Reports

One report per question you would actually ask, kept current. Every pick comes from runs of the same published test, and every update names the run behind it.

Updated October 13, 2026AI Assistant Consumer Travel Report: October 2026We tested real bookings across five AI assistants to see which handle flights, changes, seat upgrades, and cancellations best.

Latest writing

All writing

Analysis and opinion, dated and signed. Reports carry the verdicts; this is where we argue about them.

Experiment, September 30, 2026Instinct vs. Muse vs. Grok Bot: what do “elegant” and “cost-efficient” mean?I gave three assistants the same LA weekend prompt. Muse chased the lowest price, Grok Bot reduced hassle, and Instinct protected time. Then I let them learn from one another.By David Pawlan. 4 minute read.

Analysis, September 28, 2026I analyzed and compared the privacy policies of the three assistants people actually use. Here is what they say.Instinct and Muse allow training on eligible interactions by default. Grok Bot says it does not train on customer data while Privacy Mode is on. Their documents also differ on access, deletion and responsibility for agent actions.By David Pawlan. 9 minute read.

Analysis, September 25, 2026The AI assistant race just got hot. Here's where it stands.Meta shipped Muse, Instinct raised again, and OpenAI is hinting. After 250 tests across 26 assistants, an overview of where things sit and what to watch next.By David Pawlan. 10 minute read.