← TimeCart

How TimeCart matching works

Last updated: 8 October 2026

TimeCart turns one sentence into a short, ranked list of people who can do the job, with the reason for each match stored alongside it. Here is each step.

1. You describe it

One sentence, typed or spoken. “Fix my dripping tap.” “Maths tutor for my daughter on weekends.” “Someone to design a logo for a bakery.”

2. The intent is read

The sentence is read for four things: the skill needed, the place, the timing and the budget. These become a structured request. Nothing is invented. If something essential is missing, TimeCart asks only that one question, not a form.

3. Candidates are retrieved by meaning

The request is compared to provider profiles by meaning rather than by keyword. A provider who wrote “leaks, taps and bathroom fittings” is a candidate for “dripping tap” even though the words differ.

4. Trust filters run first

Trust comes before relevance. Candidates that fail safety or moderation checks never rank, however well they match. Area and availability are checked here too.

5. Candidates are scored

Each remaining candidate is scored on skills, location, freshness and interests. The score is a model output, not a bid. Providers cannot buy placement.

6. A language model re-orders the shortlist

The top candidates are passed to a language model that reads the request and the profiles together and re-orders the shortlist, catching fit that a score alone would miss.

7. Every match is explained

Each match stores why it was made: which skills matched, how far away, timing, budget. A bad match can be traced and corrected, and the engine learns from every fulfilled request.

8. You connect, privately

You see the shortlist with reasons attached and decide who to talk to. Chat and calls happen in-app. Contact details are never exposed, and you stay anonymous until you choose otherwise. The decision is always yours.

What the engine optimises for

Completed requests and satisfied people, not clicks. Marketplace health carries equal weight to relevance, every ranking decision has to be explainable, and fairness constraints are enforced.

Frequently asked questions

Does TimeCart match on keywords?

No. The request is matched on meaning. Candidates are retrieved by semantic similarity, so a provider does not need to use the same words you used.

Can a provider pay to rank higher?

No. Placement is ranked, never sold. Scores come from skills, location, freshness and interests, then a language model re-orders the shortlist.

Why did I get this match?

Every match stores the reasons it was made, so it can be explained, traced and corrected.

Does the provider see my phone number?

No. Contact details are never exposed. Chat and calls happen in-app, and you stay anonymous until you choose otherwise.


TimeCart is launching on iOS. Get early access · How matching works · What is an intent marketplace · TimeCart vs service marketplaces