Kuna Labs.

Forward-Deployed Engineers · Peru

We take your company's AI from pilot to production.

Engineers embedded in your operation, building AI systems in weeks — inside your ERP, your bank and your team. No slides. No endless PoCs.

3 systems in production right now in Lima
every delivery signed by a licensed accountant
Monthly close · May 2026 ● In production
5 days ↓ from 3 weeks

to close the month for the board

Automated bank reconciliation
Anomalous invoice detection
Close package ready for review
Reviewed and signed by your licensed accountant
  • SUNAT-compatible
  • Validated by a licensed accountant
  • No ERP migration

Before / after

How it's done today

  • Endless pilots

    Six to 18 months of slides, demos and diagnostics with no system in production. The AI lives in the pitch deck.

  • PDF consultancies

    They hand you recommendations in reports. Implementation becomes a task for your team — which is already at capacity.

  • Tools that don't understand Peru

    AI software built for Silicon Valley. It doesn't know the tax authority, the local chart of accounts, or how a Peruvian business actually operates.

How we do it

  • Embedded engineers

    We work inside your team for 4 to 8 weeks. Your AI reaches production by the end of month two, or we don't bill the closeout.

  • No system migration

    Your ERP, bank and current processes stay as they are. Your team doesn't learn new tools — they gain speed with the ones they already use.

  • Validated by a Peruvian accountant

    Every delivery is signed off by a licensed Peruvian accountant. Professional responsibility is covered; your team gains speed.

How we work

From diagnosis to production in weeks, not quarters. The time to run your AI is now.

We start on-site and build inside your team. By handoff, the system runs with your people — and the knowledge stays with you.

  1. 01

    Diagnosis

    One week on site understanding real workflows, available data, and actual blockers. No diagnostic theater.

  2. 02

    Embedded build

    Four to eight weeks inside your team, not in a parallel project. Compatible with the systems that already run your business — no migrations, no disruptions.

  3. 03

    Validation

    Your team operates what we build in parallel with the current process. We iterate until it passes the accountant's filter and the operations team's.

  4. 04

    Handoff

    Runbooks, dashboards and an optional monthly maintenance plan. The operation stays with your team; so does the code and the knowledge.

Active pilots

What we're building, right now, in Peru.

Each pilot operates inside the client's team and is validated by their licensed accountant. No names — confidentiality is part of the work. What follows is the pattern by sector.

Industrial services

Lima · Multi-site operation

In pilot
+3 weeks 5 days

to close the month for the board

Before

  • Monthly close +3 weeks late to the board
  • No real-time alerts on anomalous invoices

After

  • Monthly close in 5 days, not weeks
  • Automatic anomaly detection at 90%+ precision, accountant reviews in-cycle

Validated by their accountant

Textile manufacturing

Lima · 24/7 operation

In pilot
6 hrs/day 10×

more capacity from the same team · tens of thousands of USD/year recovered

Before

  • +2,000 invoices/month entered by hand by the team
  • ~6 hours/day lost per person on repetitive data entry

After

  • AI-assisted capture at 90%+ precision — accountant reviews and signs at month-end
  • Same team, 10× the capacity — tens of thousands of USD/year recovered

Validated by their accountant

Restaurants

Lima · Multi-account operation

In pilot
Several days 1 day/mo

of accountant review per close · ~1 working week recovered

Before

  • Monthly close fragmented across 5+ accounts and suppliers
  • Accountant assembles the package by hand every month (several days)

After

  • Unified monthly package, ready for accountant review
  • One day/month of accountant review, not several — equivalent to recovering a full week per close

Validated by their accountant

Who's behind this

Built by a Peruvian engineer who has spent years embedding AI inside real operations.

Photo of Alvaro Cuba

Alvaro Cuba

Founder & Forward-Deployed Engineer

I build internal AI platforms inside a global hospitality-tech company headquartered in Israel — automating workflows for thousands of employees. Before that, I designed AI workflows for 700+ employees at a European events scale-up. Returning to Peru to build the same thing, here, with Peruvian accountants.

  • Industrial Engineering — Universidad de Lima + Hochschule Bremen
  • MSc Finance — Esade Business School (Barcelona)
  • Data Analytics — Ironhack (Barcelona)
  • Prior experience: private banking (Inteligo), venture capital (Madero Capital, eSeed)

Let's talk

Let's talk about your operation.

Book 15 minutes on my calendar. No deck, no form — bring one question and we’ll see whether it makes sense to work together.

15 minutos. Sin presentación.

Trae una pregunta y vemos si tiene sentido trabajar juntos.

Prefer to write? alvaro@kunalabs.ai

Frequently asked

What people ask before the first call.

What is a Forward-Deployed Engineer?
It's an engineer who embeds inside your operation for weeks, understands how your business actually runs, and builds AI systems that live *inside* your processes — not in an isolated demo. It's the role Anthropic, OpenAI and Palantir hire to deploy AI inside large enterprises. What we do differently is bring it to Peruvian SMBs.
How is this different from a traditional consultancy?
Consultancies sell hours of slides and deliverables. We sell systems in production. Week one is real diagnosis; by the end of month two your team is operating something. If it doesn't reach production, we don't bill the closeout.
Do I have to change my ERP or current systems?
No. Your ERP and current systems stay as they are — your team doesn't learn new tools. We confirm compatibility with your stack during Phase 0 (diagnosis), before you sign the pilot scope.
What happens after handoff?
You get runbooks, dashboards and an optional monthly maintenance plan. The operation stays with your team; the code and runbooks too. There's no technical lock-in.
What languages do you operate in?
Spanish and English. Internal documentation, runbooks and operational meetings happen in Spanish. For clients with parent companies outside Peru, executive deliverables go in English.
Where does the name Kuna Labs come from?
*Kuna* in Quechua means *now* and *here* — the idea that the moment to act is the present. That's exactly what our work is about: taking AI from pilot to production, today. Built from Peru, with roots.