# How Tender Navigator works

Source: Tender Navigator Knowledge Base, https://tendernavigator.com/knowledge-base/how-tender-navigator-works (canonical HTML). Category: Getting started. Published 2026-09-08, last updated 2026-09-08.

> Tender Navigator works in a loop: you describe what you are looking for as a filter, Tender Navigator checks new tenders against that filter every day, an AI model evaluates each matching tender against your prompt, and the verdicts reach you on the Results page and by email.

## The five steps

Tender Navigator turns tender monitoring into a repeatable workflow. Each step below has its own article.

| Step | What happens | Where |
| --- | --- | --- |
| 1. Search | You run a live search across the tender database by keywords, countries and classification codes. Nothing is evaluated yet. | [Pre-search tool](/knowledge-base/pre-search-tool) |
| 2. Save a filter | The search becomes a filter: a saved search with an AI prompt describing your business and an email schedule. | [Save as Filter](/knowledge-base/save-search-as-filter), [Filters](/knowledge-base/how-filters-work) |
| 3. Evaluate | Every day, new tenders that match an active filter's criteria are judged by an AI model against the filter's prompt. | [AI evaluation](/knowledge-base/how-ai-tender-evaluation-works) |
| 4. Read results | Each evaluated tender appears on the Results page as Matched or Unmatched, with reasoning, and in the filter's digest email. | [Results](/knowledge-base/understanding-evaluation-results), [Digest emails](/knowledge-base/email-digests) |
| 5. Act | You open the original notice, save tenders to lists, export them, or share a tender with a colleague. | [Tender Lists](/knowledge-base/tender-lists), [Sharing](/knowledge-base/sharing-a-tender) |

## Two kinds of search, one database

The **Pre-search tool** searches the tender database directly and returns live notices. It is free, unlimited and available without an account (with a five-result preview for visitors). Every row in its results is marked **Not evaluated**, because nothing has been judged against your business yet.

A **filter** is different: it is stored on your account, it runs on its own every day, and each tender it picks up is evaluated by AI. A filter is the only way to get Matched and Unmatched verdicts. The Pre-search tool exists to help you build a good filter quickly: when a search returns the kind of tenders you want, one click saves it as a filter with the same keywords, countries and codes.

## What the AI reads

AI evaluation compares each tender against the filter's **AI prompt**: a free-text description of what makes a tender a good fit for your company, what rules a tender out, and where you actually win. The verdict and its reasoning are only as good as that description, which is why [Writing the AI prompt](/knowledge-base/writing-the-ai-prompt) is the most important article in this Knowledge Base for a new user.

Tender Navigator also keeps a private [Business Profile](/knowledge-base/business-profile) per account. A completed profile can generate a draft filter, prompt included, so the first filter does not start from a blank page.

## What runs on a schedule

- **Filtering and AI evaluation run daily** for every active filter on every plan. This is confirmed by the Tender Navigator backend team.
- **Digest emails** go out on the schedule you set per filter: daily or weekly, on a chosen day and time. The schedule only controls the email, not the evaluation.
- A **paused filter** does nothing at all. A filter created from a search or from the Business Profile starts paused, so that you can review its prompt before it consumes evaluations.

## What limits apply

Each plan allows a number of active filters and a number of AI evaluations per billing cycle. When a filter has used its share, further tenders are recorded as stuck rather than evaluated. See [Plans and what they include](/knowledge-base/plans-and-limits) and [Filter usage and limits](/knowledge-base/filter-usage-and-limits).