Alejandro Maza

AI at Kavak. Todo bien

[Dan Lahav @dan_lahav]

𝗧𝗵𝗲 𝗘𝗻𝗱-𝗦𝘁𝗮𝘁𝗲 𝗙𝗮𝗹𝗹𝗮𝗰𝘆: 𝗪𝗵𝗲𝗿𝗲 𝗜𝘀 𝗔𝗜 𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝗚𝗼𝗶𝗻𝗴?

Frontier AI models had a giant performance gain in coding in the Fall of 2025

Then with cybersecurity in April

This is now happening with open-weight models

We are optimistic about the long-term

But outside of a few players, we believe the world is not ready in the short-term

[a16z @a16z]

AI agents at Kavak sell the cars, underwrite the loans, coach the mechanics, and in one Mexican city, run the entire operation.

The Latin American used-car marketplace bet on agents three years ago. Today ~95% of interactions and transactions run end-to-end on AI: NPS tripled, sales conversion doubled, warranties down 26%, car loans approved in less than three minutes.

CPO & AI Officer Alejandro Maza joins a16z's Angela Strange and Gabriel Vasquez on how they pulled it off, why adoption without redesign fails, and why they deleted two years of working architecture to start over.

00:00 Intro
01:03 ML before transformers
02:23 Kavak's agent-per-customer architecture
04:59 Three bets: redesign the company, build superhuman agents, change the metrics
10:49 Agents with 2.1x conversion
14:23 Car loans approved in minutes
16:13 1.5x profits in a real city in six weeks
20:13 "Jedi Academy": training mechanics to ship agents
28:44 Destroying two years of work
32:52 Ford's factory: why adoption isn't enough
34:45 Advice for founders

@alehandromz @astrange @GEVS94

[Kun Chen @kunchenguid]

lol remember this org chart meme?

I just created a full simulation for all of them with agents, and the results blew my mind!

the simulation asked each organization to build and ship a web spreadsheet

want to take a guess who built the best product? reveal in thread below!

[Alejandro Maza @alehandromz]

@steipete awesome!

[Alejandro Maza @alehandromz]

@simonw Some labs might have started post-training to beat your benchmark. It's becoming big now!

[Alejandro Maza @alehandromz]

@steipete Didn't see the sponsor option till now! Thanks for everything this has been one of the most fun months of my life!

[Alejandro Maza @alehandromz]

@ByrneHobart @WillManidis any favorite transcript?

[Alejandro Maza @alehandromz]

@ArtirKel I did that! Think he got much better with time.

[Alejandro Maza @alehandromz]

@jasonfried Hi @jasonfried , I'm thinking of getting one for a similar use and aesthetic reasons. What are your thoughts after these months?

[Alejandro Maza @alehandromz]

Vibe Training

[kilian jornet @kilianj]

@jlsobrino Para la carga yo contabilizo 3 parametros basicamente en una escala de 1-5: Feeling durante el entreno, RPE durante entrenamiento, estado de humor durante el dia. Y viendo las fluctuaciones voy ajustando carga. Los otros parametros imo son especificos de 1 solo sistema.

[Alejandro Maza @alehandromz]

@elbasti https://t.co/4tobc6CNiO

[Alejandro Maza @alehandromz]

@binarybits @CsabaBiro16 Haven't found one great myself, but Deep Research put a pretty good one together from some ok sources https://t.co/PgaQgHMdIS (Second Report)

[Alejandro Maza @alehandromz]

@binarybits @CsabaBiro16 pretty bad analysis imho

[Alejandro Maza @alehandromz]

@AravSrinivas Super impressed! comparing it to Open AI's and results are on the same level so far! just way quicker (and for a 10th of the price)

[Alejandro Maza @alehandromz]

@patrickc I'm loving it for more technical reports or market research with external sources, but haven't gotten it to generate a good one with internal data. What are your usecases?

[Alejandro Maza @alehandromz]

@ryanrz ☝️

[Alejandro Maza @alehandromz]

@cramforce The man from the future by @Ananyo , and The Maniac by Labatut

[Alejandro Maza @alehandromz]

@elbasti @paulg la tarea basti!

[Alejandro Maza @alehandromz]

@rauchg @rasmic @v0 Yo también!! ☝️

[Alejandro Maza @alehandromz]

@benhylak same experience. terrible.