Services / two lines of work

We build the system,then we connect it.

Most of our clients run on spreadsheets, WhatsApp and memory. First we replace that with software shaped around their business. Then, where it earns its place, we let AI act on it directly.

01

Build

A business that has outgrown its spreadsheets doesn't need a bigger spreadsheet. It needs one system that knows its clients, its prices, its stock and its history — and behaves the way the business already does.

Below is a sample of what a system might include — not a menu, and not a boundary. Yours gets shaped in discovery.

Clients & providers

Every client, quote and conversation in one place, with the full history attached. Providers, lead times and purchase orders alongside them.

Budgets & orders

Quotes built from your real catalogue in minutes, with price snapshots frozen at the moment of sale. Last year's numbers stay last year's numbers.

Production & inventory

A pipeline that shows what's in progress, what's blocked and what's due — and stock levels that move when orders do.

Portals

Your clients check their own order status. Your providers submit their own prices. Both stop arriving in your inbox.

Reporting & documents

The reports you currently rebuild by hand, generated. Branded PDFs — quotes, invoices, work orders — produced from live data.

Integrations

Accounting, payments, e-commerce, shipping, WhatsApp. The tools you already pay for keep working, wired into the same system.

Typical first release
6–10 weeks
You own
Code, data, docs
You talk to
The engineer
02

Connect

Chat that can only talk is a demo. We give the model real, permissioned access to your systems, so asking becomes doing — and every action it takes is logged.

MCP server · alephlabs/estudio · session 0412
give me the last five budgets
→ budgets.list(limit: 5, sort: −date)
→ 5 rows · ARS 18.4M · Mar–Jul 2026
I forgot to log friday's sale — 2 Vera chairs, oak, to Marina
→ clients.match("Marina") → #0231
→ orders.create(items: 2 × VERA-OAK, price_snapshot: 2026-07-24)
→ inventory.adjust(oak_panel: −6)
✓ order #1182 created · awaiting your confirmation
MCP servers

Ask Claude or ChatGPT about your own data in plain language and have it act — read budgets, fix a missed sale, check stock. Scoped permissions, confirmations on anything destructive, a full audit trail.

Automation

The repetitive path between your tools, handled: quote accepted → order opened → provider notified → client emailed. Built as workflows you can read and change, not a black box.

Reporting & forecasting

Monthly summaries written from your actual numbers, demand and cash-flow forecasts built on your own history. Useful because the data underneath it is clean.

03 / Engagements

Three ways to start

Discovery sprint
1–2 weeks · fixed fee

We map how your business really works and hand back a scope, an architecture and a number. Yours to keep, whoever builds it.

Scoped build
6–10 weeks · fixed scope

The system, built in stages, in your hands early. Deployed, documented, and supported for 30 days after launch as standard.

Ongoing partner
Monthly · capped hours

For businesses whose software keeps growing. New modules, AI integrations, maintenance and someone who already knows your codebase.

On tooling

We don't sell a toolbox. Tools get chosen per project, for reasons we'll explain in plain language — chosen so your software is fast to change in two years, not so it matches whatever is fashionable this one.

Not sure which one you need?That's what the call is for.

Thirty minutes, no deck. Describe the mess and we'll tell you plainly whether software fixes it — and roughly what it would take.

[ Book a call ]nacho@alephlabs.dev