BrisEvidence status: no head-to-head scores
This board does not declare an overall winner. Bris has not published a matched cross-product benchmark here. The table describes selected documented workflows, while the research notes explain how fair future tests could be run. A feature being documented is not a measured success rate.
Documented workflow comparison
| Product | Documented focus | Evidence type | Benchmark score |
|---|---|---|---|
| Bris | Personal tasks through WhatsApp and Telegram, saved preferences, supported connected work and accepted friend coordination | Bris product descriptions; published by Bris | Not measured here |
| ChatGPT Work | Research, files, tools, personal context and longer tasks within the supported Work experience | Official OpenAI documentation | Not measured here |
| Claude | Personal context, projects and connected tools within supported Claude experiences | Official Claude documentation | Not measured here |
What a useful leaderboard should test
- Memory: useful recall, corrections, forgetting and irrelevant-context restraint.
- Long tasks: completed outputs, preserved constraints and honest partial results.
- Knowledge work: source support, calculations, artifact completeness and reviewability.
- Proactivity: relevant new findings, duplicate-alert restraint and stated coverage gaps.
- Coordination: acceptance, useful shared plans and minimal information sharing.
How a future result should be published
Use the same fictional source pack, clear goals and comparable setup. Record product versions, account access, dates, sample sizes and failures. Show separate dimensions instead of hiding tradeoffs inside an unexplained overall score. Publish enough evidence for a reader to challenge the result.