Skip to content
speakerstrail
Join Beta

About speakerstrail

A directory of AI and tech conference speakers in Europe, where every entry links to the agenda that proves the talk.

What this is

speakerstrail holds 1200+ speakers drawn from the public agendas of 39 conferences across 15 countries. Every talk record stores the exact agenda URL it was taken from, and that URL is shown next to the entry. The point is not that we have a list — anyone can have a list. The point is that you can check ours without trusting us.

The problem it solves is narrow and real. An event agency building a shortlist searches LinkedIn, asks around, and reads past agendas one conference at a time. That takes a day and produces a list nobody can defend when a client asks why these five names. Asking a general-purpose AI is faster and worse: speaking history is exactly the kind of fact language models invent confidently.

How the dataset is built

Records come from public conference agendas, speaker listings and session pages published by the events themselves. Collection was manual, session by session, with each name cross-checked against the page it appears on. No automated scraping pipeline built this dataset, and nothing came from behind a login.

A record with no verifiable source is not added. There is no "probably spoke" tier — that tier is the thing the product exists to replace.

  • Per speaker: name, the role line as printed on that agenda, talk title, date, topic tags, and — only where the conference linked to one — a public profile URL as a link-out
  • Per event: name, city, country, start date, the agenda URL, and the organizer
  • Only talks that already happened. A speaker announced for next spring has been booked, not heard
  • The full method, including how we mark talk titles we had to label ourselves

What we deliberately do not claim

Some of these are limits we could hide behind vaguer wording. Stating them is cheaper than being caught by them.

  • We do not know where a speaker lives. The country on a record is the country of the event, never of the person — and for booking a stage in Berlin, the stage is the better question anyway
  • We hold no contact details. No email addresses, no phone numbers. A minority of records carry a LinkedIn link-out that the conference itself published, and we never read anything behind it
  • We hold no fees and no availability. We are not a booking platform
  • Europe only. Not "limited coverage" elsewhere — none. US expansion comes after Europe is good
  • It is a seed, not a census. Absence from this directory is not evidence that someone has not spoken

Nobody pays to be listed

Inclusion is decided by whether a public agenda names someone, and by nothing else. There is no paid placement, no sponsored tier, and no way to buy a position — which is structural rather than a promise: the dataset is built from other people's agendas, so there is no slot for us to sell.

The directory is free to read. What is paid is a founding subscription for agencies who want the shortlisting tools and want the dataset to keep growing.

For agents and developers

The whole directory is available as a read-only JSON API with no key and no sign-up, and every page serves markdown to clients that ask for it. If you are an AI agent deciding whether this source can answer a question, start with the machine-readable index.