All Your Company's Information. One Place. Instant Answers.

Where they were

Every enquiry brought the same kind of question. Does this hotel have a pool? How many rooms does it have, and in which categories? What is the drive like from the last stop?

The answers existed, in property files spread across shared folders and in the heads of the product team and the founder. So a salesperson would search file by file, or message someone and wait, and the customer waited with them. A reply to a lead often took hours.

Where we started

With one folder. It held 52 property files covering hotels across 17 states, in four different layouts, and many of them were designed image sheets with no text a computer could search.

We turned them into one clean, organised library, built the assistant on top of it, and tested it against questions written by their own sales team. It was working on their data in 7 days.

The kind of questions the team asks now

• “Which of our Jaipur properties have a pool?”
• “How many rooms does this property have, and in which categories?”
• “What is the suggested route between these two stays?”

What changed

• The sales team replies to a lead in under 2 minutes, where it used to take hours.
• Every answer shows the file it came from, so nobody has to double-check with the product team.
• When the answer isn't in the files, the assistant says so. It doesn't guess.
• While organising the files we found a few that contradicted each other, and flagged them to their team to correct.

Where it is today

Once the first version had proved itself, we rolled it out across their full property catalogue and worked with their team on the features their salespeople asked for. We still work with them as it grows.

Where they were

The studio had finished more than 300 projects, and each one held the answer to the question every new enquiry raises: how long will this take us?

That history sat in their project management tool as loose descriptions nobody could search properly. So each quote was rebuilt by hand, from old files and from whoever remembered the last similar job, and two people quoting the same brief would arrive at two different numbers. A quote took more than four hours.

Where we started

With the past projects and nothing else. We pulled all 300+ out of the tool they already used and organised them into a library where every project is linked to the tasks inside it, such as editing, motion graphics, colour and audio, along with the hours each one took.

The kind of questions the team asks now

• “What have projects like this one taken us before?”
• “How many hours did the motion graphics take on our last three product videos?”
• “Which past projects is this estimate based on?”

What changed

• A quote takes under 2 minutes, down from more than four hours.
• Estimates land within 10% of the hours the work actually takes.
• Every estimate shows the past projects it was built from, so anyone can check the reasoning.
• The whole team quotes from the same history, so the number is the same whoever is asked.

Where it is today

Live, and used for every new quote.

Where they were

The company makes aluminium and glass door and window systems. Billing, contracts, production and purchases each kept their own records, in Tally, in Excel sheets and in WhatsApp groups, and no department could see what the others had.

Tracking a payment, a delivery of material or a contractor's bill meant calling the person who kept that sheet. Nobody could say when a payment would arrive or why it was late, and a missing document in one department would hold up a payment in another without anyone noticing until it failed.

Where we started

With finance. We built the first version on their Tally data, covering what they are owed, what they owe, and the accounts behind both.

Nobody was asked to move to a new tool. People kept entering things where they always had, and the assistant read from there.

The kind of questions the team asks now

• “Which customer payments are overdue, and by how many days?”
• “How much material came in this week?”
• “What is still pending on contractor billing this month?”

What changed

• More than 50 people across finance, the manufacturing units, sales and HR ask the same assistant, in plain language.
• A question that used to need a phone call to another department is answered straight from the records.
• The teams kept Tally, Excel and the routines they already had.
• It runs on the company's own Google Cloud account, so their data stays under their control.

Where it is today

Fully rolled out across finance, manufacturing, sales and HR. We continue to work with them, improving it and adding AI agents that take routine tasks off each team.