Who's writing this
Ellie Harris
I came to this work from two directions at once. I studied criminology and volunteered with victims of crime. What victims want is justice. Justice is hard to get without a clear, verifiable record of what actually happened. I'm an accredited Australian mediator under the National Mediator Accreditation Standard (NMAS). I've also spent more than twenty years in enterprise technology: sales and governance, with a specialty in change and adoption, in both product and service companies, selling into utilities, government, finance, healthcare, education, telcos and mining. I still do. Those are the institutions where, when something goes wrong, somebody has to be able to explain it.
I'm also a curious, self-taught full-stack developer. I've built a number of free tools, including heybigsister.com. Headlights is that record, built for AI agents. Free, open-source, written so the people who need it can read it themselves.
The trigger was the pattern. Every week another story: Air Canada's chatbot, AI lawyers citing fake cases, a coding agent that wiped a production database, support bots talking in the voice of the previous bot. Different industries, same failure: nobody could produce a clean record of what the AI had actually done. I've spent two decades inside organisations where we don't know what happened is not an acceptable answer, and I've spent years alongside people whose job is to make it answerable after the fact: investigators, regulators, mediators, advocates. So I started writing the field notes, and the code.
Outside work, I read philosophy, follow quantum physics, and have a long-standing interest in Taoism. Different fields, same question underneath: what are the rules behind the rules?
The opposite of a correct statement is a false statement. But the opposite of a profound truth may well be another profound truth.
Niels Bohr
Bohr was a co-founder of quantum mechanics. He spent much of his life arguing that the world rarely splits cleanly into right and wrong. Two things can both be true at once, in tension with each other, and getting to the deeper answer means holding both. That's what an audit trail is for. When an AI agent fails, the explanation almost never reduces to a single cause. The training data was stale, and the policy had just changed, and the customer's question was ambiguous, and there was no verification step in the pipeline. All of those can be true. A record that captures them all is the one that supports the conversation that actually needs to happen. Anything that collapses to a single story is doing someone's PR work.
Headlights is independent on purpose. Most of the companies talking about AI governance are tied to a vendor. They're built by the same people shipping the agents, or they're a feature inside the platform you're trying to govern. A company grading its own homework isn't an audit. I wanted something that doesn't sit inside any of the platforms it's watching.
Day job: I work as a Technology Account Director. Headlights is independent of my employer, separately funded, and free. No upsell, no pricing page, no waitlist. Just the field notes and the code.
