Most manufacturers handle inbound enquiries the same way. A lead arrives. Someone qualifies it by phone. The details go, usually by WhatsApp or email, to whoever prepares quotes. That person checks the price, drafts the email, attaches the documents, copies the right people and sends it.
On paper, that takes 15 minutes. In practice, the person sending the quote has a different main job. They run operations, sell in the field or commission machines at a customer site. The quote waits for a spare moment. That moment might come in an hour, tomorrow or later in the week. Sometimes it never comes, and nobody tracks the missed chance.
Nobody is negligent here. The business simply never had a dedicated quoting function. People did it alongside their real work, and nobody could say how many quotes went out, when, or at what price.
The workflow, rebuilt
In one Indian manufacturer, Mitochondria's agentic AI systems took over this workflow. A team member reviews a lead and submits the details. The system checks that everything essential is there: contact, company, product, assigned salesperson. If anything is missing, it asks.
Once the details are complete, the system builds the quote with the right price, attaches the company profile and product documents, copies the right stakeholders and sends it. It does this at any hour, at any volume, whoever is available.
Prices come from a structured catalogue. When a salesperson agrees a negotiated price, the system applies it. Technical notes go in when they are needed. Formatting, attachments and routing are the same on every quote.
That consistency matters more than it sounds. Manual quoting varies by person. Some remember the company profile; some forget. Some copy all the stakeholders; some miss one. Across hundreds of quotes, that variation chips at professionalism and makes it impossible to know what was promised to whom.
Within weeks, every qualified lead was receiving a complete quote the same day.
Assets the business never had
To run the system, the manufacturer first needed something it had never had: a single structured catalogue of machine codes, product names and prices. Before, that information lived in people's memories, old emails and catalogues of uncertain age.
Modern AI can work from unstructured information. Even so, the business chose to organise its data first, and the catalogue now serves far more than quoting. It would keep its value even if the system were switched off.
The system also created the business's first complete audit trail of quoting. Every quote is recorded: what went out, when, to whom, for which product, at what price and through which salesperson. For the first time, leadership can answer basic questions about its own sales activity. How many quotes went out last month? Which products draw the most enquiries? Which salespeople bring in the most leads?
This is a pattern worth knowing. Preparing an agentic system often builds information infrastructure that outlasts the system itself.
The workflow kept growing
The first workflow is never the final one. Within months, the manufacturer needed more from the same system:
Spare parts: enquiries handled alongside full machine quotes, each with its own pricing and documents.
Exports: USD pricing, with the format and commercial terms export buyers expect.
Negotiated deals: a salesperson's agreed price applied in place of the standard rate.
Non-standard builds: custom technical notes for unusual configurations.
Each request was scoped, built and deployed, some within days. This is workflow evolution, not scope creep. Scope creep means the brief was wrong. Workflow evolution means the business could only see its next needs once a working system showed what was possible. The change log becomes a record of the organisation learning about itself.
What changed in practice
Speed is the visible change. Coverage is the important one. Before, an unknown number of qualified leads never received a quote. They were qualified, passed on, and lost in the gap between someone's main job and the admin of writing a quote. Now every qualified lead receives a response.
The time returned matters too. Quoting used to take up managers' and senior salespeople's attention. That time now goes to running operations, closing deals and visiting customers.
Errors on everything the system handles have all but disappeared. Prices come from the catalogue. Attachments are included by default. Routing follows the agreed rules. Templates are maintained in one place.
Start where the work is clearly broken
This was not a complex deployment. It is a single agentic workflow: structured input, defined logic, a finished output, sent. The value it created far exceeds its complexity.
Manufacturers often reach first for the hardest problems: production scheduling, demand forecasting, predictive maintenance. Those are real opportunities. They also take a long time to deliver and depend on data foundations that may not yet exist.
A simple, repetitive workflow that is visibly broken, and that eats expensive people's time, is a better first step. It goes live quickly. The results show up immediately. The organisation gains confidence. And the catalogue and audit trail it creates become the base for everything that follows.
How to choose the first workflow
High volume, repetitive work that follows recognisable patterns.
Visibly broken today, so improvement is obvious to everyone.
Consuming senior time that should go elsewhere.
Clear inputs and outputs, such as an enquiry in and a quote out.
Low risk with review, where a person can check before anything binds.
Leaves an asset behind, such as a catalogue or a record, that serves the next workflow.
How Mitochondria's agentic AI systems support a first workflow
The approach. The first deployment starts with a visibly broken workflow that costs senior time, such as quoting. The quotes unit (unit 61) is configured to the manufacturer's catalogue, prices, templates, and routing, and grows with the business: spare parts, export pricing, negotiated deals, and custom technical notes are added as needs arise.
What it enables. Same-day quotes on every qualified lead, a structured catalogue, a full audit trail and a system the team trusts, because it asks when unsure and never sends an incomplete quote.
The impact. Coverage, not only speed: no qualified lead goes unanswered. Senior time returns to operations and customers. And the catalogue and audit trail become the foundation for the next workflow, whether technical offers, costing evaluations or drawing change control.
Read more about manufacturing.
Sources
This piece draws on Mitochondria's own deployment work; it cites no external figures.
—
Mitochondria builds agentic AI for evidence-led core operations. Research-led and model-agnostic. ISO/IEC 27001:2022 certified, with controls aligned to the GDPR, UK GDPR, India's DPDP Act and Australia's Privacy Act. Amsterdam and Pune. Request a working session →