AI and automation for small manufacturers in India

The problem

Dealers and brand buyers send orders as WhatsApp photos, PDFs or voice notes. A clerk re-types each one into Tally Prime, and mistakes in grade, size or quantity show up only at dispatch.

What we build

A WhatsApp Business API number reads each order with AI, matches items to your Tally stock-item list and price list, sends the dealer a confirmation to approve, and creates the sales order in Tally without typing.

Expected outcome

Typically removes order re-typing entirely and cuts order-entry errors to the few lines the AI flags; in our audits we look for the hours per week the office spends on this today.

How it works

1. Dealer sends the order (photo, PDF, text or voice note) to the factory's WhatsApp Business number.
2. An AI step in n8n reads the message and matches each line to the stock-item names and units in Tally Prime; unclear lines are flagged.
3. The dealer gets a tidy confirmation on WhatsApp with items, quantities and expected dispatch date, and replies ‘Yes’ or corrects it.
4. The confirmed order is written to Tally Prime as a sales order and to a Google Sheet that the production supervisor sees.
5. The sales coordinator reviews only flagged orders each morning instead of re-typing every one.
6. The owner gets a 9 am WhatsApp summary: orders received, value, and any order waiting on stock.

Systems: WhatsApp Business API · Tally Prime · n8n · Google Sheets

The problem

The daily plan is in a notebook or a whiteboard. Office staff cannot see what is running, rush orders are slotted by phone, and promised dates slip without anyone noticing.

What we build

A simple web app shows every open job, its machine, its shift and its stage. Supervisors update it from a phone in two taps. The sales desk and the owner see the same board.

Expected outcome

Expected result: the owner stops being the person who knows the plan; typically on-time delivery and machine loading become visible within the first month.

How it works

1. Open sales orders flow from Tally Prime into the board as jobs with quantity and due date.
2. The production supervisor drags each job onto a machine and shift from a phone or the office PC.
3. Operators or the shift in-charge mark stages done (cut, mould, finish, pack) with actual quantity and rejections.
4. The sales coordinator checks the board before promising a date to a customer, instead of phoning the floor.
5. If a job falls behind its due date, the owner and supervisor get a WhatsApp alert.
6. At month end the board exports output per machine, per shift and per product for costing.

Systems: Custom web app · Tally Prime · WhatsApp Business API · Google Sheets

The problem

Raw-material stock in Tally does not match the store. Reorders happen late or twice. Working capital sits in slow items while fast items run out.

What we build

Barcode scanners at goods receipt and issue keep stock live. A rules layer sets minimum levels per item and pings the purchase person with a suggested order quantity.

Expected outcome

Typically brings book stock and physical stock within a few percent, and turns reordering into a 10-minute daily approval instead of a monthly scramble.

How it works

1. Each raw-material item gets a barcode label and a minimum and reorder quantity agreed with the storekeeper.
2. The storekeeper scans goods in against the purchase order and scans issues to production on a handheld or phone.
3. Stock levels update in Tally Prime (or Zoho Inventory, if you prefer a stock module) every hour.
4. n8n checks levels daily and sends the purchase manager a WhatsApp list of items under minimum with suggested quantities and the last supplier price.
5. The purchase manager approves with one tap; a purchase order PDF goes to the supplier on WhatsApp or email.
6. A monthly report shows slow-moving items and stock value by category for the owner.

Systems: Barcode scanners · Tally Prime · Zoho Inventory · n8n · WhatsApp Business API

The problem

Inspection sheets are paper or skipped. When a customer rejects a lot, nobody can trace the batch to the shift, the operator or the raw-material lot.

What we build

A mobile checklist per product with photo capture, tied to a batch number. Results go to a shared record, and a weekly AI summary shows the top defects and where they come from.

Expected outcome

Expected: full batch traceability for customer complaints and a weekly defect picture; typically the first fixes come from seeing which shift or machine drives rejections.

How it works

1. The quality in-charge defines a short checklist per product (measurements, visual checks, pass/fail) in a mobile form.
2. The line operator or QC inspector fills it on a phone at the stage you choose, with photos, and scans or types the batch number.
3. Each record is saved to Google Drive and a Google Sheet with batch, shift, machine and raw-material lot from the production board.
4. Rejections above a set percentage trigger a WhatsApp alert to the supervisor and QC head.
5. Every Monday an AI step writes a plain-English summary: top three defects, which machine or shift, and a trend versus last week.
6. When a customer complaint arrives, the sales desk searches the batch number and shares the inspection record within minutes.

Systems: Mobile forms (AppSheet or custom) · Google Drive · Google Sheets · WhatsApp Business API

The problem

Maintenance is reactive. Service dates, oil changes and part replacements live in the mechanic's memory, so breakdowns stop production at the worst time.

What we build

A machine register with service schedules by calendar or run-hours. The maintenance technician gets WhatsApp reminders, logs each job with a photo, and the owner sees downtime per machine.

Expected outcome

Typically moves most service tasks from reactive to scheduled within two months and gives the owner a downtime number per machine for the first time.

How it works

1. Every machine is listed with its service tasks, interval (days or run-hours) and the spare parts each task needs.
2. Run-hours are taken from the production board, so schedules move with actual use.
3. The maintenance technician gets a WhatsApp reminder three days before each due task and marks it done with a photo.
4. Any breakdown is logged from a phone in 30 seconds: machine, start time, cause, fix, parts used.
5. The storekeeper sees upcoming part needs and keeps critical spares in stock.
6. A monthly dashboard shows downtime hours per machine and overdue tasks for the owner and plant head.

Systems: AppSheet or Google Sheets · WhatsApp Business API · n8n

The problem

The owner sees total sales and bank balance, but not margin per product, per customer or per month, and not until the accountant closes the books.

What we build

A dashboard that pulls sales, purchases and consumption from Tally Prime and output from the production board, so margin per product and per customer is visible weekly.

Expected outcome

Expected: a weekly margin view by product and customer; in our audits the first decision this usually changes is which low-margin items to reprice or drop.

How it works

1. Tally Prime data (sales, purchases, stock, receivables) is pulled daily into a reporting sheet or database.
2. The production board supplies actual output, rejections and machine hours per product.
3. A costing model agreed with your accountant allocates material, labour and power to each product.
4. Looker Studio or Power BI dashboards show sales, margin, receivables ageing and output, by product, customer and month.
5. A WhatsApp message each Monday gives the owner five numbers: orders, dispatch value, margin, overdue receivables and stock value.
6. The accountant and the owner review one page at the monthly meeting instead of six Excel files.

Systems: Tally Prime · Google Sheets · Looker Studio or Power BI · WhatsApp Business API