Open for new projects

A small remote team building AI agents, software and products, for our own ventures and for businesses like yours.

Products live
7
products live on the stores and the web
Research
ICANN
research published in 2026, Springer LNCS
Competitions
1st
Data & Visualization, AutoScientist Challenge 2026
Response time
2 days
to a reply from a person, not a bot

Products we build and run.

11 ventures in the register, 7 live on the stores or the web. Each one carries its real status.

Every release, dated

Published research

−41.8%

Median end-to-end latency in production voice AI.

A production voice pipeline, measured at every step: the biggest delay was one step waiting for another. Running them together took the median from 3,277 ms to 1,909 ms.

Rodrigues, C. and Vas, O. ICANN 2026, Springer LNCS, Part V, pp. 450-461.

Also: first place, Data & Visualization, Adaption AutoScientist Challenge, August 2026.

For businesses: a specific problem, solved and kept running.

  1. 01

    Assess

    Two weeks

    The problem, the data, what is feasible and what it would cost.

  2. 02

    Prototype

    Weeks

    A working proof on your data, with the measurements that decide the next step.

  3. 03

    Build

    Fixed phases

    The production system, against acceptance criteria written before the first commit.

  4. 04

    Run

    Monthly

    Monitoring, evaluation, maintenance and the next features.

What every engagement defines

Carson Rodrigues

Founder & CEO, WYTOR

Engineer and researcher. Published at ICANN 2026.

Founder-led, and small on purpose.

AI agents do real engineering work here; a person reviews and merges. Money, mail and anything published wait for a person.