Vorher

1. One-liner

Decides before the van leaves whether a pre-1993 job needs asbestos sampling, an Anzeige, or nothing at all.

2. Trend signal — why now?

On 20 December 2025 the amended Gefahrstoffverordnung came into force and rewrote who carries asbestos risk in Germany. Two changes matter.

First, a statutory asbestos presumption. Every building whose construction began before 31 October 1993 is now presumed to contain asbestos unless credible evidence says otherwise. The presumption is not paranoia: asbestos was used in roughly three-quarters of all buildings built or renovated before October 1993 (Containerhelfer, 2026).

Second, and this is the fight, the burden landed on the contractor rather than the client. Under §5a the Veranlasser owes only an Informationspflicht — hand over the building year and whatever they happen to know about construction history. No technical investigation duty was created for them. As the legal analysis puts it: “Eine verbindliche Asbesterkundungspflicht für den Veranlasser folgt hieraus nicht”, while “Die Hauptverantwortung liegt auch im Verhältnis zum Veranlasser… bei dem Bauunternehmen” (FPS Law).

The trades associations lost that argument and are still furious, on the record, by name:

• Felix Pakleppa, Hauptgeschäftsführer ZDB: “Statt Bauherren in die Verantwortung zu nehmen für ihre Sanierungsprojekte, sollen unsere Betriebe und Beschäftigten nun allein sicherstellen, dass sie sich nicht einem erhöhten Gesundheitsrisiko aussetzen. Das ist realitätsfern und ein absolutes No-Go” (bau.bi)
• Helmut Bramann, Hauptgeschäftsführer ZVSHK: “Die jetzige Regelung ist insofern so praxisfremd wie unverantwortlich” (same)
• Mathias Bucksteeg, Hauptgeschäftsführer Bundesverband Farbe: “Wir sind wütend”, calling it “bürokratischer Irrsinn” (same)
• Alexander Neuhäuser, Hauptgeschäftsführer ZVEH: “Die Politik wälzt die Verantwortung vom Veranlasser einer Baumaßnahme auf den Auftragnehmer ab” (elektro.net)
• Ralph Mahler, owner, Beton Bearbeitung Mahler GmbH — an operating firm, not a lobbyist: “Die neue Gefahrstoffverordnung ist schön und recht, aber unterm Strich hat man die Verantwortung auf die Handwerksbetriebe abgewälzt” (DHZ, 2025-09-08)

The ZDH’s formal objection is that the rule “die Bringschuld des Veranlassers in eine Holschuld des (Bau-)Unternehmers umwandeln will” — converting the client’s duty to provide into the contractor’s duty to go and fetch — and it calls the result impractical and unimplementable (ZDH). The Bundesverband Farbe warns the reading of “Abbrucharbeiten” could “normale Sanierungsaufträge lahmlegen”.

Then there is the clock. A new company-level Genehmigungspflicht now covers demolition work in the low and medium risk bands, not just high risk — and it must be demonstrated from 19 December 2026: “Die Genehmigungspflicht für Abbrucharbeiten im niedrigen und mittleren Risikobereich muss erst ab dem 19. Dezember 2026 nachgewiesen werden” (BVN, 2026-01-19). Approval runs six years, and a four-week Genehmigungsfiktion applies if the authority stays silent. The notification must now name the individual employees and attach proof of their Fachkunde and occupational health screening.

And here is the gap the regulation left wide open: the employer must decide, alone and per job, whether the work is approval-required Abbruch or exempt Instandhaltung.

Nobody agrees how many firms that captures. The BMAS impact assessment put the number affected by the Genehmigungspflicht at 1,165 Betriebe. The Bundesverband Farbe says that if “Abbrucharbeiten” is read to include partial removal during routine maintenance — stripping wallpaper off a wall, lifting a floor covering — the real figure is 220,000 (handwerksblatt, 2025-08-20).

A 190-fold disagreement between the regulator and the trade about who is even in scope, three months from the deadline, with the TRGS 519 rewrite that would settle it still unpublished. Every firm in that gap has to decide for itself, this week, on jobs it is quoting now.

Provenance:

• Signal 1 (demand): Regulator and trade disagree 190-fold on who is even in scope — BMAS impact assessment says 1,165 Betriebe affected by the Genehmigungspflicht, Bundesverband Farbe says up to 220,000 if “Abbrucharbeiten” covers routine partial removal; five named association chiefs on the record calling the rule “realitätsfern”, “praxisfremd wie unverantwortlich”, “bürokratischer Irrsinn” — https://www.handwerksblatt.de/handwerkspolitik/gefahrstoffverordnung-malerhandwerk-kritisiert-gesetzentwurf — 2025-08-20
• Signal 2 (feasibility): The whole chain is unbuilt and the filing is manual — a vendor sweep across EHS platforms, Handwerker ERPs, basik-net, labs and TRGS 519 training providers found nothing producing the Erkundung decision, the §11a Anzeige packet or the Fachkunde expiry tracking; Anzeige goes to 16 Länder authorities as PDF-and-email (Berlin LAGetSi: “schriftlich per Post oder per E-Mail”, no online form) one week before work starts — https://service.berlin.de/dienstleistung/330763/ — in force 2025-12-20
• Signal 3 (economic): Firms are already paying to over-sample defensively — ZDB’s Katrin Mees: “Sie führen jetzt häufiger Beprobungen durch, um sich abzusichern” — at €50–232 per sample and 10–14 days’ standard lab turnaround, against a hard 2026-12-19 approval deadline and ~150,300 firms in the four core exposed trades — https://www.deutsche-handwerks-zeitung.de/asbest-handwerk-schein-366205/ — 2025-09-08 Category: Regulatory arbitrage

3. The opportunity

Most compliance software sells the artefact — generate the Gefährdungsbeurteilung, fill the form, store the certificate. That is not where this hurts. The trades firm’s problem happens earlier and is a judgement call: a customer in a 1974 building wants a bathroom rebuilt. Is this Abbruch or Instandhaltung? Do I have to sample first? Do I file an Anzeige a week before I can start? Do I need the approval that becomes mandatory in December?

Get it wrong in the cautious direction and you burn money and — worse — time. A lab sample runs €50–150, TÜV SÜD charges €232, and a full Gutachten with site visit is €450–650. The self-sampling kits sold through OBI and toom that trades firms actually reach for cost €69.90 for bound materials and €139.90 for tile adhesive and plaster — but their standard turnaround is 10–14 days, with express at 3–5 working days (bau.bi). Two weeks of waiting on a job that might be half a day of work is the economic core of this market.

Get it wrong the other way and you have an Ordnungswidrigkeit under §§49/50 GefStoffV, a fine, civil claims from the client or your own employees, and in individual cases criminal liability under the StGB — on top of having put your people in front of friable asbestos.

So the firm faces a decision it is not equipped to make, several times a week, with money on both sides of the error. The regulator declined to define the boundary. The client has no duty to investigate. The lab only answers the question you already decided to ask.

That is the product: not the paperwork, the verdict. Answer “does this job trigger anything, and what exactly,” in two minutes, before anyone drives anywhere — and leave behind a dated record of why that answer was reasonable.

We know what firms do instead, from the ZDB’s own head of sustainable building, Katrin Mees: “Sie führen jetzt häufiger Beprobungen durch, um sich abzusichern, da sie nach wie vor allein in der Verantwortung für die Gefährdungsbeurteilung stehen” — they now sample more often just to cover themselves, because they alone carry the Gefährdungsbeurteilung (Deutsche Handwerks Zeitung, 2025-09-08). That is the market, already spending money, in the least efficient way available.

The incumbents are aimed elsewhere, and the gap is total. A vendor sweep across EHS platforms (Quentic, WEKA, iManSys), Handwerker ERPs (Streit, Label, pds, Sander & Doll, TAIFUN), the Handwerk-specific Gefährdungsbeurteilung tool basik-net, the labs (IVARIO, TÜV SÜD), and the TRGS 519 training providers found not one product that produces the Erkundung decision, assembles the §11a Anzeige, or tracks the Fachkunde expiries. basik-net comes closest — Handwerk-targeted hazard-assessment software at €25/mo — and its product page does not mention Asbest, GefStoffV, TRGS 519, Anzeige or Genehmigung at all.

What exists instead is: editorial content explaining the rules, blank Musterformulare from BAuA and the Länder, a lab that analyses a sample you already decided to take, and a two-day course selling the certificate the Anzeige requires. The Gutachter and remediation firms sell the sampling — they have no incentive to build a tool whose main output is often “you don’t need us on this one.” The ZVSHK’s own asbestos page is a link hub to other people’s PDFs.

There is also a reason the market will not fix itself. The ZVDH makes the structural point: a firm that samples properly and quotes compliantly loses the job to the firm that does not. Without a cheap, fast, defensible way to reach the right answer, the honest firm is competitively punished — which is precisely the wedge.

4. Target market

• Primary customer: Owner or Meister of a German trades firm with 3–25 employees working in existing buildings — Elektro, SHK, Dachdecker, Maler, Bodenleger, Trockenbau. Revenue roughly €400K–4M. The kind of firm where the boss personally decides which jobs to take and has no safety department.
• Why they buy: They are being handed criminal-adjacent liability for a substance they did not install, on buildings they did not survey, by a rule their own association calls unimplementable. Pre-1993 stock is most of their work. Every quote on an old building is now a risk decision.
• Rough TAM reasoning: The four core exposed trades alone are ~150,300 firms: Elektro 49,113 (ZVEH, 2025), SHK ~48,000 (ZVSHK, 2025), Maler und Lackierer 37,954 (Bundesverband Farbe, 2025-06-30) and Dachdecker 15,241 (ZVDH, 2025-12-31). Add Bodenleger, Trockenbau and Fliesenleger and it is plausibly 180–200K. The size band fits: ~78% of Dachdecker firms employ fewer than 10 people. At €79/mo, 2,000 firms is ~€1.9M ARR; 5,000 is ~€4.7M — around 3% of the core four.
• Why now for them: The presumption, the §5a asymmetry and the Anzeige duty are already live since 20 December 2025. The company-level approval requirement has a one-year Übergangsfrist and must be evidenced from 19 December 2026 — three months out — with up to four weeks for the Genehmigungsfiktion to run. The filing window is closing now, and nothing has shipped to help.

5. Product sketch (MVP)

• Job triage in two minutes. Address and build year in, and a short guided path through what is actually being touched — which surfaces, which materials, how much, what method. Out comes a verdict: no trigger / sample first / Anzeige required / approval-required Abbruch.
• Abbruch vs Instandhaltung call. The undefined boundary, made explicit and consistent, with the reasoning shown and the relevant TRGS 519 and GefStoffV passages cited — so the firm can defend the call, and so two estimators at the same firm answer alike.
• Building-age lookup. Pre- or post-31 October 1993 resolved from the address where possible, with the client’s §5a information request generated when it is not.
• §5a information request, sent for you. A dated written request to the Veranlasser for building year and construction history — the document that shifts what can be shifted and proves you asked.
• Anzeige packet per Bundesland. The right form for the right authority, pre-filled, including the named employees with their Fachkunde and arbeitsmedizinische Vorsorge attached, ready a week before work starts.
• Fachkunde expiry ledger. TRGS 519 Anlage 4C certificates run six years and the Anzeige names individuals — so the roster tracks who is currently valid and warns before a certificate lapses and quietly disqualifies a crew.
• The file that survives an inspection. Every verdict, request, sample result and filing kept per job, dated, exportable.

6. AI angle — what’s load-bearing

The load-bearing work is classification under an ambiguous rule, which is precisely what the regulator refused to specify and what a decision tree alone cannot carry.

A trades firm describes a job in its own words and its own trade vocabulary — “Bad raus, Fliesen runter, neue Leitungen” — and the product has to map that onto an unsettled legal boundary, identify which building materials in that era plausibly contain asbestos in that position (tile adhesive, floor coverings, Nachtspeicheröfen, roof sheets, window putty, spray coatings), and produce a verdict with a defensible rationale. The output is not a form; it is an argument the firm can stand behind.

Second load-bearing job: keeping that reasoning current across 16 Länder with different forms, authorities and submission channels, plus TRGS updates and the clarifications still expected on the Abbruch boundary. Remove the language work and you have a static PDF checklist — which already exists, free, from every Handwerkskammer, and which nobody uses on a Tuesday morning with a customer waiting.

7. Localization angle

Germany-only and deliberately so. The 31 October 1993 date, §5a’s asymmetry, TRGS 519’s Fachkunde tiers, the risk bands at 10,000 and 100,000 fibres/m³, the 19 December 2026 approval deadline and sixteen different Arbeitsschutz authorities are all German-specific. The product is German-language by necessity — trades firms will not work in English — and sold through German channels: Handwerkskammern, Innungen, Fachverbände and the trade press.

It is not a global play wearing a local coat. The nearest neighbours — Austria and Switzerland — have their own asbestos regimes and would each need the same depth of work. That narrowness is the moat.

8. Business model — path to $1M–$5M ARR

• Pricing: €79/mo for firms up to 10 employees, €149/mo up to 25; €49 one-off per Anzeige packet for firms that will not subscribe yet, which doubles as the low-friction entry.
• ACV: ~€950 for the small tier, ~€1,790 for the larger. Well inside a trades firm’s software wallet — they already pay comparable sums for ERP and time-tracking seats.
• Rough math to $1M ARR: ~950 firms on the small tier (€79 × 12 × 950 ≈ €900K) plus a modest packet-fee line. Under 1% of the plausibly exposed base.
• Rough math to $5M ARR: ~4,500–5,000 subscribed firms with tier mix, which realistically needs a channel deal — a Handwerkskammer or Fachverband bundling it for members — rather than pure direct sales.
• Expansion path: asbestos verdict → the rest of the GefStoffV hazard set the same firms face (KMF/mineral wool, PAK, PCB, lead) → the full Gefährdungsbeurteilung file → multi-site and Generalunternehmer tiers where a main contractor screens its subcontractors’ filings.

9. Go-to-market wedge — first 100 customers

1. Sell against the December deadline, in the trade press that is already covering it. handwerk magazin, Deutsche Handwerks Zeitung, handwerksblatt, bau.bi and the trade-specific outlets (boden-wand-decke, malerblatt, dachdecker.org) have all run the association fight. A free “are you ready for 19 December?” checker that produces a real per-firm answer is a story those outlets already want, aimed at readers who already know they have a problem.
2. Lead with the scope question: “Are you one of the 1,165 — or one of the 220,000?” That gap is the regulator’s own number against the trade’s, it is genuinely unresolved, and every firm wants to know which side it is on. A free scope check answering it per-firm is the single most compelling hook available, and it is also the product’s core function.
3. Go through the Innungen and Handwerkskammern, not around them. These bodies are in the business of telling members about exactly this and are currently doing it with PDFs and evening seminars. Offer the tool free to an Innung’s members for a quarter in exchange for one mailing; a single mid-sized Innung is several hundred firms. Their public position — that the rule is unworkable — predisposes them to anything that makes it workable. Caveat from the numbers: only about 25% of Maler firms are Innung members, so this channel reaches a committed minority and cannot be the whole plan.
4. The TRGS 519 course rooms are the qualified lead list. Every firm sending someone to a €595-and-up Anlage 4C course has self-identified as in scope and has just spent two days being told how exposed it is. Partner with training providers (Asbest-Akademie, NAV e.V., HWK course programmes, Dachdecker-Verband) to put the tool in the course pack.
5. Free §5a request generator as the top of funnel. The client information request is genuinely useful, takes thirty seconds, and requires the firm to enter the job — which is the same data the paid verdict needs. Give it away without a signup wall.
6. Direct to the loudest trades first. Bodenleger and Maler have the sharpest wallpaper-and-adhesive exposure to the Abbruch boundary and the least in-house safety capability. Start where the ambiguity costs the most.

10. Build complexity — justification

Medium. The application itself is a guided intake, a classification layer and a document generator — no integrations required to be useful on day one, which is what keeps time-to-value short. The real work is domain depth: encoding TRGS 519 risk bands and Fachkunde tiers, the era-and-material knowledge that drives the sampling call, and sixteen Länder notification variants with their own forms and authorities. That last part is unglamorous, slow, and precisely why it is defensible.

A technical founder with a safety-engineering or Baurecht advisor ships a credible v1 for two or three trades in 12–16 weeks, then widens trade and Land coverage. What it cannot be is a weekend wrapper — get the verdicts wrong and you have shipped liability.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Decision-support and document preparation. Must be positioned as such, never as a Gutachten or an accredited analysis — the sampling itself stays with accredited labs and sachkundige personnel.
Ethical — no harm / dark patterns	✅	Tension to manage honestly: the firm wants “no sampling needed” and the product must not be tuned to please them. Calibrate cautious, log every verdict, and route genuinely uncertain jobs to sampling rather than guessing. Worker exposure to friable asbestos is the thing being prevented.
Market exists (evidence above)	✅	Named associations on the record, a dated statutory deadline, ~564,000 firms, real fines.
1–5 person team can build this	✅	One or two builders plus a domain advisor, 12–16 weeks to a usable v1.
Launchable with <$50K / ₹40L	✅	Main costs are domain advice and Länder form coverage, not infrastructure.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	17/20	Hair-on-fire: live duty, criminal-adjacent liability, felt on every quote for a pre-1993 building, with money lost whichever way the firm errs. Docked because a share of firms will keep ignoring it until an inspector or an incident arrives.
Demand evidence	15	14/15	Five named association chiefs plus a named firm owner on the record, a Bundesrat clarification request, a dated statutory deadline, a regulator-vs-trade scope gap of 1,165 vs 220,000 firms, and firms already paying to over-sample defensively (ZDB: “häufiger Beprobungen … um sich abzusichern”). Docked one point because nobody yet pays for this decision specifically.
Build feasibility	15	11/15	No integrations needed for v1 and the document generation is standard. Docked for genuine domain depth: TRGS 519 encoding, era-material knowledge, and sixteen Länder variants that must be right rather than approximately right.
Distribution clarity	15	12/15	Unusually concrete: Innungen and Kammern with member mailings, TRGS 519 course rooms as a pre-qualified list, a trade press already running the story, and a dated deadline to sell against. Docked because association deals move on association timescales.
Revenue mechanics	15	11/15	€79–149/mo fits the wallet and $1M needs under 1% of the exposed base. Docked because German trades SMBs are famously slow software buyers and the €5M path leans on a channel deal that is not yet in hand.
Time to first revenue	10	7/10	The December deadline creates urgency and the €49 packet pulls revenue forward, but v1 needs real domain build before it can be trusted with a verdict. Realistically 8–12 weeks post-launch.
Defensibility	10	6/10	Sixteen Länder of PDF-and-email filing variants, verdict calibration and association relationships compound, and a verified zero-competitor field gives real runway. Still docked: an incumbent Handwerker ERP or basik-net could bolt this on, and the imminent TRGS 519 rewrite could flatten the ambiguity that creates the value.
Total	100	78/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required — and German fluency plus a credible safety or Baurecht advisor are mandatory, not optional. This is a market where being wrong is a liability event, and where a founder who cannot hold a conversation with a Meister will not sell anything.

Key assumptions to validate (3–5)

1. Assumption: Firms will trust a software verdict on a question carrying criminal exposure. How to test: Put 30 real past jobs in front of 15 Meister, show the verdict and its reasoning, and ask whether they would act on it or still phone the Kammer. If they still phone, the product is a lead magnet, not a business.
2. Assumption: The Abbruch/Instandhaltung boundary can be classified consistently enough to be safe. How to test: Have a safety engineer and a Baurecht lawyer independently classify 50 representative jobs, then measure agreement — with each other and with the product. Low human agreement is itself the finding: it would mean the honest output is a risk range and a recommendation to sample, not a confident verdict.
3. Assumption: The 19 December 2026 deadline is actually driving behaviour rather than being ignored. How to test: Ask five Arbeitsschutz authorities how many low/medium-risk approval applications they have received to date, and ask three Innungen what their members are doing.
4. Assumption: Firms will pay rather than use the free Kammer PDF. How to test: Run the €49 Anzeige packet as a standalone offer before building the subscription and count conversions.

Risk flags

1. Liability transfer. A wrong “no trigger” verdict could expose workers and implicate the vendor. This shapes the product: calibrate cautious, state limits plainly, carry insurance, and keep the sampling decision auditable. It also argues for positioning as decision support with a documented rationale rather than as an answer.
2. The TRGS 519 rewrite is the sharpest timing risk, and it is imminent. A fully revised TRGS 519 is expected in the second half of 2026 — summer at the earliest — and it is explicitly expected to clarify which activities count as Abbruch and which as Instandhaltung (dbz.de). That cuts both ways and must be watched weekly. A clear boundary makes the rules easier to encode and the verdicts safer, but it also shrinks the ambiguity that creates today’s willingness to pay — the product would shift from resolving the boundary to applying it, which is a smaller job and an easier one to clone. Build so the classification layer is swappable, and assume the rewrite lands mid-build rather than after launch.
3. Deadline drift. A postponement of the 19 December 2026 approval requirement, or a softening that pushes the Erkundung duty back onto the Veranlasser, would blunt the urgency the whole go-to-market leans on. The associations are lobbying for exactly that. Track the Bundesrat and BMAS actively and verify the deadline still stands before any launch push — vendor blogs have invented commencement dates in this space before.
4. Slow-buying market. German trades SMBs buy software reluctantly and through trusted channels. Direct-to-web conversion will likely disappoint; the Innung and Kammer route is probably the real business and it is slower.
5. Incumbent bolt-on. Streit, pds, Label, Taifun and the Gefährdungsbeurteilung platforms all have the customers already. Speed and depth are the only defence in year one.
6. Adverse selection on the ethics. The firms most eager for a tool that says “no sampling needed” may be the ones least inclined to sample when they should. Product tone and calibration have to resist the customer’s preference, which is commercially uncomfortable and non-negotiable.

14. Structured verdict

Score:                  78/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       German-fluent technical founder with a safety-engineering or
                        Baurecht advisor on the cap table; comfortable selling through
                        Innungen and Kammern rather than through a growth funnel
Time to revenue:        8–12 weeks post-launch; €49 packet fees sooner
Capital to launch:      €30–45K ($33–50K) — mostly domain advice and Länder coverage
Top 3 assumptions to validate first:
  1. Meister will act on a software verdict for a criminal-adjacent call
     — 15 interviews over 30 real past jobs
  2. The Abbruch/Instandhaltung boundary classifies consistently — two independent
     experts over 50 jobs, measure agreement before trusting any automated verdict
  3. Firms pay rather than use the free Kammer PDF — sell the €49 packet standalone first
Kill criteria:
  - Abandon if independent expert agreement on the Abbruch boundary is below ~70% —
    the product cannot be more confident than the experts it encodes
  - Abandon if fewer than 5 of 15 interviewed Meister say they would act on the verdict
    without also calling the Kammer
  - Abandon if the 19 December 2026 approval deadline is postponed by more than a year
    or the Erkundung duty is shifted back onto the Veranlasser

15. Next step — 1-week validation sprint

• Day 1–2: Assemble 50 real jobs from three friendly firms across different trades — actual quotes on pre-1993 buildings. Have a safety engineer and a Baurecht lawyer classify each independently as Abbruch or Instandhaltung, sampling or not. Measure their agreement with each other. This single number decides whether the product is a verdict engine or merely a triage-and-document tool.
• Day 3–4: Build nothing but the §5a information request generator and the Anzeige packet for one Bundesland. Take both to two Innungen and ask directly whether they would mail it to members.
• Day 5: Offer the €49 Anzeige packet to 40 firms sourced from a TRGS 519 course list and the trade press comment sections. Count paid conversions, not sign-ups.

Falsifiable outcome: proceed only if (a) the two experts agree on at least ~70% of the 50 jobs, (b) at least one Innung commits to a member mailing, and (c) at least 3 of 40 firms pay €49 without a call. Miss any of the three and the shape is wrong — most likely it becomes a documentation tool sold through associations rather than a verdict engine sold direct.