AI with Michal

AI-native

For TA and recruiting teams: an operating style where models, skills, and automation are assumed in the design of work, with clear handoffs and QA, not one-off chats when you remember to open ChatGPT.

Michal Juhas · Last reviewed May 2, 2026

Who this is for

TA leaders, full-cycle recruiters, and sourcers who are past the novelty phase and need repeatable quality: same tone, same structure, same checks on every req.

In practice

  • Document how you hire: briefs, red lines, tone, and disallowed phrases live where the model can read them (Markdown beats pasted Word for tokens and diffs).
  • Reuse, do not re-prompt: custom GPTs, Claude skills, or Gems carry context so a short user message still expands into on-brand output.
  • Own the handoff: anything a candidate or hiring manager sees gets a second pair of eyes until the system earns trust.

Where it breaks

If nobody maintains the knowledge base, if prompts live only in one person’s chat history, or if legal and HR never weighed in on automation, you get fragile “AI theater” instead of AI-native operations.

From recent workshops

In live sessions on AI in recruiting and sourcing automation, the same pattern repeats: teams that connect skills, knowledge bases, and (when ready) APIs ship durable workflows. Teams that chase the newest model without fixing inputs stay stuck re-editing generic drafts. Interface matters less than whether the assistant can see your SOPs and constraints.

Boolean versus systemized work

ModeWhat you doRisk
Ad hoc chatRe-type context each timeInconsistent output, no audit trail
Systemized (Gems, GPTs, skills)Pre-load tone, format, must-havesNeeds owners to update when brand or policy changes
Automated flowsTools like Make or n8n move rows and draftsNeeds monitoring, GDPR, and API hygiene

For the sourcing angle on when to stay literal versus semantic, read Boolean search vs AI sourcing.

Related on this site

Frequently asked questions

What does AI-native mean in a recruiting team day to day?
It means you design recurring work (intake, sourcing, screening notes, outreach) assuming a model or automation can help: saved instructions, shared skills or Gems, structured inputs like scorecards, and a human who owns verification before anything candidate-facing ships.
How is AI-native different from "we use ChatGPT sometimes"?
Occasional chat is reactive. AI-native teams embed the model in repeatable steps, version the instructions, and debrief what broke when a hire closes. Live workshops surface the gap: the same sourcers get wildly different output from identical tools when only one side systemizes.
Where should a team start without boiling the ocean?
Pick one high-volume artifact (for example outbound for one role family or intake-to-brief). Ship a small chain: inputs, model pass, reviewer, where the output lands (ATS field, sheet, email). Expand only after that loop is trusted. The Starting with AI: the foundations in recruiting course is built for that progression.
What goes wrong when teams declare themselves AI-native too early?
Burnout from half-wired automation, inconsistent candidate experience, and skipped verification on employers, dates, and policy-sensitive wording. Naming limits (hallucinations, bias, data retention) is part of mature AI-native practice, not an admission of failure.
Do we need engineers to become AI-native?
No, but you need literacy at the right depth: Markdown for readable knowledge bases, light Git or folder discipline for skills, and clarity on who approves external sends. Advanced workshops go deeper on APIs; many teams still win with Gems, GPTs, and tight prompts first.
Which internal resources pair with this mindset?
Read What is AI-native work? on the blog, browse AI tools for the stack you standardize on, and use Guides by role to align sourcers, recruiters, and TA leads on the same vocabulary.

Annual membership

All workshops, courses, and premium materials

Members get live workshops, courses, recordings, and member-only resources. One pass for the year.

See membership →

← Back to AI glossary in practice