Vorzeige-Produktfamilie

Betriebsplattform für technische Dienstleister

Von der ersten Nachricht bis zum unterschriebenen Arbeitsschein — schneller kalkulieren, sauberer disponieren und die abendliche Büroarbeit abschaffen.

🌐Die Zusammenfassung ist auf Deutsch verfügbar. Die detaillierten Abschnitte unten werden noch auf Englisch angezeigt, während die Übersetzung läuft.

Ein Betriebssystem für Reparatur-, Installations- und Wartungsbetriebe. Es verwandelt unstrukturierte Anfragen — Fotos, Sprachnachrichten, WhatsApp-Nachrichten — in strukturierte Aufträge, erstellt kalkulierte Angebote zur menschlichen Freigabe, schnürt Arbeitspakete für den Techniker und macht aus einem gesprochenen Einsatzbericht einen sauberen Servicedatensatz, Rechnungsgrundlage und Wartungshistorie.

Produktkonzept

The Technical-Service Operations Platform is a complete operating layer for businesses that send people out to fix, install or maintain things: appliance and electronics repair, heating and plumbing, electrical, HVAC, vehicle and equipment workshops, IT and facilities service. Their revenue depends on quoting quickly, arriving prepared, and billing accurately — yet most of that work is done by hand, twice, and often late at night.

It is not a single chatbot and not one clever automation. It spans the whole job lifecycle: an enquiry arrives with a photo and a vague description; the system extracts what it can, spots what is missing, and asks one precise question; it retrieves the asset’s history; it drafts a priced quotation that a human always approves; it schedules the visit and hands the technician a job package; and afterwards it turns a 40-second voice note into a structured service report, invoice input and an updated maintenance record.

The unifying insight is that technical-service businesses do not lose money on the wrench-time — they lose it in the paperwork around it: slow quotes that let a competitor win, technicians arriving without the right part, and reports written from memory at 21:00 that are thin, late, or never billed. The platform keeps the human in charge of price and safety while removing the transcription, the chasing and the re-typing.

Kundenproblem

Today an enquiry arrives as a WhatsApp photo of a broken boiler and the message “can you fix this, how much?”. Someone has to recognise the model, guess the fault, decide whether it is worth a site visit, and reply — often hours later, by which point the customer has messaged two competitors.

Quotations are the revenue bottleneck. Preparing one means finding the right parts, prices and labour estimate, then typing it up. Because it is slow, quotes are delayed or never sent, and jobs are lost purely on response time.

Technicians are dispatched with incomplete information. The address is right but the model number, the access notes, the parts likely needed and the last service are all in someone’s head or a different system, so a first visit becomes two.

After the job, the report is written from memory in the evening. It is inconsistent, misses billable materials, and arrives days later — delaying the invoice and weakening any warranty claim. Maintenance history is scattered across notebooks, photos and email.

The economic consequence is a business that is busy but leaky: quotes lost to slowness, revenue lost to under-billed materials, capacity lost to repeat visits, and cash lost to invoices that go out a week late — none of it measured.

Vorher und nachher

5Today

1. Photo enquiry answered hours later, if at all
2. Quote typed by hand — often too late to win
3. Technician dispatched without model or history
4. Report written from memory at 21:00
5. Materials forgotten; invoice sent days later

Busy but leaky — quotes, materials and cash all slip.

5With the platform

1. Enquiry read and quoted within minutes
2. Quote priced from the firm’s list, human-approved
3. Technician arrives with model, history and parts
4. Voice note becomes a structured report on the spot
5. Materials captured; invoice input ready same-day

Same crew, same tools — fewer leaks, faster cash.

	Today	With the platform
Quote turnaroundillustrativ	hours / next day	minutes
Billable materials capturedillustrativ	from memory	from the report
Invoice sentillustrativ	days later	same day

Textalternative (barrierefreie Beschreibung)

1. Today: Photo enquiry answered hours later, if at all → Quote typed by hand — often too late to win → Technician dispatched without model or history → Report written from memory at 21:00 → Materials forgotten; invoice sent days later
2. With the platform: Enquiry read and quoted within minutes → Quote priced from the firm’s list, human-approved → Technician arrives with model, history and parts → Voice note becomes a structured report on the spot → Materials captured; invoice input ready same-day

Wie es zusammenpasst

1 / 6

KundeKanalKIDatenspeicherGeschäftsregelnGeschäftssystemPersonal

Phase 1: Enquiry with media — A photo, message or voice note arrives.

Textalternative (barrierefreie Beschreibung)

1. Enquiry with media — A photo, message or voice note arrives.
2. Read & understand — Image, speech and text become structured fault and intent.
3. Retrieve context — Asset history and likely parts are pulled in.
4. Draft the quote — Priced from the firm’s own list — never invented.
5. Human approves — A person signs the price before it is sent.
6. Dispatch & report — Technician gets a package; the report and invoice follow.

End-to-End-Arbeitsablauf

1 / 8

1. Voice note after the jobPersonal

   “Pump swapped, two seals, plus the valve I had.”

2. Transcribe & structureKI

   Work, parts and time separated.

3. Match parts to catalogueKI

   Pump and seals matched; valve low-confidence.

4. Cross-check job packageGeschäftsregeln

   Extra parts weren’t expected.

5. Flag unconfirmed itemGeschäftsregeln

   Valve held for office verification.

6. Office confirms from stockPersonal

   A human verifies the extra part.

7. Report + invoice inputGeschäftssystem

   Service record and billing assembled.

8. Maintenance history updatedGeschäftssystem

   Per-asset record kept current.

PersonalKIGeschäftsregelnGeschäftssystem

Schritt 1: Voice note after the job — “Pump swapped, two seals, plus the valve I had.”

Textalternative (barrierefreie Beschreibung)

1. Voice note after the job — “Pump swapped, two seals, plus the valve I had.”
2. Transcribe & structure — Work, parts and time separated.
3. Match parts to catalogue — Pump and seals matched; valve low-confidence.
4. Cross-check job package — Extra parts weren’t expected.
5. Flag unconfirmed item — Valve held for office verification.
6. Office confirms from stock — A human verifies the extra part.
7. Report + invoice input — Service record and billing assembled.
8. Maintenance history updated — Per-asset record kept current.

Ausführliche Praxisbeispiele

Happy path

A photographed boiler fault becomes a priced quote before the competitor replies

A four-technician heating and plumbing firm near Linz. — A WhatsApp message at 08:12 with a photo of a boiler error display and “kein warmes Wasser, was kostet das?”.

1. The image is read: the boiler brand and model are identified from the label, and the fault code on the display is extracted.
2. The message is understood in German; intent “no hot water · repair · quote” is captured.
3. The phone number matches an existing customer; the last service and installed model are retrieved.
4. The fault code maps to two likely causes and the parts usually needed; the system marks the diagnosis as probable, not certain.
5. A draft quotation is prepared from the firm’s own price list — labour band, likely parts and a call-out fee — with a clear “subject to on-site inspection” clause.
6. The dispatcher reviews the draft, adjusts one part, and approves. The quote is sent in German within minutes.
7. On acceptance, a visit is scheduled and a job package is created with model, fault code, access notes and the parts to load on the van.

Ergebnis: A quote goes out in minutes instead of hours, built on the customer’s actual model and history, with the human setting the final price and the technician arriving prepared.

Exception path

A voice report flags a billable material the technician forgot to mention

The same heating firm, end of day. — A technician records a 45-second voice note after a repair: “Pumpe getauscht, zwei Dichtungen, plus das Ventil das ich noch hatte.”

1. The voice note is transcribed in German and structured into work performed, parts used and time on site.
2. The parts “pump, two seals, one valve” are matched to the catalogue; the valve is low-confidence because it was “one I still had”.
3. The system does not silently bill it. It flags the valve as an unconfirmed billable item for the office to verify against stock.
4. It cross-checks the job package: the pump was expected, the seals and valve were not — so it asks the technician to confirm the extra parts before the report is finalised.
5. The dispatcher confirms the valve from van stock; the service report, invoice input and maintenance history update together.
6. Because the material was captured, it is billed rather than absorbed as invisible cost.

Ergebnis: A material that is routinely lost to memory is captured and billed, while an uncertain item is verified by a human instead of guessed — protecting both revenue and trust.

Different market

A vehicle-workshop intake turns a WhatsApp voice message into a prepared inspection

A two-bay independent car workshop in Graz serving a mixed Austrian and Albanian-speaking clientele. — A returning customer sends an Albanian voice message: “Makina po bën një zë kur frenoj, a mund ta shoh nesër?”

1. The voice message is transcribed and understood in Albanian; intent “noise when braking · inspection · tomorrow” is extracted.
2. The vehicle is matched by the customer’s number plate on file; the last service and mileage are retrieved.
3. The system generates the missing questions a brake complaint needs — front or rear, how long, any warning light — and asks them in Albanian.
4. An inspection quotation is prepared from the workshop’s labour rates; a qualified mechanic approves it before it is sent.
5. An appointment is offered against real bay availability; on acceptance the technician receives a job card with history and the reported symptom.
6. After the inspection, a voice report becomes the service record, the invoice input and an updated maintenance history for that plate.

Ergebnis: A vague spoken complaint becomes a prepared, correctly scoped inspection in the customer’s language, with the mechanic — not the AI — approving the work and the price.

Wer bei jedem Schritt verantwortlich ist

1 / 9

CustomerAIBusiness rulesOffice / technicianSystems

Schritt 1 — Customer: Sends photo + question — “What’s wrong, how much?”

Textalternative (barrierefreie Beschreibung)

1. Customer: Sends photo + question — “What’s wrong, how much?”
2. AI: Read label & fault code — Model and error extracted.
3. Systems: Retrieve asset history — Last service and model.
4. AI: Map fault → likely parts — Marked probable, not certain.
5. Business rules: Price from list — Labour band + parts + call-out.
6. Office / technician: Approve & adjust quote — Human sets the final price.
7. Systems: Send quote — In the customer’s language.
8. Customer: Accepts — Customer confirms.
9. Systems: Build job package — Model, history, parts to load.

Produktmodule

Multichannel enquiry intakemvp

Captures enquiries from WhatsApp, phone, web form and email so nothing is lost, and understands them in German, English and Albanian.

Media & document capturemvp

Reads photos of equipment, labels, error displays, receipts and documents; transcribes voice notes and calls.

Missing-information detectioncore

Knows what a given job type needs and asks the one or two precise questions required before quoting or dispatch.

Quotation accelerationcore

Drafts a priced quote from the firm’s own price list and labour bands — always for human approval, never auto-sent.

Parts & materials identificationcore

Identifies likely parts from the model and fault, matches them to the catalogue and flags uncertain matches for review.

Scheduling & job packagescore

Books against real availability and hands the technician a package with model, history, access notes and parts to load.

Voice-to-service reportmvp

Turns a spoken field summary into a structured, editable service report with work done, parts used and time on site.

Customer updatesextension

Keeps the customer informed — on the way, delayed, completed — in their language, with the office in control of tone.

Invoice preparationcore

Assembles invoice input from the approved report so billing goes out same-day, not next week.

Maintenance history & warrantyextension

Builds a per-asset service history and organises warranty and defect cases with the evidence attached.

Owner analyticsextension

Shows quote turnaround, win rate, repeat-visit rate and under-billed materials — making the leaks visible.

Eingaben und Integrationen

• WhatsApp, phone, web-form and email enquiries
• Photos of equipment, rating labels, error displays and damage
• Voice notes and voicemail from customers and technicians
• Supplier parts catalogues and price lists
• The firm’s own labour rates and quote templates
• Existing customer, asset and vehicle records
• The invoicing/accounting tool

Nutzer und Käufer

Täglicher Nutzer

The technician (voice reports, job packages) and a back-office dispatcher (intake, quoting, scheduling).

Prozessverantwortlicher

The workshop manager or owner who defines price lists, labour rates, quote templates and approval thresholds.

Wirtschaftlicher Käufer

The owner — the person who feels lost quotes, repeat visits and late invoices directly in cash flow.

Technischer Administrator

Usually us (managed) or a tech-comfortable staff member; connecting a WhatsApp number, a price list and the invoicing tool is the setup.

Letzter Entscheidungsträger

The owner or managing partner. The buyer and the person in pain are the same, so cycles are short.

KI-Fähigkeiten

• Speech recognition for customer and technician voice notes
• Image understanding for models, labels, fault displays and damage
• Language detection and understanding across DE/EN/SQ
• Information extraction (fault, asset, parts, time, intent)
• Matching described parts to the supplier catalogue
• Drafting quotes, reports, clarifying questions and customer updates

Deterministische Fähigkeiten

• Prices, labour rates and quote totals come from the firm’s price list — the AI never invents a price
• Approval thresholds decide what a human must sign before it is sent or billed
• Parts availability and stock are checked against authoritative data
• Scheduling respects real availability, skills and travel — no double-booking
• Warranty eligibility and deadlines are evaluated by rules, not guessed

Objekt-Lebenszyklus

1 / 7

Status 1: Enquiry — Captured from any channel.

Textalternative (barrierefreie Beschreibung)

1. 1. Enquiry — Captured from any channel. (→ quote)
2. 2. Quoted — Human-approved quote sent. (→ schedule)
3. 3. Scheduled — Visit booked, package built. (→ dispatch)
4. 4. In progress — Technician on site. (→ report)
5. 5. Reported — Voice report structured. (→ invoice)
6. 6. Invoiced — Same-day invoice input. (→ close)
7. 7. Closed — History & warranty updated.

Menschliche Verantwortung

• The owner sets price lists, labour rates, templates and approval limits.
• A qualified person approves every quotation and every final price.
• The technician confirms parts and work; low-confidence items are verified, not assumed.
• Any safety-relevant diagnosis or certification stays with a licensed human.

Wirtschaftlicher Nutzen

• More quotes won because they go out in minutes, not hours.
• Higher revenue per job because billable materials stop leaking into memory.
• More capacity because technicians arrive prepared and second visits fall.
• Faster cash because invoices follow the job the same day.
• Evening paperwork returned to people, and — for the first time — quote turnaround and win rate as managed numbers.

Es werden keine Marktstatistiken oder finanziellen Versprechen impliziert. Alle Zahlen in den obigen Visualisierungen sind illustrative Beispiele, keine gemessenen Ergebnisse.

Risiken, Grenzen und Fehlerfälle

• A wrong price or a wrong part erodes trust fast — hence quotes and parts are drafted for human approval, never auto-committed.
• Voice transcription is imperfect on a noisy site; the system confirms billable items rather than assuming them.
• A confident-sounding but wrong diagnosis is dangerous in trades — the platform labels diagnoses as probable and keeps a human accountable.
• Under-scoped quotes can lose money; a clear “subject to on-site inspection” boundary and approval thresholds protect margin.
• Integration with the price list and invoicing tool must be exact — a stale price is worse than none, so sync and reconciliation are non-negotiable.

Produktentwicklung

Kleinste glaubwürdige erste Version

A voice-to-service report for one crew: a technician records a note, gets a clean structured report and invoice input, with the office editing before it is billed.

Professionelles Produkt

Full lifecycle — multichannel intake, missing-info detection, quotation acceleration, parts matching, scheduling, job packages, reporting and invoice preparation with an owner dashboard.

Optionale Erweiterungen

• Preventive-maintenance reminders
• Warranty & defect case management
• Customer self-service status page
• Deeper accounting and parts-supplier integration

Langfristige Plattform

A shared service-operations layer that vertical front-ends (heating, vehicles, IT, facilities) plug into, reusing the same intake, quoting, dispatch and reporting engine.

Gemeinsame Anleitung

Allgemein anwendbare Methoden — wie man validiert, pilotiert, Preise festlegt und Menschen verantwortlich hält — stehen im gemeinsamen Leitfaden, damit diese Seiten spezifisch bleiben:

• AI vs deterministic automation
• Human approval & accountability
• Measuring economic value
• Pricing hypotheses
• Design a pilot