Skip to content

Case studies

Two engagements written up in full: what the business was doing before, what we built, and what it does now.

Written in Lahore

Short summaries of the same work sit on the portfolio. This page is the long version, and it is the one to read if you want to know how the work was actually done.

The engagements written up on this page
EngagementWhat it isWhere it runs
Polaris ERP and POSOur ERP, deployed on siteRetail and wholesale, multi-branch
Bonnet.aiOur agent platform, built on itBack office workflows

Polaris ERP and POS

A retail group ran each branch on its own till software and reconciled the day by hand every evening. Stock counts disagreed between shops, and the only way to answer a question about last week was to ask someone to build a spreadsheet.

What we built

  • One system of record across branchesStock, billing and reporting sit in a single PostgreSQL database instead of a spreadsheet per branch. A transfer between two shops is one row, not two people phoning each other.
  • Questions in Urdu or English, answered from live dataThe owner types a question in the language they think in. Polaris resolves it against the live schema and returns rows, not a chart it invented. Nobody on the floor writes SQL.
  • Every write gated, logged and reversiblePermissions decide who can take each action, an audit row lands with every write, and a mistake caught within five minutes can be undone instead of corrected by hand at month end.
  • Your server or oursThe deployment target is a decision the business makes. The data stays wherever it is put, which is the answer most owners want before anything else.
See PolarisFull product detail at polariserp.app

Where it stands

Polaris is our own product and it is in production. The counts here come out of the system itself; everything else about the product lives on polariserp.app.

Live businesses
19
Built-in reports
52

Asked on the shop floor

اس ہفتے سب سے زیادہ کون سا پروڈکٹ بکا؟

Which product sold the most this week? The answer comes back as rows out of the live database.

Bonnet.ai

Our agent platform, and the second thing we run in production. It exists because back office work is usually too irregular to script: every attempt to automate it breaks on the exceptions, and the exceptions are most of the job.

  1. 1Read the process as it actually runsWe sit with the person doing the work and write down every step, including the ones that only happen when something goes wrong. Most of the useful automation is in those.
  2. 2Model it as a graph, not a scriptSteps that can fail, retry, or need a human become nodes in a LangGraph workflow with explicit state. A linear script hides the failure paths; a graph forces you to name them.
  3. 3Put a human at the decisions that cost moneyApprovals, refunds and anything that writes to a ledger pause for a person. The agent prepares the decision and shows its working. It does not sign off on its own.
  4. 4Ship it behind the tools they already useThe workflow runs against the systems already in place. Staff keep their existing screens; the change they notice is that the queue is shorter.

What we are not claiming

No percentage saved, no hours reclaimed, no efficiency figure. Bonnet.ai has not been running long enough for us to hold a number up and defend it, and the only numbers on this site are ones we can open the system to prove. What shipped and what it does is everything above.

Every figure on this page, and where it comes from

If a number here cannot be traced to a system we can open in front of you, it should not be here. Ask us to show any of these on a call.

Figures claimed on this page and their sources
FigureValueHow it is counted
Live businesses19Businesses running Polaris in production today.
Permissions157Counted from the permission table that gates every action.
Built-in reports52Reports shipped in the product, run against the live ledger.

A third write-up is not ready

There is retrieval work in progress that we would rather describe once it has run in production for a while. Until then it stays off this page. If it is close to what you need, ask and we will talk through it directly.

  • Retrieval over internal documents
  • Multi-step tool use
  • Human review at every write

Tell us what your evenings are spent fixing.

A 30 minute call. We look at how the work moves through your business now and say where the time goes. Messages on the WhatsApp line get an answer the same working day.

Book a callSee the short version