup the evidence original must stay unambiguous — one authoritative record, not a folder of diverging copies scattered across continents and vendors. Multilingual capture is the visible feature; custody discipline underneath is what makes the feature trustworthy.

Residency by choice, not by default

Gewerkton’s data stance can be stated in one sentence: EU cloud or your own infrastructure — your choice. Residency is not a region locked in at signup; it is a decision the customer makes. A European contractor with strict procurement rules can keep data in the EU cloud under a familiar regime. An operator that wants everything inside its own perimeter can run on its own infrastructure. The platform treats both as first-class deployment paths rather than grudging exceptions.

That posture is more respectful of reality than the usual one-size-fits-all hosting page. Different projects answer to different owners, different auditors and different national rules, and those answers can change from one job to the next. A platform that hard-codes a single residency answer forces customers to route around it. Gewerkton instead exposes the decision — which is exactly what “data custody as a feature” looks like in practice.

The Asia angle makes the design intent clearest. Gewerkton explicitly targets projects in Asia — Chinese, Korean and Vietnamese crews, multilingual from capture to report — with data residency by choice as part of the offer. A platform that dictated storage geography to its customers would have to walk away from some of those projects. A platform that treats residency as a customer decision does not have to. Once again: the commercial reach follows from the custody model, not the other way round.

Bring your own AI — region included

Voice-first software lives or dies on its speech pipeline, and speech pipelines live or die on where the audio is processed. Gewerkton’s answer is BYO-AI: 13 AI providers, bring your own keys, region selectable — EU, US, and Asia including mainland China — with no vendor lock-in. The customer picks the provider, supplies the credentials, and chooses the jurisdiction in which transcription and analysis happen.

This is the residency philosophy extended to the model layer. An EU project can keep its voice processing with EU providers. A project with crews from mainland China can select Asian providers — including mainland China — without the platform treating that choice as exotic. And because the keys are the customer’s own, switching providers is a configuration change rather than a migration. Lock-in is where custody quietly dies; Gewerkton engineered that failure mode out at the design stage.

The operational meaning of “your own keys” is worth spelling out: the relationship with the AI provider belongs to the customer, not to the platform. Gewerkton sits in the middle as tooling, not as a broker of your audio. Combined with region selection across EU, US and Asian providers, that gives project teams a concrete, auditable answer to the question every evidence workflow eventually faces: where did this data go, and under whose arrangement?

There is a second, subtler benefit. When the platform itself is not the AI vendor, its incentives stay clean. Gewerkton does not need your audio for anything; it needs your workflow to work. That is the same posture as the tracker-free website, one layer deeper: the fewer reasons a platform has to want your data, the easier it is to trust it with your evidence.

Three product lines, one custody model

The product family is a branded house with three lines, and the data stance runs through all of them. Gewerkton Field is the voice-first construction site app: dictation to evidence, defects, daywork reports, takt planning and the site portal. This is where capture happens — in the field, in the crew’s own language, often offline in the dead zones that wind farms and infrastructure projects specialise in. Offline capture matters to the custody story too: evidence recorded without connectivity has to stay intact on the device until it can be synced, with no improvising in between.

Gewerkton Studio is the browser workspace for plans and models. Notably, it assumes no perfect inputs: where no model exists, the site team creates one in the browser. That is a pragmatic admission about real projects — most existing buildings and much of the world’s infrastructure come with no tidy model attached — and it keeps the workflow inside the same custody perimeter instead of shunting teams out to external modelling tools with their own data appetites.

Then there is Gewerkton Cloud, the operations and model/data coordination layer between Field, Studio and third parties. Cloud is where the custody story becomes operational: which models and records move, to whom, in which direction, under which residency choice. Coordination is the moment data is most likely to leak — exports, shares, integrations — and it is exactly the layer Gewerkton has made its own product line rather than leaving it to email attachments and ad-hoc file drops.

Where this actually runs

The deployment fields read like a list of worst cases for data discipline, which is presumably the point:

• Wind farms and renewables: distributed sites, rotating crews, field acceptance, offline capture in dead zones where connectivity is a rumour.
• Data centers and industrial plants: many trades in parallel under tight deadlines, where meeting decisions become trade-sorted task lists.
• Housing and building construction: defects with photo and deadline, dictated daywork reports, signature on device at handover.
• Infrastructure and tunnels: long durations and many change orders, with site instructions backed by the original audio.
• Cross-border teams: EU, US and APAC on the same project, each in their own language, the evidence original unambiguous.
• Projects in Asia: Chinese, Korean and Vietnamese crews, multilingual from capture to report, with data residency by choice.

Across all of these, the pattern is the same: the data that matters is created at the edge — spoken, photographed, signed — and its value depends on everything after capture being controlled. Egress-free is not just a website property in this product; it is the product’s thesis.

Built by one person and a fleet of agents

The engineering backstory fits the architecture. Gewerkton is built by a solo founder directing a fleet of coding agents — Codex and Claude. In one night, the fleet shipped 21 software packages, verified with negative controls and mutation tests. Whatever one thinks of agent-driven development, that verification detail matters. Mutation testing is what you reach for when you distrust your own test suite — it deliberately breaks the code to check that the tests notice. Negative controls are what you run when you want expected failures to prove the checks are real. A team that verifies this way is a team that thinks about evidence. The website, the residency model and the testing discipline are the same instinct at three different altitudes.

The market positioning has the same dual character. Gewerkton was born in the German market and carries the deepest German commercial integration — GAEB, REB, XRechnung and DATEV, the unglamorous acronyms that decide whether a tool survives German procurement. But it was built for global markets from the start: 27 content languages, regional AI-provider choice across the EU, US and Asia, and deployment targets from European housing to Asian infrastructure. Depth at home, reach abroad, custody in the customer’s hands throughout.

Status: beta, openly

One thing should be stated plainly, because Gewerkton states it plainly: the product is in beta right now, with a public beta planned for fall 2026. The architecture is opinionated, the three product lines are defined, and the verification discipline is real — but this is a young platform inviting scrutiny, not an incumbent demanding trust. For an evidence tool, that honesty about its own status is arguably part of the custody story too.

The tracker-free website, then, is not a marketing gimmick. It is the same architecture decision made visible: what you capture is yours, where it lives is your choice, and the platform’s job is to keep it provable. If that sounds like the way construction evidence should be handled, Gewerkton Cloud — the coordination layer where those custody decisions actually run — is the place to look next.