> ## Documentation Index > Fetch the complete documentation index at: https://docs.apollospace.ai/llms.txt > Use this file to discover all available pages before exploring further. # Manufacturing & Industry > Factories, manufacturers, and industrial suppliers: chase suppliers on POs and delivery dates, turn inbound requests into order records, and the backoffice that links the floor and the office — with purchase approvals, payments, and contracts always in your hands. It's 6:40am. The morning shift is about to start and one of the cells will stall in two hours: the raw material that was due yesterday never arrived, and nobody called the supplier to find out where it is. Meanwhile, three purchase orders issued last week have no delivery confirmation, a customer emailed asking for a quote that got buried in the inbox, and the invoice for the last steel batch is in a PDF someone forgot to file in the right folder. It isn't sloppiness — it's too few hands to watch suppliers, orders, and the floor at the same time. **What leaves your plate:** chasing each supplier on delivery dates, turning a loose inbound request into a structured record, flagging the material that's about to be late, keeping documents organized, and the daily operations digest all start happening on their own — with context, within budget, and with purchases, payments, and contracts always waiting on you. ## How it works, in one line In manufacturing the bottleneck is rarely making the thing — it's **keeping suppliers, orders, and dates under control without a late delivery ambushing the line.** The agent doesn't approve purchases or sign contracts: it makes sure every PO is followed up, every request becomes a record, every delay shows up before it turns into downtime — and it hands you what needs judgment and money. ```mermaid theme={null} flowchart LR A[Trigger
PO issued, new request,
date approaching, end of day] --> B[Agent pulls context
from Company Brain + CRM] B --> C[Decides + uses tools
CRM, Boards, Outbound,
Routines, Docs, Folders] C --> D{Sensitive?
purchase, payment,
contract} D -->|No| E[Chases / records /
flags / files / digests] D -->|Yes| F[Proposes + calls you
with full context] E --> G[Becomes memory
the next step already knows] F --> G ``` ## Concrete flows ### 1. Chase the supplier — purchase orders and delivery dates **Trigger:** a purchase order was issued and the promised date is approaching, or it passed without confirmation. The agent pulls the supplier, the item, and the agreed date from the [CRM](/en/features/crm), checks that supplier's delivery history in the [Company Brain](/en/features/brain) — who tends to run late, what the real lead time is — and drafts a crisp chase: "confirming delivery of PO 4821, 1020 steel, due 06/28; still on track?" The send goes out by [email or WhatsApp](/en/features/outbound) via [Marcus](/en/agents/marcus), with the option to **ask for your approval before each send** while you don't yet trust the tone. The supplier's reply lands as activity on the card, and the order's stage updates — confirmed, late, rescheduled. The result: you don't find out about the delay when the line stops. You find out when the agent chases, **days ahead**, and it only calls you if the supplier goes quiet or gives a bad date. What to ask and when to chase lives in the Brain, not in a fixed script. A critical-component supplier with a long lead time gets chased earlier and more firmly than a shelf item — the agent pulls that calibration from the history you already ingested. ### 2. Turn a loose inbound request into a structured order record **Trigger:** a request arrives by email, [WhatsApp](/en/integrations/whatsapp), or a form — a customer asking for a quote, a distributor sending a purchase order as plain text. The agent reads the message, extracts what matters — item, quantity, desired date, terms, who asked — and creates a structured record: a [lead or opportunity in the CRM](/en/features/crm) if it's a quote, or a [task](/en/features/boards) on the order board with every field filled and the original attachment alongside. Nothing sits in a lost screenshot; the conversation history becomes activity. If the request matches an item already in the [Company Brain](/en/features/brain) — spec sheet, price list, availability — the agent notes the reference on the card. ```mermaid theme={null} flowchart TD A[Loose request arrives
email / WhatsApp / form] --> B[Agent extracts
item, qty, date, who] B --> C[Creates structured record
CRM or task on the board] C --> D{Becomes a quote
or proposal?} D -->|No| E[Stays organized
and traceable] D -->|Yes| F[Drafts it and
proposes it for you to approve] E --> G[Becomes memory] F --> G ``` The win isn't "a bot that types orders." It's that **no request gets lost in the inbox** and whoever opens the board in the morning finds each one already in the right shape — ready to become a quote, which you review and approve. ### 3. The routine that flags late inbound materials — and keeps documents in place **Trigger:** a [routine](/en/features/rotinas) the agent schedules for itself sweeps the open purchase orders every morning. The routine matches POs against promised dates and surfaces what's **late or about to be late**. For each one, the agent opens a [proactive notification](/en/features/rotinas) in your bell, drafts the supplier chase (flow 1), and — if the delay threatens a customer delivery — flags it on the production board. In the same motion, when an invoice, a quality certificate, a catalog, or a supply contract arrives, the agent files the document in the right [folder](/en/features/pastas) — by supplier, by order — and ingests the content into the [Company Brain](/en/features/brain), so the next question ("what's the certificate for batch X?") has an instant answer. **The delay that becomes downtime was almost always visible in advance.** A routine has no hectic Monday: it watches the dates every day, at the same time, and turns "nobody saw it" into a flag in your bell before the problem reaches the floor. For new suppliers or inputs, [Scout](/en/agents/scout) runs the public web research — who supplies that component, price range, references — and returns a summary with sources, no guessing. ### 4. The daily operations digest **Trigger:** a [routine](/en/features/rotinas) at the end (or start) of the day. [Athena](/en/agents/athena) — the organization's chief of staff — gathers what happened and what's coming: purchase orders that changed status, deliveries confirmed and late, new requests that came in, dates due tomorrow, documents still missing. She assembles a short, direct digest and delivers it to your bell and the channel you use. With the [Composio integration](/en/integrations/composio), that same digest can drop into the operations team's Slack or become an item in Notion — without anyone compiling a spreadsheet by hand. **Why this matters:** the picture of the operation usually lives in three heads and five tabs. A daily digest, always at the same time, with the right numbers pulled from the CRM and the boards, is what makes the backoffice truly link the floor to the office. ## What stays human The agent **proposes**, never decides alone, whenever the subject is: * **Purchase approval** — issuing or approving an order, choosing a supplier, locking the quantity and amount of an order. * **Payment** — paying a supplier, advances, refunds, any outflow of money. * **Supply contract and signature** — any commitment that creates an obligation with the supplier. * **Commercial proposal to a customer** — the quote or formal offer that goes out the door. * **Heavy external, legal, or regulatory communication**, governance decisions, and destructive actions. In those cases the agent prepares everything — the compared quotes, the supplier history, the draft order — and **waits for your approval.** Every action lands in the auditable trail, with author and justification, and autonomy grows like a ratchet per action class, not in a leap. ## Where to start Ingest your supplier list with lead times, the item spec sheets, and the documents you consult every week into the [Company Brain](/en/features/brain). That's where the agent pulls the right chase and the instant answer from. Connect a [channel](/en/integrations/whatsapp) and let the agent track open purchase orders and chase delivery dates. Start with human approval before each external message and loosen it as it earns trust. Create a [routine](/en/features/rotinas) that sweeps the dates every morning and another that delivers the operations digest. One agent, one cadence, one channel — and autonomy grows slowly. ## Next steps The full pattern of the office that sustains the operation — records, documents, chases, and the digest that connects the dots. Tracking order, date, and delivery end to end — of which manufacturing is a high-volume case. The daily date sweep and the operations digest — proactive, at the same time, every day. Take the record and the digest where the team already works — Slack, Notion, calendar, and hundreds of apps.