Synthetic users that do something.
Most synthetic users answer interview questions. Ours try to get something done. They walk your real docs step by step, and when they can't go on, they show you the paragraph that stopped them.
Opinions are cheap. Attempts are evidence.
An AI persona saying “I'd find this confusing” is a guess you can't check. A persona that stalls at step four because section 3.2 doesn't cover their situation is a finding you can open, read and fix.
How it works.
- 01Bring your docs.Subjourney builds the journey they describe, and the personas who'd walk it.
- 02Run them.Each persona pursues a goal, node by node, using only what your docs give them.
- 03Read the evidence.Every stall is pinned to a step and a paragraph.
Interview-style vs journey-style.
| Interview-style synthetic users | Subjourney personas | |
|---|---|---|
| What they do | Answer questions | Try to reach an outcome |
| Grounded in | A description of who they are | Your docs, one step at a time |
| Output | Quotes and opinions | Stalls traced to paragraphs |
| How you check it | You can't, really | Open the paragraph and read it |
What they do
- Interview-style synthetic users
- Answer questions
- Subjourney personas
- Try to reach an outcome
Grounded in
- Interview-style synthetic users
- A description of who they are
- Subjourney personas
- Your docs, one step at a time
Output
- Interview-style synthetic users
- Quotes and opinions
- Subjourney personas
- Stalls traced to paragraphs
How you check it
- Interview-style synthetic users
- You can't, really
- Subjourney personas
- Open the paragraph and read it
“Aren't synthetic users unreliable?”
For opinions, often. That's why Subjourney doesn't ask what personas think. It shows what your docs let them do, and every finding points to text you can verify yourself.