Pre-submission audit for clipping campaigns. Every clip graded against the campaign's own brief by two AI reviewers from different model families. The stricter verdict stands, and reports land in 48 hours.
Innov8iveLabs is an independent software studio. The work behind it: 15+ years of production AI shipped in banking, adtech, consumer packaged goods, and Big 4 consulting, now turned into small, focused software products.
shipped across the web, iOS and the command line
graded against their campaigns' own briefs
tiered and cited, at pennies each
of production AI behind the build: banking, adtech, CPG, Big 4 consulting
Each product ships on the same engine: adversarial review, agent research, owned compute.
Pre-submission audit for clipping campaigns. Every clip graded against the campaign's own brief by two AI reviewers from different model families. The stricter verdict stands, and reports land in 48 hours.
Agent-native prospecting for niche communities. Research agents sweep community sources, tier every lead against a rubric with cited evidence, and draft outreach. A human approves every send.
Two are products you can buy today. The rest is the studio's own tooling, built on the same engine and running the business daily.
Pre-submission audit for clipping campaigns.
Every clip graded against the campaign's own brief by two AI reviewers from different model families. The stricter verdict stands.
valid8.i8l.tech >Clip annotation and labelling.
The labelling app behind valid8's evaluation set: human judgements, one protocol, so accuracy claims rest on measured ground truth.
annot8.i8l.tech >Agent-native prospecting.
Research agents sweep community sources, tier every lead with cited evidence, and draft outreach. A human approves every send.
investig8.i8l.tech >Content studio.
The pipeline that drafts, audits and schedules the studio's own media. Nothing publishes without a human disposition.
gener8.i8l.tech >Cross-project task store.
One board behind every project, readable by the agents doing the work and the human deciding what matters.
STUDIO TOOLINGShared memory across machines.
Sync for claude-mem, so one set of memories follows you across every machine you work on. Hosted, or run it on your own hardware.
mem.i8l.tech >Remote for self-hosted media.
A native iPhone and iPad remote for fourteen self-hosted media services, built for people who run their own servers.
lodestar.i8l.tech >One remote, two devices.
Controls an Apple TV and the television it is plugged into from a single pad, so the second remote stays in the drawer.
clickpad.i8l.tech >The same board, on the phone.
Native client for the task store: what is open, what is blocked, and what shipped, without opening a laptop.
STUDIO TOOLINGOne home screen for the studio.
Launcher for every app above, plus the internal documentation, on iPhone and iPad.
STUDIO TOOLINGThe pipelines under valid8 and investig8 are the same four pieces.
Adversarial review. Two model families grade the same work independently and the stricter reading stands.
Agent research. Parallel agents sweep marketplaces, rankings, and press into tiered lists with evidence attached to every name.
Owned compute. Pipelines run on studio hardware, a Mac and DGX two-node setup, which keeps unit costs fixed and data in-house.
Human sign-off. Audit trails on every verdict and an approval gate on every send. Nothing publishes on its own.
“I spent 15+ years building production AI systems across banking, adtech, consumer packaged goods, and Big 4 consulting. Innov8iveLabs is where that discipline turns into products.”
Jonathan Poeder · jonathanpoeder.com
Pilots, partnerships, or something adjacent you're building.