Our work in action

Real systems for real operators.

Practical AI builds that turn messy operational work into decisions, follow-through, and measurable outcomes.

01 / Document collection

Automated document collection

A Telegram agent that identifies client documents, files them against the right monthly checklist, and drafts the follow-up for anything missing.

Open live demo

Automated document collection Document collection product demo

The problem

Clients send their CA firm documents on WhatsApp and email, as unnamed PDFs and phone photos spread across staff inboxes. Staff spend hours working out whose file is whose, filing it by hand and chasing what’s still missing, and every month it starts again.

The solution

Kagaz is a Telegram agent. Staff forward any client file to it, and it reads the document, identifies the client by GSTIN, PAN or CIN, and files it on that month’s checklist. When it isn’t sure, it asks before filing, then lists what’s missing and drafts the WhatsApp follow-up.

02 / Finance · Reconciliation

Payment reconciliation

Order-level payout reconciliation for restaurant operators, so every short-paid rupee becomes a dispute, explanation, or deliberate write-off before the window closes.

Open live demo

Payment reconciliation Finance · Reconciliation product demo

The problem

A restaurant on Swiggy and Zomato gets paid by weekly bank transfers. Each transfer has already had commission, GST, platform fees, promo shares and refunds taken out, and nobody checks those deductions order by order. On a ₹400 dish, a restaurant can receive around ₹220 once every deduction is taken (The Week, July 2026).

The solution

Settle reads each platform’s order report and payout file for every outlet. It works out what each order should have paid under the agreed rate card and shows the gap in rupees: commission, promo share, unexplained refunds and orders that were never paid.

03 / Real estate

Real estate dev optimizer

A launch planner that reads MahaRERA filings, compares unit mixes in rupees, and keeps pricing decisions grounded in sell-out speed, interest carried, and profit.

Open live demo

Real estate dev optimizer Real estate product demo

The problem

A housing project’s profit is mostly set before construction, when the developer picks the unit mix and price, usually on instinct. When that call misses, flats sit unsold while construction loans keep charging interest. Unsold homes in India’s top seven cities rose 12% in a year to about 6.3 lakh units (Anarock, Q3 2026).

The solution

A launch planner reads every MahaRERA filing near the parcel and compares mix options in rupees: months to sell out, interest carried and profit. After launch it tracks each new filing and proposes price changes for the owner to approve.

04 / Manufacturing

Manufacturing error management

A single complaint tracker that finds the real cause, assigns the fix, sends the 8D reply on time, and shows whether the fix held.

Open live demo

Manufacturing error management Manufacturing product demo

The problem

OEM complaints reach suppliers through WhatsApp, email, portals and debit notes. No one sees them all in one place. The same defect keeps coming back, debit notes keep growing, 8D replies go late, and the supplier rating takes the hit.

The solution

We built one tracker that collects every complaint, finds the real cause and assigns the fix to the right person. The 8D reply goes out on time, and the numbers show whether the fix held.