Quoting well takes time, and the time you spend quoting is time you do not spend selling. It happened to us: every proposal was a hand-built document, with prices from memory and one more email to follow up. So we did what we do best: we built the system we wanted to have. Today we use it every day to quote in minutes.
The problem: every quote started from scratch
An agency quotes similar services again and again, but rarely in the same way. The price depends on the scope, the scope depends on the conversation and the conversation gets lost between chats and emails. The result is familiar: slow proposals, inconsistent prices between whoever quotes and no visibility into what happened after they were sent.
We wanted three things: quote fast, quote with the same criteria every time and know what happens to each quote, even when it becomes a project.
What we built
A custom quoting system, designed for service teams and for internal use:
- A service catalog with categories and base prices, and templates per project type, so you never start from zero.
- A quote editor with items you reorder by dragging, rich-text sections, a Gantt-chart timeline, autosave and an edit lock so two people do not step on each other.
- A PDF with the company's brand, identical to the preview, ready to send by email from the same system.
- An AI assistant that quotes over the catalog (more on that below).
- Projects: when a quote is accepted, it becomes a project with milestones, meetings, minutes and accumulated value.
- Sellers and payments: a margin per seller, revenue split between participants, deposits with receipts and balances.
- Prospects and clients: lightweight tracking of whom to contact, when and with what result.
- A dashboard with accepted quotes by month, a funnel by status, conversion, and top clients and services.
The AI that quotes over your catalog
What changed our day-to-day the most is the assistant. It is not a generic chat that makes up prices: it works with tools that query real data from the system.
You ask, for example, "I need a quote for a six-page corporate site with a blog". The assistant looks up the client, checks the catalog, looks at similar quotes we have already made, calculates the totals and shows you a proposal. Only when you confirm does it create the draft, marked as AI-generated, and a person reviews it before it is sent.
- Answers from real data: prices come from the catalog and totals are calculated by the server, not by the model.
- Suggestions you apply in one click: the assistant proposes changes to a quote as cards you can apply, discard or undo.
- Search by meaning, not just by keywords, to find similar quotes.
- Attachments: upload a PDF brief and it becomes a starting point.
- With limits: the AI key is stored encrypted, there are usage caps per user and attached content is treated as untrusted.
A quote from start to finish
The editor keeps everything on one screen: the client, the project type, the seller, the validity and the detail with its totals. The quote's sections (summary, scope, deliverables, timeline) are copied into each proposal when it is created, so what a client signed never changes even if the templates do.
From there you review the document as the client will see it, send it by email with the PDF attached and it is recorded in the email history.
From "accepted" to delivered project
An accepted quote does not end the work: it starts it. The system turns it into a project with its milestones, and each milestone carries a planned date and an actual date, so you see at a glance what is on time and what is late. Meetings are scheduled with a video-call link and their minutes, and the project value is calculated by adding up the accepted quotes.
Engineering: built to last
- Money always on the server side: the browser displays, the server calculates.
- A single source of permissions for the server and the interface, with three roles: administrator, team and seller (who only sees their own).
- Data frozen on sending: company data and taxes are fixed in each quote.
- No magic server actions: every mutation goes through API routes, with a tested service layer.
- One installation per company, ready to adapt to other service teams.
Results
- Professional quotes in minutes, not hours.
- Consistent prices: the whole team quotes from the same catalog.
- Every quote, project, seller and payment visible in one place.
- A useful AI because it works with our data and always has a person at the end.
We built it for ourselves, and that is exactly why we built it well: we use it daily and improve it every week.
Does your team quote by hand and lose time on every proposal? At Codegea we develop custom systems and AI automations that adapt to the way you work. Let us talk about your project.