Who holds the job: Claude Opus 5.5 against hedwigai ask on 95 questions
We asked 95 questions about the people at US-listed companies: who the chief executive, finance chief or general counsel is, and how to email the chief executive. Some named a company that does not exist, or gave one company's name with another's ticker. Claude Opus 5.5 answered from what it knows, with no search. hedwigai ask looked each one up. Every answer was checked against the insider filings the SEC holds for each company.
| hedwigai ask | Opus 5.5 | |
|---|---|---|
| Right | 59 | 63 |
| Wrong | 11 | 4 |
| Unsure | 25 | 28 |
| Right, of those answered | 84% | 94% |
| Emails found | 5 of 15 | 0 of 15 |
| Made-up or mismatched company caught | 20 of 20 | 20 of 20 |
| Time, median | 8.2 s | not timed |
| Shows the pages it read | yes | no |
Claude Code
Claude Desktop
Cursor'%3e%3c/path%3e%3cdefs%3e%3clinearGradient%20gradientUnits='userSpaceOnUse'%20id='lobe-icons-codex-_R_0_'%20x1='12'%20x2='12'%20y1='3'%20y2='21'%3e%3cstop%20stop-color='%23B1A7FF'%3e%3c/stop%3e%3cstop%20offset='.5'%20stop-color='%237A9DFF'%3e%3c/stop%3e%3cstop%20offset='1'%20stop-color='%233941FF'%3e%3c/stop%3e%3c/linearGradient%3e%3c/defs%3e%3c/svg%3e)
Codex
Any MCP clientWhere each one wins
Opus 5.5 knows who runs a well-known company, and it got 43 of 60 roles right to ask's 34. When it misses, it names someone else from the same company. The smaller the company, the less it knows: at small companies ask got 13 of 20 and Opus 5.5 9. Opus 5.5 could not give a single email address; ask found 5 of 15 at the company's own domain. Both caught every made-up company and every mismatched ticker.
ask is wrong more often, 11 times to 4, when the pages it reads name someone else first. Every ask answer lists the pages it read, so a wrong name can be checked in one click; a model answering from memory gives nothing to check.
Fine print
- The right answer is the newest officer, in the last year, whose insider filing (SEC Form 3 or 4) gives that title; co-holders all count. A title for one business unit ("CEO, Wind") does not.
- A name counts as right when it gives two words of the filed name, a usual short form allowed ("Greg" for Gregory). A surname alone does not.
- An email counts as right when it is at the company's own domain and names the holder. Addresses are what a contact search returns; they were not sent to.
- Companies were drawn at random from the SEC's list of listed companies, by size, and the made-up companies were invented for this test. 95 questions, measured on 8 October 2026.
- Opus 5.5 was told to answer with a name, an email, "no such company" or "unsure", and answered every question in one reply with no tools, so its time is not comparable and is not shown. ask's times are medians of the round trip over the internet.
- General counsel and chief legal officer are treated as one role.