Skip to content

Find companies that are hiring for the work you sell, every day

Starts twice a day on a schedule. Ends with scored leads in a table and a daily digest of the newest top leads. Nobody is contacted.

AI in one step only: classifying new postsStress test not run yetSee the 8 checks

Bring the step you are stuck on: 30 minutes, a straight answer and the next step. If you would rather have it built, a reply within 5 working days.

The route

Daily digest

5 steps: 2 rule steps, 1 AI step and no person step

Copy, share, discuss

Copy

Available once this workflow passes its checks.

Share

todkett.com/automations/sales/hiring-signal-lead-finder/

Discuss

Opens when the first workflow passes its checks. No GitHub account? Ask about this by form.

How the library, the tracks and the talk work: the community page.

At a glance

Starts when
A schedule, twice a day, plus a daily digest email.
Input
Posts from job boards, public job APIs, forums, freelance sites, social posts and career pages.
Output
Scored leads in a table and a daily digest of the newest top leads.
AI
AI in one step only: classifying new posts
Tools
Public job board APIs and feeds, Automation community forums, Company career pages, n8n data tables, Google Sheets, Gmail, OpenRouter
Last stress test
No recorded run yet

01Steps

The steps

5 steps: 2 rule steps, 1 AI step and no person step. Companies posting jobs for work you could do are warm leads, but checking job boards, forums and freelance sites by hand takes hours and misses most of them.

  1. 01Trigger

    Posts from public sources

    Posts from job boards, public job APIs, forums, freelance sites, social posts and career pages.

  2. 02Rule step

    Prefilter, skip seen posts

    Cheap rules prefilter; posts seen before or reposted are skipped using an evidence ledger.

  3. 03AI step

    AI classifies new posts

    AI classifies only the new posts: signal type, relevance and organisation.

  4. 04Rule step

    Score and rank in code

    Code scores and ranks each lead, P1 to P3, saves it to a table and syncs a sheet.

  5. 05Output

    Daily digest

    A daily digest of new P1 and P2 leads and any failing source. Nobody is contacted.

    Daily digest

02Where AI is used

Where AI is used, and what stays with you

AI reads, judges or drafts. Fixed rules check, decide and route. A person approves anything that goes out.

AI step

Step 03

Classifying new posts

Posts are messy text from many sources. Rules remove most items before any AI call.

What the model reads

  • New posts that passed the prefilter rules, in batches of 10.

What it never sees or decides

  • Posts seen before or reposted: the evidence ledger skips them.
  • The score or the rank: code computes both.
  • Anyone's inbox: nobody is contacted.

Model calls: Not counted from the workflow yet. It is counted before this page goes live.

Fixed rules

They check, decide and route. 2 rule steps.

  • 02Prefilter, skip seen posts
  • 04Score and rank in code

A person

No step waits for a person. Human sign-off (ST7): fixed approval rule: nobody is contacted automatically.

03Checks

Checks around the AI

What runs before the model, and what checks its answer after, in plain words. Build these into your version too.

  1. 01Per-source schedule and health tracking, so a broken source never looks healthy.
  2. 02Prefilter rules run before AI.
  3. 03Posts are deduped against an evidence ledger and saved results; reposts are merged.
  4. 04AI runs in batches of 10.
  5. 05The score is computed in code from demand, relevance, freshness, fit, an identified organisation and commercial value.
  6. 06A 16-case labeled test set, including a prompt-injection case, checks the prompt and the scoring.
  7. 07It never contacts prospects.

04The 8 checks

Stress test

The 8 checks we run on every workflow, and you can run too. A check marked Not needed says why.

Status
Not yet0 of 8 that apply have a recorded pass
Last run
No recorded run yet
Model review
Still open

Checks marked Not yet are still open. 5 of them were seen in the workflow design on 23 Oct 2026; a check shows Passed only once its test run is recorded.

  1. ST1Test set

    Status: Not yet

    Seen in the workflow design on 23 Oct 2026. The recorded test run is still open.

  2. ST2Prompt attack

    Status: Not yet

    Seen in the workflow design on 23 Oct 2026. The recorded test run is still open.

  3. ST3Run twice

    Status: Not yet

    Seen in the workflow design on 23 Oct 2026. The recorded test run is still open.

  4. ST4Failure recovery

    Status: Not yet

    No recorded test yet.

  5. ST5Load and limits

    Status: Not yet

    No recorded test yet.

  6. ST6Failure alert

    Status: Not yet

    No recorded test yet.

  7. ST7Human sign-off

    Status: Not yet

    Seen in the workflow design on 23 Oct 2026. The recorded test run is still open.

  8. ST8Rules around the AI

    Status: Not yet

    Seen in the workflow design on 23 Oct 2026. The recorded test run is still open.

0 passed · 0 not needed · 8 not yet

What we test

We show what we tested, never the test data. The example below is made up.

05Test replay

How to test it

Before you trust a workflow, try to break it. The sample record runs the route step by step. Each break test below gets its own replay once its recorded run exists.

Record
SIGNAL-0482
Data
Sample run, made-up data
  1. 01Posts from public sources

    A new job post on a public board

    Done
  2. 02Prefilter, skip seen posts

    Not seen before; passes the prefilter rules

    Done
  3. 03AI classifies new posts

    Signal type, relevance and organisation classified

    Done
  4. 04Score and rank in code

    Ranked P1 in code; saved to the table

    Done
  5. 05Daily digest

    Listed in the daily digest; nobody contacted

    Daily digest
    Arrived

SIGNAL-0482 reached the last station: Daily digest. Listed in the daily digest; nobody contacted.

Sample run on made-up data. No stress-test run is recorded yet, so this plays the example, not a test. Nothing runs live.

The break tests for this workflow

  • ST3

    Same input sent twice

    Seen in the workflow design. No recorded run to replay yet.

    Not yet
  • ST2

    Instructions hidden in the input

    Seen in the workflow design. No recorded run to replay yet.

    Not yet
  • ST4

    A connected service fails

    No recorded run to replay yet.

    Not yet
  • ST8

    The AI returns a broken answer

    Seen in the workflow design. No recorded run to replay yet.

    Not yet

06Example

The example, in and out

Made-up data, written to show the steps; no recorded run stands behind it yet. Fernway is a made-up small software company, and its address ends in .example, so it cannot belong to a real business.

InputExample

Example on made-up data: one run over public job boards and forums for Fernway's services.

OutputExample

Posts seen before skipped by the evidence ledger. Prefilter rules drop most items before AI. New posts classified, scored in code, and the new P1 and P2 leads listed in the daily digest. Nobody contacted.

Measured result

From the run log, 16 Oct to 23 Oct 2026: about 9 percent of 13,236 collected items reached the AI step, over 10 runs.

Source: Run log, 10 runs, 16 Oct to 23 Oct 2026.

07Limits

When not to automate this

  • It never contacts anyone. It finds and ranks.
  • Sources change their pages; a broken source shows in the daily digest.

08What you need

What you need to run it

Each connected tool and the access it needs
ToolAccess
Public job APIs and feedsRead only, within each source's terms
n8n data tablesSettings, the evidence ledger and the results
GmailSend the daily digest to you
OpenRouterOne API key with a credit limit, for the classifier step
Where it runs
In n8n, self-hosted or n8n cloud.
Model calls
Not counted from the workflow yet. It is counted before this page goes live.
Running costs
Yours, paid to each service directly. Scope in writing first. A quote before any build. No prices on the site yet.

How to adapt it

Changes we have made or tested on this workflow.

  1. 01Change the sources and search queries in a settings table, without editing the workflow.

What changed

  1. ContentFirst version, written from the 23 Oct 2026 catalog.

09FAQ

Questions people ask about it

Does it contact the companies it finds?

No. It finds, scores and lists them in a daily digest. Nobody is contacted automatically.

How much of the input reaches AI?

Over 10 runs from 16 Oct to 23 Oct 2026, about 9 percent of collected items reached the AI step. Rules and a ledger of posts already seen removed the rest.

How is it tested?

A labeled test set of 16 cases, including a prompt-injection case, checks the prompt and the scoring.

10Related

Stuck on your version of this workflow?

Bring the step that does not work. In 30 minutes you get a straight answer and the next step. If you would rather have it built, tell us your tools and the rules you already follow. Scope in writing first. A quote before any build. No prices on the site yet.