How a building materials distributor learns each morning which RFQs are worth quoting

Who is the company and what problem does it have?

Picture a distributor of building and plumbing materials, with one warehouse and a sales team of three. It sells pipes, fittings, adhesives, boards and insulation to construction companies and installers.

Every day, Markonex shows requests for quotation from companies buying exactly these products. The problem is not a lack of requests but time: each morning every salesperson opens the list, reads the same requests and decides in their own way which ones are worth quoting.

How did the team work before the agent?

The team had following filters and received notifications, but triage was still manual. Requests with short deadlines got lost among the others, and the big ones, with long product lists, were left for "later" and sometimes expired before anyone quoted.

• The same request was read by two or three people.
• There was no clear order: what expires today, what is close to the warehouse, what best matches the stock.
• The sales manager only found out at the end of the week which requests had gone unanswered.

What do you need to start an agent like this?

Markonex Agents are Claude agents, from Anthropic, that a company creates in the Markonex Agents web app. The agent works for the company, not for a single employee, and uses the company's own Anthropic API key: model usage is paid directly to Anthropic, from the company's prepaid credits, not to Markonex. The cost shown in the app is an estimate.

• The company is verified by the platform (the "Verified by the platform" badge).
• The administrator turns on Agents and MCP under My company → AI assistants → Agents and ticks the information notice.
• In Markonex Agents, under "API keys", the company adds its Anthropic key; only the last 4 characters remain visible.
• The "New agent" wizard has these steps: API key, Model, Instructions, Permissions, Schedules, Limits (monthly budget, per-run budget, maximum runs per day) and Summary.
• Permissions in plain language: the agent can read requests for quotation, following-filter matches, the company profile and the catalogue, and can create new following filters. It is not allowed to send offers or messages.

How do you try it safely, before using real data?

Every company has a separate test environment: a "(TEST)" company and up to two test partner companies, with their own data, no emails, no push notifications and no ANAF or e-Factura. There you can publish a few sample requests from the test partner company and watch how the agent finds and ranks them.

In the test environment the agent confirms its own actions, so you can see the whole flow quickly. In the real account the same actions wait for a person to approve them. The test environment can be reset or deleted at any time.

Step 1: write the instructions, as you would for a new colleague

Instructions are ordinary text, in Romanian or English. You write them once, at the "Instructions" step of the "New agent" wizard, and you can change them at any time.

1. Agent instructions

   Every working day, look for the requests for quotation published since the last run that match our catalogue of pipes, fittings, adhesives and insulation. Put first the requests expiring within 3 days, then those in the counties around our warehouse, then those with long product lists. For each request, explain why it matches and which catalogue products would go into an offer. Do not send offers and do not write to anyone.

Step 2: set the schedule and the cost limits

Under "Schedules" you choose Monday to Friday at 8:00, Romanian time. Under "Limits" you set a monthly budget, a per-run budget and a maximum number of runs per day, so the agent never spends more than you decided. The agent can also be started by hand with "Run now".

Step 3: the agent searches, ranks and writes the report

On every run the agent searches active requests for quotation, reads the matches from the company's following filters and compares what is requested with the products in the catalogue. The report appears in Markonex Agents, in the agent's "Runs" tab, where the team reads it over the first coffee.

1. The agent

   I found 9 new matching requests. High priority: a request for PPR pipes and fittings, 46 lines, expiring the day after tomorrow, about 30 km from the warehouse; a request for facade adhesive and mesh, expiring tomorrow. Medium priority: 4 insulation requests from neighbouring counties. The rest match partially. I also noticed 3 requests for ball valves that no existing following filter covers.

Step 4: the agent fills the gaps in the following filters

Creating a following filter is an internal action the agent may perform directly, without approval, if it has the permission. When the instructions ask for it, the agent adds a new filter for the kinds of requests the existing ones miss, and from the next day the team also gets the usual notifications for them.

Offers stay in the hands of the salespeople: they open the requests from the report and quote from the platform, as before.

What changes for the sales team?

The morning starts with a short, ordered list instead of a search. Each salesperson knows which requests to take, and short-deadline requests no longer slip by. As an illustrative estimate, daily triage can drop from about an hour shared by three people to a few minutes of reading the report; the real result depends on the volume of requests and on how clear the instructions are.

What limits and safeguards does the agent have?

The agent has a single tool: the Markonex MCP server. It does not browse the web, open files or run code. It has no memory between runs: each run starts from the data in the platform and the instructions the company wrote. In this scenario the agent has no commercial write permissions at all, so there is nothing for anyone to approve: it only reads, reports and, if asked, creates following filters.

• It works only with the permissions ticked when it was created, within the rights of the user whose identity it uses.
• Schedules run at most once an hour, on Romanian time; a run may start a few minutes late.
• Prepared commercial actions wait for approval by a user with the agents-management permission, who is notified in the app, by push or by email; the approval is valid for 24 hours and only while the run is still active.
• An approved action runs in the name of the person who approved it, and the activity log shows "Name (via Anthropic agent #id)".
• The agent cannot issue or cancel invoices, sends nothing to e-Factura, makes no payments and cannot administer the company or its users.
• If the administrator turns Agents off for the company, all agent access is revoked at once.