What an AI-native product studio looks like
How a small studio uses AI to report, build products, and maintain them under human direction.
An independent, AI-native product studio
We run media sites and share the tools and research behind them.
Operating our own sites teaches us how the whole business fits together: reporting, production, distribution, audience relationships, and revenue.
We’re making tools developed for our own use available to other publishers, starting with AI visibility and citation tracking.
We release the underlying components so others can inspect the code, adopt it, and build beyond our original use.
We publish the reports that inform our product and business decisions, including the sources and assumptions readers need to assess them.
How a small studio uses AI to report, build products, and maintain them under human direction.
Start with a person’s decision, then work out what information would help.
Source records, verification, and corrections belong in product development as well as reporting.
AI & journalismPublic beta
What does AI change for journalism?
An agentic publication about AI and journalism. It gives newsroom readers briefings and a research record they can follow as claims develop.

Public data & civic lifePublic beta
What’s happening in your schools?
School profiles, budget explanations, and comparisons built from official New Jersey data. Started by a parent, for others trying to understand their schools.

Hardware & buying decisionsPublic alpha
Buy, wait, or use what you have?
Price tracking and reporting for hardware buyers, starting with Mac desktops. The guides examine what to buy, when to wait, and what older machines can still do.

Publishers & information ecosystemsPublic alpha
Who serves a community’s information needs?
A planned public reference connecting publisher records with ownership, business models, and the communities they serve.
Our betas are open for use and feedback. Alphas are earlier experiments; their scope and features may change.
Developed for our own publishing, now being prepared for other newsrooms.
Test the questions your audience actually asks, see which sources AI systems use, make a real change, and measure what happens next.
About this toolFind mentions and republications of your journalism, trace them to the original story, and see what changed.
About this toolThree components from our research and publishing practice. Each can be used independently.
A reporter’s notebook for people and agents
Keep sources and claims together after an AI session ends. Flip preserves the record needed to check an answer, correct it, or continue the research.
Model routing and telemetry
Record model calls locally and route requests across providers. Somm helps us investigate what a task costs and which models suit it.
Portable reports and web tools
Package a web artifact with the assets it needs to run. Artoo separates the presentation you share from the private research that produced it.
The research we use to decide what to build and how to run the studio.
August 2026
What AI-assisted visualization can do, where it fails, and how practitioners assess the results.
Read the field guide ↗August 2026
Inference costs, infrastructure, and model choices, with dated prices and explicit forecast assumptions.
Read the report ↗I’m Marc Lavallee. My career has been spent building technology for journalism. Lyra Forge is where I’m developing publications and tools with AI.
I direct the studio and remain responsible for what it publishes. The consulting practice at Lyra TK works with organizations on strategy and transformation.
There is room for creative experiments, too. Caz Games preserves playable games and the development record of an agentic studio. Its site still carries a dated pause notice.
For project feedback or a conversation about working together, get in touch.