Why do factory owners wait days for industrial repair technicians?

A machine stops, and the owner spends days phoning dealers and local mechanics while the line sits idle. Here's "FixLine", a breakdown desk that one person could build with Claude in a weekend.

The problem

Who feels it: Owners and plant heads of small and mid-sized factories running CNC machines, compressors, injection moulders, boilers and control panels. When a machine fails, they call the original dealer, wait for a service engineer, or hunt for a local technician through contacts. Nobody has a clear picture of the fault, the spare part needed or when help will arrive.

Breakdowns are frequent. In an ABB global survey of 3,215 plant maintenance decision-makers conducted in July 2023, 88% of Indian industrial businesses reported unplanned outages at least once a month, against 69% globally (ABB).
Each hour is expensive. The same survey put the cost of unplanned downtime for a typical Indian business at close to Rs 70 lakh an hour. This is based on respondents' estimates, and it reflects larger plants more than micro units (ABB).
Many still wait for things to break. A report citing the ABB survey says 19% of Indian respondents still rely on run-to-fail maintenance, and NITI Aayog sees about 1% of GDP in reducing breakdown and downtime losses (Asianet Newsable / ANI).
Market size signal. The same report projects India's asset performance management market to grow from USD 810.6 million in 2024 to USD 2.25 billion by 2033 (Asianet Newsable / ANI).

Who's already working on it

Predictive maintenance startups such as Presage Insights (Rs 5.6 crore seed, 2025) use sensors to catch faults early, which suits larger plants. EasyFix runs managed repair for multi-location enterprises. IndiaMART lists local industrial repair shops. We didn't find a simple tool that helps a small factory diagnose the fault, reach the right technician fast and keep a record of who fixed what.

The solution idea

"FixLine" is a breakdown desk for small factories: describe or photograph the fault, get a likely diagnosis and parts list, and have the job sent to a ranked list of nearby verified technicians who must accept within minutes.

Fault intakeThe operator sends a photo, the machine nameplate, the error code and a voice note on WhatsApp; the app turns it into a clear job card with make, model and symptoms.

First-line diagnosisSuggests likely causes, safe checks the in-house team can try, and the spare parts to keep ready, with a clear note that it is guidance, not a certified repair.

Technician dispatchSends the job card to technicians filtered by machine type and distance, asks for acceptance and arrival time, and moves to the next name if nobody accepts.

Technician recordTracks response time, first-time fix rate and repeat failures for each technician and dealer, so the owner knows whom to call next time.

Machine historyKeeps every breakdown, part and invoice per machine, and flags machines that fail repeatedly for planned maintenance or replacement.

How Claude builds it

Which Claude capabilities do the work

Claude CodeBuilds the web app, the WhatsApp intake, the dispatch logic, the database and the tests.

Claude API in the productReads nameplate photos and error screens, structures the job card, suggests causes from manuals and drafts messages to technicians.

SkillsA "breakdown-playbook" skill with safety rules, triage questions per machine type and message templates in English and Hindi.

ArtifactsA clickable job card and technician board mock-up to show factory owners before the build.

Connectors (MCP)Google Drive for machine manuals and Google Sheets for pilots that already keep a maintenance register; Vercel to deploy.

Scheduled tasksAcceptance timeouts that trigger the next technician, a daily open-jobs summary and a monthly machine reliability report.

Weekend build plan

Saturday morning — interviews and data: Talk to three factory owners and two local technicians. Collect the last ten breakdowns from one factory: machine, fault, who fixed it and how long it took.
Saturday midday — prototype: Ask Claude for an artifact showing the job card, the dispatch screen and the technician record. Show it to the owners.
Saturday afternoon — core app: Use the starter prompt below in Claude Code to build machines, technicians, jobs and the status timeline.
Saturday evening — intake: Connect WhatsApp Business or email and have the Claude API turn photos and voice-note transcripts into structured job cards. Test on real samples.
Sunday morning — dispatch: Add the technician ladder with acceptance timeouts as scheduled jobs, plus messages to the owner at each step.
Sunday afternoon — records: Compute response time and first-time fix rate with plain code and unit tests. Add the machine history page.
Sunday evening — ship and pilot: Deploy to Vercel. Onboard two factories and ten technicians in one industrial area, and log every breakdown for a month.

Starter prompt for Claude Code

Build a Next.js + TypeScript + Postgres web app called "FixLine" for small Indian factories.

Entities: Factory, Machine (make, model, serial, location), Technician (skills by machine type, area, phone, verified flag),
Job (machine, symptoms, photos, status, accepted_by, eta, resolved_at, parts_used, cost), JobEvent.

Flow: operator reports a fault (photo + text or voice transcript) -> Claude API returns JSON
(make, model, error_code, symptoms, likely_causes[], safe_checks[], parts_to_check[], urgency)
-> app offers the job to the top 3 matching technicians -> if none accepts within 15 minutes, a scheduled job offers it to the next 3
-> owner sees a timeline from report to resolution.

Rules:
- Diagnosis is shown as guidance only; never tell anyone to bypass guards, interlocks or lockout procedures.
- Response time and first-time fix rate are deterministic code with unit tests, never model output.
- Store every message and photo with a timestamp.
- Templates in English and Hindi live in /skills/breakdown-playbook.md.

Start by proposing the schema and file structure, then build job creation and the timeline first.
Add fixtures of 15 real-style fault reports and a test that checks the parser output.

Monetisation

Suggested pricing, not yet validated.

Free: one plant, up to 5 machines and the breakdown log.
Rs 1,999 a month per plant: unlimited machines, diagnosis, dispatch and reports.
Technicians join free; a small booking fee per completed job once volume exists.
Not validated: test willingness to pay in the pilot.

Riskiest assumption, and how to test it

Assumption: The idea assumes that enough skilled, reliable technicians exist in an industrial area and will accept jobs through an app. For specialised machines, only the dealer's engineer may be able to fix the fault.

Test first: Before building, run it by hand in one industrial estate for a month: a WhatsApp group of ten technicians, a shared sheet of jobs, and a phone call when a factory reports a breakdown. Measure time from report to technician arrival against the factories' usual experience. A reasonable bar: most jobs accepted within an hour and owners asking to keep using it.

Repair advice on industrial machines carries safety risk; keep diagnosis as guidance, insist on lockout procedures, and avoid steps that could void warranties. Verify technician identity before listing them.