- What an AI workflow actually is, in plain English
- The four building blocks: triggers, conditions, actions, and AI judgement
- A complete walkthrough of the 'Monday morning shortlist' workflow
- Three more workflows you can ship this week
A workflow is just a recipe. When this happens, do that. When a candidate accepts an interview, sync my calendar. When a new role opens, start sourcing. When Friday hits, summarise the week.
Recruiters have been using workflows forever. They have just been calling them “the way I do things”. The new thing is that AI can now do the steps that used to need a human. That is what changes the game.
The four building blocks
- Trigger: the thing that starts the workflow. A new candidate, a stage change, a schedule, a message received.
- Condition: the gate that decides whether to keep going. Score above 70? Has 5+ years experience? Based in the EU?
- Action: the thing the workflow does. Send an email, book a meeting, update a record, post to Slack.
- AI judgement: any step where the workflow asks an AI to think. Score this candidate. Summarise this call. Draft this message.
The 'Monday morning shortlist' workflow
The most common workflow we see new customers build first. Goal: every Monday at 9am, you wake up to a shortlist of 10 fresh candidates for your top three open roles, scored, with reasoning.
Here is the recipe in plain English:
Trigger: Every Monday at 06:00. Step 1 (AI): For each open role tagged 'priority', search across 30 data sources for candidates who match the role rubric and changed jobs in the last 6 months. Step 2 (Condition): Filter to candidates with a match score >= 75 and not already in our database. Step 3 (AI): For each remaining candidate, find three personalisation hooks from their profile. Step 4 (Action): Compile a shortlist for each role, email the recruiter at 09:00 with the scored list, reasoning, and personalisation hooks ready to use.
Building it step by step
Inside any modern workflow tool (Vitae, Zapier, n8n, Make), the build looks like this. We will use Vitae naming, but the pattern works in any of them.
- Add a Schedule trigger. Set it to Mondays at 06:00.
- Add a Loop step over your active roles, filtered to tag = 'priority'.
- Inside the loop, add an Agent step. Give it the role rubric and tell it to source 10 candidates per role.
- Add a Filter step. Match score >= 75 AND not in candidate database.
- Add another Agent step. Generate three personalisation hooks per candidate.
- Add an Email Send step. Format as a clean digest, one section per role.
- Test it manually with one role first. Refine. Then turn the schedule on.
Total build time: 25 to 40 minutes the first time. After that you can clone it for any pattern.
Three more workflows worth shipping this week
1. Auto-screen new applicants
Trigger: Application received. Step 1 (AI): Score the candidate against the JD. Output: score 1-100, top 3 strengths, top 3 gaps. Step 2 (Condition): - If score >= 80: advance to recruiter review queue. - If score 50-79: move to talent pool with a tag. - If score < 50: send polite decline (drafted by AI, recruiter approves before send).
2. Friday digest to your client
Trigger: Every Friday at 16:00. Step 1: Pull all activity for client X this week. Step 2 (AI): Summarise into 5 bullet points. Step 3: Email the client with the summary + a link to the live portal.
3. Re-engage silver medalists
Trigger: Every quarter on the 1st. Step 1: Pull all candidates marked 'silver medal' in the last 12 months. Step 2 (AI): For each, check if they have changed roles, posted recently, or moved location. Step 3: Draft a personalised re-engagement message per candidate, queued for recruiter approval.
Where most teams get stuck
The first workflow takes 30 minutes. The second takes 20. The fifth takes 5. The hardest part is not the building, it is the deciding. Most recruiters can list 50 things they wish were automated. Pick the one that runs the most often and bothers you the most. Build that one. Ship it. Repeat.
The bottom line
Workflows are how you move from AI-as-a-chat-tool to AI-as-an-employee. Start with one. Get one win. Build five over a quarter. The compounding effect on your time is not subtle.


