They bought the decisions behind the data.
Google paid $10 million for the operational data of a defunct airline. Not the passenger records. The record of how a complex business made decisions under real constraints, and what that says about the agents most companies are deploying.
Deterministic AI for enterprise governance, on the AI InterConnect podcast.
A 34-minute conversation with Ryan Hanley on why regulated operations need answers they can re-trace, where vector search runs out, and what a rules-driven graph found in Alberta's public records.
AI InterConnect, YouTube
Second place at Agency 2026, Alberta's national AI hackathon.
In 48 hours a five-person team turned 14 public datasets into a 38-million-node graph, then answered the hardest question in the brief with no language model in the query path.
Ottawa, hosted by Alberta's Ministry of Technology and Innovation
Model comparison: two extraction systems, four things measured.
An internal benchmark of two Taskd email-extraction pipelines on synthetic, production-inspired fixtures. The cheaper model was not the better operational choice.
Taskd joins the Mila Ventures Founder in Residence programme.
Taskd has been accepted into Mila's residency for founders building AI-native companies, with a base at Mila's Montreal headquarters and access to its research community.
Mila, MontrealTaskd featured in the Linux Foundation's report on open source AI in Canada.
The Value of Open Source AI for the Canadian Economy, commissioned by Meta and published in February 2026, cites Taskd's use of Llama to build governed AI agents from fragmented operational data.
Linux Foundation Research, commissioned by Meta
Work with Taskd
Bring us one critical workflow.
See how Taskd turns the fragmented sources behind a recurring workflow into current context, governed decisions and reviewable evidence.
Book a demo