AI Agents' 'Unmapped' Maya Site Was Already in a 2025 Paper

Pietro Miotti, whose post about 48 AI agents finding an "unmapped" Maya site in Mexico reached 846 points on Hacker News, added an update on Nov. 11, 2026: the site had already been mapped in a 2025 paper by Estrada-Belli and colleagues.

In the update to his post, Miotti wrote that "the agents missed it while going through the literature, so they rediscovered it rather than discovered it." The post opens with the line "By analyzing NASA lidar data, my AI agents discovered a Maya site in Mexico." It says he is "not an archaeologist" and does not state an employer. The Hacker News submission was titled "Google SWE points 48 AI agents at NASA Lidar, finds unmapped Maya pyramid in 72h", and collected 312 comments.

What the 48 agents did

Miotti said the agents turned NASA lidar strips covering about 1,200 square kilometers of jungle into images, then reviewed them for monumental features while excluding natural landforms, sinkholes and roads. He said no specialized model or classifier was used. The agents also checked the raw point cloud to confirm the shapes were on the ground and not vegetation. The post does not name the models behind the agents.

The result was a hilltop site with a levelled plaza, an oval pyramid he estimates at "at least 8-12 m tall", a raised courtyard, a flat-bottomed hollow that may be a reservoir, and three platforms. Miotti describes it as a possible Classic-period civic centre, roughly AD 250 to 900, and calls that speculation from a chat with the agents. The reconstruction image is a Gemini illustration generated from the lidar layout.

The October timeline

Date	Event
Oct. 18	Idea forms while Miotti travels in Peru
Oct. 19	Agents produce reports overnight
Oct. 20	Report revised, translated to Spanish and sent to INAH
Oct. 21	INAH replies it has no registered site at the location
Nov. 11	Update: the site was already mapped in 2025

INAH is Mexico's National Institute of Anthropology and History. Its reply, which Miotti quotes, said "no registered archaeological site was identified at the location given in your report." That statement concerns INAH's register, not the published literature, and the post says the finding is "not official yet." A field check has been offered but not scheduled. No archaeologist is named as having verified the interpretation.

Why the lidar record made "unmapped" a risky word

Lidar mapping of Campeche is dense and published. A Tulane University study in the journal Antiquity, announced Oct. 29, 2024, surveyed about 50 square miles and identified more than 6,500 pre-Hispanic structures, including a previously unknown large city with stone pyramids near the area's only highway. Luke Auld-Thomas, the doctoral student who led it, said "we have not found everything, and yes, there's a lot more to be discovered."

Auld-Thomas's team found a new city in a survey of that size, so lidar does turn up sites. The literature search is the step that checks whether a find is new, and by Miotti's own account it is the step the agents missed.

For another agent claim that needed checking against the vendor's own numbers, see the report on Vercel Labs' agent-browser token claims. Related reading on agent tooling is in the piece on agent skills repositories. Miotti's post says the exact coordinates sit in a confidential annex, and INAH's local centre will decide whether to send a team.