Case study · Sell and Serve · Global industrial distributor

An industrial distributor turns quotes around in under an hour instead of a day.

How a global distributor put an AI layer inside its customer inbox, so routine requests are answered from the company's own systems and specialists handle only the exceptions.

  • 80%+of routine requests answered without a person drafting
  • Under 1 hourquote turnaround, from same day at best
  • 20+countries on the same layer

A global distributor that wins on response time.

The company

A group selling and servicing industrial components through branches and field teams in more than twenty countries, with a catalogue of several hundred thousand parts. Its customers run plants and infrastructure. When a machine stops, the supplier who answers first usually wins the order.

  1. IndustryGlobal industrial distribution and service
  2. FunctionSell and Serve: customer inbox through to order
  3. ScopeShared mailboxes across the group's branches, in several languages
  4. SystemsTwo systems of record from different eras, connected and left in place

Specialists were the bottleneck on every request.

The challenge

Several hundred requests a day arrived by email and phone, and most described a part rather than quoting a code. Answering one meant recognising the part, finding it among several hundred thousand items, checking stock, looking up that account's contract price and confirming the dispatch cut-off. Then writing back, in the customer's language.

The people doing that were the same specialists the business needed for technical selling. Quotes went out the same day at best, often the next. Follow-ups happened when someone remembered. The business could only sell as much as its people could type.

Nobody was doing the job badly. They were doing it by hand.

An AI layer inside the inbox, answering from the company's own systems.

What we built

We put an operating layer between the shared mailbox and the two systems of record. It reads each request, identifies the account and the part, and drafts the reply with live stock, contract price and dispatch cut-off drawn from the client's own data. Nothing is estimated. One request, taken from the system running today:

09:14 · inbound · shared mailbox

“Do you have the part with the angled fitting in stock? What does it cost us, and can it still make today's dispatch?”

A twelve-year account, asking three questions at once, in a language nobody on shift that morning speaks fluently.

  1. 01

    Identifies the account and the part

    The sender is matched to their contract terms, location, credit position and order history. The described part is narrowed by meaning to the item this account has bought eleven times before.

  2. 02

    Pulls the facts from the systems of record

    Live stock, the contracted price and today's dispatch cut-off, each shown with its source.

  3. 03

    Decides whether it is certain enough

    Clean answers, and it drafts. An ambiguous description, stock split across sites or a price off contract, and it hands the email to the right person with the research done. It never guesses at a price or a date.

  4. 04

    Drafts, and learns from the edit

    In the customer's language and the company's voice. A person reads it and clicks send. Whatever they change feeds back in.

Elapsed: about nine seconds.

From quoting to order placement. Once quoting had proved itself in the numbers, the same layer was extended to the next step. When a customer accepts a quote or sends a purchase order, the system reads it, matches it line by line to the quote and the account, and places the order in the system of record. A person approves anything that does not match. Order placement had been the largest block of manual work after quoting, and automating it freed a further 30% of the team's capacity.

The client decides what goes out on its own.

Who holds the dial

On day one the system drafts and the team approves. As accuracy on a category proves itself month after month, the client releases that category to go out without review. Stock and order-status answers usually go first. Pricing and technical advice stay with a person far longer, sometimes for good. Every step is the client's decision, on their own evidence.

Anything that is a commitment, whether a price, a stock figure, a delivery date or a warranty term, can only come from data read out of the client's own systems. Where the system cannot find the answer, it routes to a person rather than filling the gap.

Same team, far more ground covered.

Results

  1. Routine requests answered without a person drafting: 80%+Specialists read and send. They no longer research and write.
  2. Quote turnaround: under an hour, from same day at bestIn a market where the first answer usually wins the order.
  3. Order placement automated: a further 30% of capacity freedAccepted quotes and purchase orders go into the system of record without re-keying. The time released has gone into outbound sales, so the same team now calls customers instead of only answering them.
  4. Every quote followed upQuotes that go quiet are chased on schedule, recovering business that had been won and then forgotten.
  5. Rolling out across more than twenty countriesThe same layer is now moving further down the order cycle: purchasing and stock, invoicing, cash. Each function is committed on the strength of the previous one's numbers, not a slide.

How it is built: inside the client's own cloud, in their own region. No copy on our side, named and revocable access, and audit logs held by the client, under a signed data protection agreement.

Start with whether you are this shape.

Is this you

01

Is a shared inbox the front door to your revenue?

Orders, quotes and service requests arriving by email and phone, in more than one language.

02

Do customers describe what they need rather than quote it?

A large catalogue, negotiated prices per account, and answers that depend on live stock.

03

Are your specialists spending the day on answers a system could draft?

The people who win technical accounts, tied up re-typing the routine.

Three yeses and it is the same shape. The first move is a short scan of your inbox and your systems of record, so the business case rests on your own numbers before anything is built.

Start a conversation Next case study: global apparel manufacturer All case studies