01 / COLLECT
Bring the evidence together
Planned inputs include screenshots, pasted log excerpts and written observations, with source references kept alongside the draft.
Idea / validation stage · Not yet launched
Less guesswork.
More reproducible bugs.
Amnezi is a product concept for software teams: turn screenshots, logs and bug notes into evidence-linked reproduction steps and regression test drafts that a human can review.
Discuss a QA workflow →Seeking discovery conversations with web and mobile teams.
Illustrative example · Not a live product
From fragments to a useful starting point.
Screenshot: payment errorLog: request timeoutNote: after retry
Proposed output
- Open checkout and select a saved payment method.
- Submit payment, then retry after the timeout.
- Check whether the error persists and capture the request ID.
Missing context: Which browser and build were used? Was the first payment accepted?
Synthetic scenario. Steps are hypotheses to verify; evidence alone may not establish the cause.
The problem
A screenshot rarely tells
the whole story.
Bug evidence often arrives across chats, screenshots and partial logs. We want to help QA engineers and developers assemble that context, see what is missing and prepare a report worth investigating.
02 / REASON
Make uncertainty visible
Generate candidate reproduction steps, flag unsupported assumptions and ask targeted questions when the evidence is incomplete.
03 / REVIEW
Prepare a repeatable check
Draft regression test cases for human review. The intended workflow keeps approval and execution with the team.
A focused starting point
Build with real workflows.
Our first goal is to validate whether evidence-linked drafting can reduce the time spent clarifying bug reports.
- Interview QA engineers and developers about their current process.
- Build a small prototype around text and screenshot inputs.
- Evaluate draft quality against manually prepared reports.
Where the project stands
Amnezi is a proposed venture at the idea and validation stage, with no registered company. There is no launched product, customer base or measured performance to report.
Planned AI approach
We intend to use the Claude API for multimodal evidence analysis and structured drafting. Sensitive log redaction, data handling controls and human review are planned design requirements, not implemented features.
Help shape Amnezi
What makes a bug report useful?
If you work on QA or software delivery, we would value a conversation about your workflow. Please avoid sending sensitive logs or customer data.