General-purpose AI and Solon
Solon vs ChatGPT for consulting proposals
This is not a question of which writes better prose. General-purpose models write well. The question is what happens to the work once the conversation ends.
What ChatGPT and Claude do well.
Genuinely well, and worth saying plainly: they draft quickly, they adapt tone, they restructure an argument on request, and they are already in your team's hands. For an individual consultant working on a single document, a general-purpose chat is often enough. Most of the value in a first draft is already available to any firm that wants it.
Where a firm-wide workflow needs more.
The gap opens when the work has to outlive the conversation, be reviewed by someone who was not in it, or reuse something the firm already knows.
| ChatGPT or Claude | Solon | |
|---|---|---|
| Context | Assembled by hand in each conversation | Persists around the firm, client, opportunity and artifact |
| Output | A message in a thread | An artifact with a version and an owner |
| Review | Happens elsewhere, on a copy | Attached to the exact version, with sources and changes |
| Firm knowledge | Pasted in, if someone remembers to | Applied from approved methodology, with the source visible |
| Client boundaries | Managed manually by the person prompting | Set per source, per client and per output |
| After approval | The next document starts empty | Approved context is the basis for the next artifact |
Solon is not a replacement for the model.
Solon works around the AI models your team already values. Its role is to provide the consulting-specific memory, artifact and review layer that a general-purpose chat does not provide on its own. If your firm is happy with how proposals get written today, that is a reasonable answer.