> build in public

The Lab

I track everything I build in 2-week sprints and share the results.
Wins and failures, with real numbers.

See experiments Explore GTM OS

Since January 2026

0
Experiments
0
Winners
0
Failed
0
Running
stratega-lab / active

Currently running

Active experiments in this sprint.

TrendJourney validation
256 prospects
ICE 8
CI pretotype (fake door)
Kill criteria set
ICE 7.3
Stratega Site deploy
Before 5 Mar
ICE 7
Fractional HoG campaign
804 contacts
ICE 6.5
stratega-lab / highlights

Best experiments

The ones worth talking about — including the ones that didn't work.

WIN Duomo
Broad targeting vs LAL
Switched from Lookalike to Broad+Interest. CPL went from €118 to €7.30 in 10 months.
-94% CPL
WIN Chatty / Stratega
CI multi-agent system
Built a multi-agent CI pipeline: briefing, research, analysis, battle cards. 6 competitors deep-dived.
$236M mapped
WIN Academy
Academy DM campaign
Direct messages to founders for beta access. 20 out of 23 said YES.
87% reply
FAIL Stratega OS
13GB model on 16GB RAM
Pulled gpt-oss:20b (13GB) on Mac Mini with 16GB RAM. Forced reboot. New rule: max 8GB.
System crash
WIN Stratega OS
Claude Code from iPhone
iPhone → Termius → Tailscale SSH → tmux → claude. Full dev environment from my phone.
Zero new deps
FAIL TrendJourney
Delegated agent rebuild
Let an autonomous agent rebuild my fashion researcher. It delivered 4 out of 8 requirements.
40% delivery
stratega-lab / portfolio

Projects I've built

Real projects, real metrics. Not case studies — operating systems.

Duomo Design
adslead gen
active
Full-funnel Meta Ads management. 103 qualified leads/month at €7.30 CPL. Best performing month in company history.
-94% CPL
TrendJourney
ai productfounder
building
Fashion trend intelligence. 5 AI agents research trends, extract Pantone palettes, and generate collection-ready reports.
5 agents
PickEat
csoai ops
active
Sports F&B startup. Built 10 AI agents, managed dev team, shipped 3 releases, closed 3 venue contracts.
10 agents built
Stratega OS
ai infraagents
active
15+ agents, persistent memory, MCP integrations. Multi-model orchestration running daily.
15+ agents
stratega-lab / method

How I run experiments

Every experiment follows the same framework. No exceptions.

01

Hypothesis

Every experiment starts with "If I do X, then Y will happen." Clear input, measurable output.

02

ICE Score

Impact, Confidence, Ease — scored 1-10. Only experiments above 7.0 make the sprint.

03

2-week sprint

Ship fast, measure, decide. No experiment runs longer than 2 weeks without a verdict.

04

WIN or KILL

Every experiment ends with a clear verdict. Winners scale. Losers get documented and killed.

stratega-lab / connect

Want to run experiments
like this?

30-minute call. I'll share the frameworks and tell you if I can help build your growth system.

Book a free call Explore modules

or email matteo@stratega.co