Sonnet Task Profile: General Professional Work (CCAO-F)

The sensible default in the middle

Every selection framework needs a default -- the choice you make when nothing about the task pushes you to an extreme. In the tier spectrum, that default is Sonnet, the balanced middle. The CCAO-F exam tests this at the understand level because recognising Sonnet as the sensible starting point, rather than reaching for an end-tier by reflex, is what keeps most model choices correct with the least effort.

Sonnet's role is to handle the broad middle of professional work: drafting, synthesis, and analysis that is demanding but not extreme. Most knowledge work lives here. By balancing quality against speed better than either end-tier for typical tasks, Sonnet lets you cover the bulk of requests well without stopping to deliberate. That role makes it the anchor of the unified decision logic.

What "general professional work" means

The tasks Sonnet is built for are the everyday substance of knowledge work: writing a draft, pulling several inputs into a coherent synthesis, analysing a situation and drawing conclusions. What these have in common is that they demand real capability but not extreme reasoning under ambiguity. They are hard enough that a purely structured, fast tier would fall short on nuance, yet not so ambiguous or high-stakes that they need the deepest reasoning available.

That middle band is wide. Most professional requests -- reports, memos, analyses, plans, comparisons -- land in it. Sonnet handles them well because its balance of quality and speed is tuned for exactly this kind of work. The point to internalise is that "not the top tier" does not mean "not analytical." Sonnet handles analysis competently; it is only the genuinely complex, ambiguous, high-stakes analysis that reaches past it toward Opus.

Why starting with Sonnet is the smart move

Beginning from Sonnet is not a compromise; it is a strategy that avoids two symmetric errors at once. Reach for the fastest tier by default and you under-provision capability, risking weak output on work that needed more nuance. Reach for the top tier by default and you over-provision, spending speed and budget on capability the task never used. Sonnet sits where neither error occurs for the typical case, so starting there and only stepping off when the task's profile clearly demands it is the efficient path.

This is why the exam frames Sonnet as the default. The discipline is to justify moving away from it, in either direction, rather than justifying using it. Drop to Haiku only when the task is genuinely structured and speed or volume dominates; rise to Opus only when complexity, stakes, or ambiguity genuinely warrant it. For everything in between -- which is most work -- Sonnet is the answer, and defaulting to it keeps the bulk of decisions right without over-thinking.

What the CCAO-F exam trips candidates on

Two errors are tested. The first is defaulting to Opus for routine drafting tasks that Sonnet already handles well, over-spending capability. The scenario describes ordinary professional work with no extreme ambiguity or stakes; reaching for the top tier there wastes speed and budget for no better outcome. The credited answer keeps routine work on Sonnet.

The second is assuming Sonnet cannot handle any analytical task because it is not the top tier. Sonnet handles analysis without extreme ambiguity perfectly well. A question may imply that "analysis" automatically means Opus -- but only genuinely complex, high-stakes, ambiguous analysis reaches that far. Most analysis is Sonnet work.

Common misreadings to avoid

How this shows up on the exam

Questions describe ordinary professional tasks and ask for the tier, often with an Opus distractor dressed up as the safe choice. The reliable answer defaults these to Sonnet and reserves Opus for genuine complexity or stakes. Treat Sonnet as the starting point and require a specific reason -- extreme structure and volume, or extreme ambiguity and stakes -- to move off it.

This knowledge point builds on the tier spectrum and sits between the Haiku and Opus profiles, together feeding the unified decision logic.