Case study · PlasterTEX
Assessment deliveredAI agent builtOne AI co-pilot for the whole operation, not just the sales reps
How a growing building-materials manufacturer in Bali chose to build its AI around the way its team already works, on WhatsApp, with customers and with each other, instead of asking a field team to learn a new system.
PlasterTEX is a fast-growing decorative-plaster manufacturer in Bali, supplying builders and villa projects across Indonesia.
PlasterTEXPlasterTEX did not start by choosing software. It started with how its people work. Its business is sold in person, on building sites and in villa projects across Bali, by a small team that knows the product and its customers well, and almost every conversation, with a customer, a rep or the warehouse, already happens on WhatsApp. So the question was never which system the team should move to. It was how to give the tool they already use a memory, the company's rules and its numbers, so that what a rep learns on site is kept, a quote goes back in the same conversation, and leadership sees the day without chasing anyone. That is how a company that sells on relationships grows without losing them: the same people, the same way of talking, and the repeating work handled in the background.
PlasterTEX makes decorative plaster finishes for architects, developers and villa projects across Bali, and it is growing: more projects, more quotes, more product shipped, and a plan to reach new islands. Its strength is in the field. Reps meet customers on site, learn what each project needs and when, and build relationships that turn into orders. Success brings its own weight. Every quote is worked out by hand from the area to cover, the finish, the primer and the discount, and goes back and forth on WhatsApp before it is final. What a rep learns on a site lives in their phone and their memory until someone writes it down. Samples leave the warehouse for site visits, paid orders have to reach the warehouse with the right address, and the founders need to know what happened today across the whole team. The team carries all of it, and carries it well, because the relationships matter too much to hand to tools that do not fit how they work. The founders asked the next question: how AI could take the repeating work off their people, in the tool they already use, without changing what makes them good at selling.
We started with the founders, and then sat with the operations lead who runs the sales team in the field. Across four sessions we went through the work as it is done every day: how a site visit becomes a lead, how a lead becomes a quote, how a sample leaves the warehouse, how a paid order reaches it, and how leadership learns what happened that day. We read the company's own contact base and its quotes, and worked out its pricing with the team: the coverage of each bucket, the primer that goes with it, and the discount each person may give.
One finding shaped everything after it: the business already runs on WhatsApp. So the answer was not a new system to learn. It was one agent, on one WhatsApp number, in English and Indonesian, that customers and the team talk to the way they already talk to each other.
Before any build, the specification was written with the founders and approved by them line by line. Then we built it to that specification and handed it over.
One rule holds it together: the agent prepares, and a person decides. A quote goes to the customer when the rep sends it on, and a discount above a rep's limit goes to the founder.
- The first message. Today customers' questions arrive on WhatsApp and wait for whoever is free to answer. With the agent, they are answered in English or Indonesian from the company's own answers, and a request for samples goes straight to the rep who covers that district.
- The quote. Today a quote is worked out by hand from the area, the finish, the primer and the discount, and goes back and forth before it is final. With the agent, the rep sends a voice note from the site, the agent works out the buckets and the primer by PlasterTEX's own rules, and a branded proforma comes back on the phone for the rep to send.
- The lead. Today what a rep learns on a site lives in their phone until someone writes it down. With the agent, a voice note after the meeting is enough: the agent asks only what is missing, the record is complete, and the next follow-up is dated.
- The contact base. Today the company's contacts sit across several sheets. With the agent, they are one clean list, sorted by type of buyer and ready to be followed up at the right moment in each project.
- The day. Today the founders find out what happened by asking. With the agent, the day's priorities become each rep's morning list, the agent checks what was done, the operations lead gets a daily brief and the founder a weekly one.
- The paid order. Today a paid order reaches the warehouse by message and memory. With the agent, the warehouse receives the paid invoice and the delivery address directly, and the proof of delivery comes back to the operations lead.
- An Opportunity Assessment across sales, operations and the warehouse
- A written specification, approved by the founders line by line
- One AI agent on WhatsApp, in English and Indonesian, for customers and the team
- A quote engine that works to PlasterTEX's own pricing and coverage rules
- A clean, deduplicated contact base
- A daily brief for operations and a weekly brief for the founder
What we found is a company that sells in person and runs on WhatsApp, whose knowledge lives with the people in the field. The agent keeps that way of working and gives it a memory: the quote comes back in the conversation, what was learned on site is kept, the contacts are worked at the right time, and the day is visible without anyone having to chase it.
Built to the specification agreed with PlasterTEX. This page describes the work and the targets: no figure on it is a measured outcome.
Let's talk
Let's find what AI can do for your business.
Every company is different. The best place to start is a chat.