Cardinal

Your workflows, redesigned with business judgment and AI.

Cardinal finds the workflow where AI can move your business the most, builds it on your systems and rules, and gets it running.

See redesigned workflows Let's talk

This page is the English summary of cardinal.cl. The rest of the site is in Spanish.

Cardinal takes the workflow to level 3 (redesigned around AI), whether it starts at level 1 or level 2.

  1. Manual

    The workflow runs on spreadsheets, emails and people typing.

  2. AI added on

    There are systems and AI in some steps; the workflow is still the old one.

  3. Redesigned around AI

    The workflow is rethought as if it were being designed today: agents run it end to end and people decide the exceptions.

What a redesigned workflow looks like.

Three examples from three areas, with their screens. The first one is running; the other two are proposals with no client yet. The screens are in Spanish, the language they were built in.

Accounts payable

Area
Finance
Level
Redesigned around AI
Status
In production

The agent reads each invoice, validates it with the tax authority, proposes the account, emails a question only when something is unclear, and posts it to the ERP.

The inbox of invoices to review: supplier, amount, claim deadline and the reason each one did not go through on its own. The batch of invoices posted to the ERP, with the system's response for each line.
Illustrative example · fictitious data

Event budgeting

Area
Commercial
Status
Proposed, no client yet

From budget to event close: AI drafts each budget from past events, every payment is matched to its line, and management sees the real margin.

A new budget built from similar past events, with the budgeted and actual margin of each one and the draft that AI proposes. The management dashboard for the quarter: budgeted and actual margin for each event, and the income statement.
Illustrative example · fictitious data

Project office

Area
Management
Status
Proposed, no client yet

The master plan recalculates with every change: what to start today, which decision is due, and how confident each milestone date is.

The project dashboard: business milestones at risk, workstreams holding up progress and the committee's decision agenda. A milestone date with its confidence level: the spread of many simulations of the plan and the risks that move it most.
Illustrative example · fictitious data

Business judgment and building on the same team.

A senior who understands the business and a forward deployed engineer (FDE) who builds, with agents instead of a pyramid. The model grows by adding pairs.

Eduardo Cyterszpiler.

Eduardo Cyterszpiler founded Cardinal. He ran Integration Consulting's Southern Cone offices with full P&L responsibility, led revenue and pricing engagements as Senior Director at Simon-Kucher, and headed Marketing Strategy for the Americas at NTT DATA, with a 100+ consultant team across 8 markets. Global Executive MBA, IESE.

Eduardo on LinkedIn

Shared risk.

Cardinal invests in the diagnostic and earns its return once the system is running.

  1. No-cost diagnostic*. One workflow, two weeks. It delivers how the workflow runs today (with its starting cost, where that can be measured), the redesign, a first version running on the company's own data, and a priced proposal.

  2. A fixed-price build, close to cost, invoiced against deliverables once they are implemented.

  3. Operation for a monthly fee that follows usage where there is a unit to count (invoices processed, budgets approved) and is fixed where there isn't. With a cap. Unit, fee and cap are agreed before anything is built.

*Subject to eligibility

Cardinal doesn't charge by the hour or for recommendations.

Building with AI is cheap. The expensive part is building something with no impact.

Eduardo Cyterszpiler, founder of Cardinal