Aidan N. Gomez
Co-author of the 2017 paper and a co-founder of Cohere, which dates its founding to 2019.
Read Aidan N. Gomez’s profile →
About Cohere ↗Company history · checked 15 September 2026Cato & Veil / AI corporate intelligence
AI businesses.
The people behind them.
Investigating team movements, information exposure and competing interests in the AI sector. We connect research and professional histories with the evidence needed to understand a commercial concern.
Explore the relationship study ↓AI corporate intelligence
The people, interests and access behind the business.
A competitor is recruiting a connected group of engineers. Confidential work appears outside the company. A supplier or research partner has interests that were not disclosed. We investigate the relationships and events behind these concerns.
Discuss an investigation →AI talent intelligence →Recruitment & team movements
Establish who moved, their responsibilities and the sequence of departures. Examine whether a pattern affects the company’s capability or warrants further enquiry.
Information loss & insider concerns
Examine relevant access, communications and records to test an allegation of information loss. Distinguish legitimate professional contact from evidence of unauthorised disclosure.
Competing interests & partnerships
Investigate relevant advisory, commercial, research and ownership relationships around a person, partner or supplier.
Supply chains & geopolitical exposure
Identify counterparties, dependencies and cross-border interests relevant to a transaction, relationship or emerging concern.
Applying the method / A client scenario
Several engineers leave.
What does the company need to establish?
A cluster of departures can raise different questions. The investigation follows the evidence, including evidence that the moves were ordinary recruitment.
01 / Establish the sequence
People and responsibilities.
Confirm dates, roles and the work each person was responsible for. Map documented professional connections and the organisations involved.
02 / Test the concern
Access and records.
Where authorised, examine relevant access records, communications and document history. A relationship identifies an enquiry; it does not prove a leak.
03 / Explain the findings
A decision the client can act on.
Report what is established, what remains unresolved and whether further technical, legal or protective work is justified.
This scenario explains the investigative method. It is separate from the researchers shown in the historical graph.
Research connections are loading. The full account is available alongside.
01 / The question
One published collaboration. Three later companies.
In 2017, eight researchers co-authored Attention Is All You Need, the paper that introduced the Transformer architecture. Four of those authors later co-founded the three companies shown here. Follow those documented relationships.
One research collaboration. Several subsequent company-founding links.
Evidence behind this chapter ↗02 / 12 June 2017
Eight names. One shared contribution.
Attention Is All You Need introduced the Transformer architecture. Its author list gives us a precise starting point: eight people connected by a published piece of work.
The lines here mean co-authorship—not ownership, influence scores or a private alliance.
Evidence behind this chapter ↗03 / 2019 → 2023
The company boundaries change.
Aidan Gomez co-founded Cohere. Llion Jones co-founded Sakana AI. Ashish Vaswani and Niki Parmar founded Essential AI. These are selected founding relationships supported by company sources.
Four of the paper’s authors connect to three later companies in this view. It is not a complete career map.
Evidence behind this chapter ↗04 / Historical relationships
A founding link is not a current role.
A professional network has more than one clock. A paper remains co-authored after someone changes jobs; a founder can leave the company they established. We keep the relationship type and the event separate.
Explore a person below to read the exact relationship and its source.
Evidence behind this chapter ↗05 / The intelligence question
What does this tell a client?
These companies share part of their founders’ research history. For an investor, partner or hiring team, that identifies people and experience worth examining. It does not establish that today’s teams share technology, work together or perform equally well.
The map establishes connections. An assessment of dependence, access or intent requires further evidence.
Evidence behind this chapter ↗The evidence index
Choose a person.
Examine the connection.
Eight paper authors. Selected founding links. Every relationship shown here has an accessible source.
The Cato Brief / Research notes
Read the people.
Understand the business.
Network analysis, incentives and research assessment. Three ways to examine the questions beneath a company’s public account.
01 / Network analysis
The company boundary is only the beginning.
A published research collaboration can reveal connections that an organisation chart leaves out. The meaning of those connections needs careful handling.
Read the analysis ↗02 / People & incentivesA large hiring package is a starting point for questions.
Compensation can help explain organisational pressure. It cannot, by itself, tell you why someone moved or whether a team will remain intact.
Read the analysis ↗03 / Research assessmentLook at the contribution before the ranking.
When assessing emerging AI expertise, the most useful question is what someone has demonstrably contributed—and how that work matters to the decision in front of you.
Read the analysis ↗From research to an instruction
Know what you need to establish.
Tell us the company, relationship or decision you need to understand. We agree the questions, available sources, gaps, reporting and fees before starting.
Commission a private briefing ↗Receive the Cato Brief ↗Before committing capital
Examine the people behind a proposed investment, relevant professional histories and the dependencies that deserve closer enquiry.
Before entering a partnership
Understand the organisations and interests involved, and identify what the public record does not resolve.
When circumstances change
Scope monitoring around named organisations and material developments, with a clear distinction between an event and its possible significance.
Method & limits
Every connection has a meaning.
This selected historical study combines a published author list with company-founding accounts. Positions in the visual are an arrangement for reading, not a geographic map or an influence score. A professional connection is not evidence of misconduct. The founding links do not assert current employment, ownership or control.
The wider ATIU research corpus—market assessment, movement, compensation, emerging talent, rankings and collaboration data—informed the questions. Only independently source-checked relationships from that work are included here. This is a dated research study, not a live monitoring feed.
Attention Is All You Need ↗Vaswani et al. · arXiv · 12 June 2017The eight named authors and the Transformer architecture.About Cohere ↗Cohere · Company history · checked 15 September 2026Cohere was founded in 2019; Aidan Gomez is a co-founder.Introducing Essential AI ↗Essential AI · company announcement · Founding announcement · checked 15 September 2026Ashish Vaswani and Niki Parmar founded Essential AI.Corporate information ↗Sakana AI · Company history · checked 15 September 2026Sakana AI was founded in 2023 by David Ha, Ren Ito and Llion Jones.