FIDENIT
OPEN SOURCE · Planned

Agent Harness

A frame for AI models to work as a team

PROBLEM

Could several AI models work together like one team?

For several AI models to work like one team, we think they need a frame that splits the work, checks each other's results and keeps a record. We want to build that frame as open source and share it with people facing the same problem. As with KITE, we would like to keep it open and refine it together.

HOW IT WORKS
01 · INPUT

Connect

A task brief

02 · ENGINE

Refine · Calculate

We aim to split the brief across several models and have them check each other's results.

03 · OUTPUT

Result

Verified results and a work record

GOALS
01

Splitting the work

02

Cross-checking results

03

Keeping a work log

04

Setting cost limits

SPEC
FieldOpen source
StatusPlanned
InputA task brief
OutputVerified results · work record
DeliveryGitHub open source (goal)
MORE IN OPEN SOURCE

Let's talk about
Agent Harness

business@fidenit.com