Show us a process where humans are doing work software should be doing

Kemeny Studio · AI operating capacity

Learning · Revenue & customers · Documents & cases

Results, not software. AI does the work.

We build and operate AI capacity around a measurable business result. One workflow and one KPI, in production in 4 to 6 weeks. Then we expand.

Define the first resultSee how it works
  • OpenAI Select Partner
  • In production in 4 to 6 weeks
  • A KPI from the start
  • No licenses

Who does what

  1. You

    Receive the result.

    You buy a bounded, measurable result with an agreed KPI. Not licenses, not development hours.

  2. The AI

    Does the work.

    It answers, follows up, qualifies, grades, validates and reconciles. Continuously, inside your rules.

  3. Kemeny

    Builds, runs and improves the capacity.

    We design, connect, deploy, supervise and continuously improve the operating system behind the result.

    The people who run it

    • Francisco Kemeny

      Managing Director

    • Catalina Aliaga

      Operations Director

    • Jonathan Olivares

      Technology Director

    • Benjamin Martinez

      Software Developer

    • Arturo Jaime

      Associate Sales Rep

    • Mauricio Baron

      AI Developer

    • Gabriel Aramburo

      Software Developer

What we put in production

Not software. Operating capacity.

You don't buy another tool to operate yourself. We put AI capacity into production around work that needs to get done.

  1. Week 1

    Define the result

    Pick the workflow, the users, the systems and the KPI.

  2. Weeks 2 to 6

    Build the capacity

    Connect real data and tools. Ship it to real users.

  3. Ongoing

    Run & improve

    Monitor, evaluate, secure and handle incidents. Push the KPI up, cut costs and find the next workflow.

How buying works

One result first.

You don't hand us your whole operation. You start with one piece of work, prove it, and expand from there.

  1. 01

    One piece of work

    Bounded: what is in and what is out.

  2. 02

    One measurable output

    What gets delivered, and where you see it.

  3. 03

    One KPI

    Baselined before we start, so the change shows up in numbers.

  4. 04

    One responsibility boundary

    What the AI does, what we supervise, what stays with your team.

Prove. Operate. Expand.

  1. 01 · Prove

    First production workflow

    4 to 6 weeks

    From USD 14K

    • One workflow, one KPI
    • Real users and real data
    • A person as fallback while it proves itself

  2. 02 · Operate

    Ongoing operation

    Monthly

    From USD 25K a month

    • Monitor, evaluate and secure
    • Improvements and model upgrades
    • Cost optimization and incidents

  3. 03 · Expand

    Full AI platform

    6 to 12 months

    From USD 60K

    • More workflows and users
    • More channels and business units
    • More volume

Reference prices. Each engagement is priced once the workflow, scope and KPI are defined. Ongoing operation is sized to the system, its usage and how critical it is to the business.

01 / 03

Start with: One type of learning content

Learning Operations

The AI produces and grades. Your experts approve.

We take over

Producing your training, and the practice and assessment around it.

What you get

More approved content. More graded practice.

Content your experts approved, and practice that gets graded and returns to each person as feedback.

The work · done by AI

  1. 01Turns your experts' material into course content
  2. 02Keeps the content up to date
  3. 03Generates activities and assessments
  4. 04Runs roleplay simulations
  5. 05Grades against your rubrics
  6. 06Gives each person feedback and coaching
  7. 07Records progress in your LMS

Measured by

  • Time from expert material to approved content
  • Practice sessions per person
  • Grading turnaround
  • Completion rate
  • Score change between assessments

Stays with your team

Your experts approve the content and own the rubrics. Your managers decide what to do with the results.

Clients today

  • Laureate ·
  • Laboratorio Chile / Teva
Define your first training result

02 / 03

Start with: One class of leads or requests

Revenue & Customer Operations

The AI answers and follows up. Your sellers close.

We take over

The layer between an inbound lead and a meeting or an order.

What you get

More qualified conversations. Fewer lost opportunities.

Every lead answered and logged, with a pipeline your sales leadership can see.

The work · done by AI

  1. 01Answers every inbound lead
  2. 02Follows up on every open conversation
  3. 03Qualifies against your criteria
  4. 04Recovers dormant opportunities
  5. 05Books meetings and processes orders and bookings
  6. 06Keeps the CRM or ERP up to date
  7. 07Escalates exceptions to a person, with context

Measured by

  • Response time
  • Contact rate
  • Qualification rate
  • Meetings booked
  • Opportunities recovered

Stays with your team

The rules and the close: who qualifies, what can be offered, when a person steps in, and the conversations that need a seller.

Clients today

  • GPG ·
  • Autorenta ·
  • Librería Nacional
Define your first sales result

03 / 03

Start with: One document or case process

Document & Case Operations

The AI prepares every case. Your team signs off.

We take over

Cases, documents and exceptions, from intake to close.

What you get

Cases closed within SLA. Evidence on file.

Shorter cycles, fewer errors and reconciliations that no longer wait on a person.

The work · done by AI

  1. 01Receives and classifies requests
  2. 02Reads and validates documents
  3. 03Builds case files
  4. 04Reconciles information across systems
  5. 05Routes approvals
  6. 06Resolves exceptions or escalates them
  7. 07Keeps the evidence trail
  8. 08Tracks SLAs and closing times

Measured by

  • Cycle time per case
  • Cases closed within SLA
  • Error rate
  • Backlog
  • Share reconciled without manual work

Stays with your team

Approvals that need a signature, policy decisions and the exceptions you choose to keep.

Clients today

  • Laureate ·
  • GPG
Define your first operations result

Published cases

Work already in production

Figures from our published cases. When a figure is reported rather than measured, we say so. We describe clients by sector, not by name.

  • Client
    School-supplies retailer, Chile
    In production since Dec 2025
    Work we took on
    Reading school supply lists and matching every item to the catalog
    Output
    The client reports seasonal headcount held while the number of schools rose
    Evidence
    97.9% of lists extracted successfully
    190,325 items across 7,563 lists. 4.3% go to human review.
  • Client
    Higher-education group, Mexico
    In production, on the client's infrastructure
    Work we took on
    Producing AI learning content and delivering it inside the LMS
    Output
    Activities inside the LMS, with grade passback. Releases go through the client's change-control committee
    Evidence
    20,000+ students
    66,000+ feedback responses so far.
  • Client
    Car rental, Mexico
    In production, with real customer traffic
    Work we took on
    Serving customers on WhatsApp and web
    Output
    Anything that needs a person goes to the client's own agents, through three escalation paths
    Evidence
    Real customers, every day
    Each escalation is taken by one agent, with automatic re-notification.
  • Client
    Marketplace operator, Chile and Mexico
    In production, with nominated users
    Work we took on
    Giving planning, logistics and management a data layer they query with AI
    Output
    Planning went from two-week test loads to a complete monthly plan
    Evidence
    14+ months of daily use
    Efficiency gain reported by us, not independently measured.
See all published cases

Clients

The number marks the domain where we take work for that client today.

  • Laureate(Learning Operations, Document & Case Operations)
  • Laboratorio Chile / Teva(Learning Operations)
  • GPG(Revenue & Customer Operations, Document & Case Operations)
  • Autorenta(Revenue & Customer Operations)
  • Librería Nacional(Revenue & Customer Operations)
  • Banorte
  • Midea
  • BBDO
  • MediaBrands
  • Beta San Miguel
  • Unifrutti
  • UVM
  • UCentral
  • CIAL
  • BNAmericas
  • Examedi
  • Finvero
  • GMS
  • Instance
  • QServus
  • SES
  • AutoExpertos
  • Akya

Questions

Before you buy

What exactly does Kemeny take responsibility for?

For the capacity that produces the result: that it runs, that exceptions get handled, that performance is measured and that it keeps improving. You receive the result; we operate what produces it.

What is a “result” in practice?

A bounded piece of work with a measurable output. For example: every inbound lead answered and followed up, every trainee's practice graded, or every case of one type closed within SLA. It has a KPI and a baseline.

How do we define the first result?

Start with the Workflow Fit Check. It is free and returns a first read on the work, a suggested KPI and a scope. If it fits, we agree on the work, its boundary and how it will be measured in a working session.

How long does the first result take?

4 to 6 weeks. In week 1 we pick the workflow, the users, the systems and the baseline KPI. In weeks 2 to 5 we connect the systems and go live with real users, with a person as fallback. In week 6 we measure the KPI before and after.

What does it cost?

Reference prices: the first production workflow starts at USD 14,000. Ongoing operation starts at USD 25,000 a month. Expanding into a full AI platform starts at USD 60,000, over 6 to 12 months. We set the final number once the workflow, scope and KPI are defined.

What does ongoing operation include?

We stay accountable for keeping the system working, measured against its KPI. We monitor production health, incidents, quality tests and model evaluations. We update prompts and models, ship small improvements, and optimize cost, performance and security.

How is performance measured?

Against the KPI we agree when we define the result, with a baseline taken before we start. We report it to you as we run the work.

What happens when the AI cannot complete the work?

It escalates. Every result has a defined exception path: what the AI can't close goes to our supervision or to the person on your team we agreed on, with the context attached. Those exceptions are also what we use to improve it.

What remains with our team?

The decisions: the rules, the approvals that need a signature, and the work you choose to keep. Each domain shows what usually stays with you, and we write it down when we define the result.

How does it expand after the first result?

Once the first workflow proves its KPI, we add workflows, users, channels and business units, up to a full AI platform. That typically takes 6 to 12 months, one step at a time and each with its own measure.

How does Kemeny supervise the AI?

Our team follows the work the AI does: reviews exceptions, checks performance against the KPI, adjusts rules and instructions, and improves the system over time. You set how far it can act on its own.

Is our data secure?

We connect only what the work needs and keep a record of what the AI did. Where you want a person to sign off, there is an approval step. Security and integration constraints are part of defining the first result.

Do you replace our systems?

No. The AI works on top of the CRM, ERP, LMS and channels you already use.

Do we need a technical team?

Not to run it. You need someone who owns the process and access to the data and systems. We run it day to day.

Got another question?

Tell us what work you want off your plate and what's holding you back. We'll answer for your case.

Talk to us

What do we take first?

Pick where to start. The Fit Check turns it into a first result with a KPI and a scope. It's free, and the answer shows on screen.

Define the first result

If it's not a fit, we'll tell you.