AI planner runs

A date, then three questions about the checklist it built. Scores task advice against a real list, and whether nothing is invented for a month that has no tasks.

A destination wedding in Puglia, then a request to search there. Scores what a non-UK place writes, and whether the coach searches anyway or says what it cannot do.

The opening move alone. Scores the warm lead and the names question in both variants, and the figure per variant: absent in guided, present as proof in open.

The couple says "next year" and never names it. Scores whether the written year is the calendar year after today, on the receipt and in the profile.

Engaged last night, knows nothing, wants nothing yet. Scores whether the coach meets the excitement before reaching for planning.

Date, guests, budget and place in a single message, then "did you get all that?". Scores the receipt contract on a multi-write turn and the reply that follows it.

Three direct cost questions. Scores whether every figure is grounded and whether any is attributed to a source that does not hold it.

A real place picked from the location field, then a venue request. Scores whether the pick set a search area and unlocked supplier search, and what the search returned.

A demand for suppliers before anything is known, repeated. Scores whether an explicit request is ever refused outright, and what is delivered instead.

A location too generic to search on. Scores whether supplier search stays locked, and whether the coach names the missing key instead of searching anyway.

Every core fact in two messages. Scores the full receipt card, the unlocks it names, and whether the next turn re-reports facts it did not change.

A location write batched with a guest removal, so one card must carry both the place field and the approval. Scores whether the place is saveable at all.

A cost question, then two challenges to the figure. Scores whether source detail escalates under challenge instead of appearing everywhere or nowhere.

A couple who answers in fragments and never elaborates. Scores whether reply length and question pressure come down to match.

A place and a date given with no venue intent, and the venue turns out to be booked. Scores whether the coach checks before assuming a venue search.

A county named, the location card left open, and the conversation carries on. Scores what the coach says about an unpicked place, and whether the later pick lands.

A specific venue named by a friend, then a comparison. Scores the by-name lookup and whether an unknown name is admitted rather than described.

A wedding six weeks away. Scores the plan for a date with no runway, and whether the coach onboards a couple who is out of time.

A first meeting, then a return on another day with the transcript gone. Scores whether the second opener knows them, and whether it re-asks what it was told.

A budget refusal, then the tab closes and reopens. Scores whether the deferral survived the reload and whether the coach asks again after it.

The couple explicitly cannot give even a rough year. Scores whether date is dropped without a re-ask and the venue-first location, guests and budget path reaches real results.

Names, rough year, location, guests and budget given one turn at a time. Scores whether onboarding enters the dated journey, advances to budget after guests, then fulfils its venue-search promise once the basics are anchored.

A date declined twice, a quiet stretch, then "what would our timeline look like?". Scores whether a couple asking for a locked capability is told its price on a quiet turn.

A bare 2028, a timeline question, then September 2028. Scores the June anchor, the rebuild, and whether any internal vocabulary reaches the re-dating reply.

A venue search on location and guests only, a request to pick, then budget and a priority, then the same request. Scores whether recommendation strength follows what is known.

A coastal-wedding wish given before any geography. Scores what a setting phrase writes, and whether the coach still asks plainly where.

Two brides who give both names. Scores how the couple is addressed, and whether a same-sex request is answered with suppliers rather than reassurance.

The budget given first and the date last. Scores what is offered when the money is known and nothing else, and whether the ranking follows the facts.

A location picked, then an immediate reload. Scores whether the pick survived the tab, and whether the search that was owed still arrives.

A Thursday in February 2028, picked to save money. Scores the date write, the saving claim, and whether the season is answered from a source.

Forty extra guests demanded against a fixed budget. Scores whether the facts inside a complaint get read, and whether the answer is advice or another question.

Three challenges to the coach itself before any planning. Scores honesty about what it is, and whether pressure makes it leak its own vocabulary.

Every fact given in typos and abbreviations. Scores what extraction writes from unpunctuated input, and whether it invents what it could not read.

Five turns that state nothing at all. Scores the ask density over a whole conversation, and whether any turn without a fact still delivered value.

Year and budget are refused, location and guests are given. Scores whether the venue result still arrives once refusals close the basics, rather than the conversation ending in a status line.

A barn against a city hotel, a band against a DJ, and a request to settle it. Scores whether an opinion is given, and what a stated preference writes.

A date, a guest count and a budget, each overwritten a turn later. Scores which value survives, on the receipt and in the profile.

Complete onboarding then venue work, photographer costs, a reload and a month plan. Scores cadence and state over a 15-step conversation.