What Level Are You At? The Four Stages of AI in Staffing — And How to Move to the Next One

Published: March 16, 2026 | By: The Crossing Report | 8 min read

---

Summary

Most small staffing agencies are using AI at Level 1 or Level 2. They're prompting tools, getting suggestions, and manually handling every next step. The firms that are pulling ahead in placement speed and revenue growth are operating at Level 3 — where AI proactively runs workflows and the recruiter manages exceptions. This piece maps the four levels, shows you where your firm is, explains what Level 3 actually looks like in practice, and gives you one move to make this week to advance.

---

The Question Worth Asking First

A staffing agency owner recently described their AI setup this way: "We use AI constantly. We paste job descriptions into ChatGPT, we generate candidate outreach messages, we ask it to summarize resumes. It's a huge part of how we work."

When I asked whether their placement speed had improved, there was a pause. "About the same."

This is the gap that matters. Many staffing firms in 2026 are using AI tools every day without moving up the maturity curve. They're doing Level 1 work with Level 1 tools — and the competitive firms are operating two levels above them.

The difference is not which tools you're using. It's how the work flows.

---

The Four Levels of Staffing AI Maturity

The industry research from platforms like Bullhorn, Avionté, Atlas, and IRIOnline has converged on a four-level maturity model. Here's the plain-language version for a 5–25 person agency.

---

Level 1 — Assistive: AI Does What You Tell It

What it looks like: You open a tool (ChatGPT, Copilot, your ATS's basic search), give it a prompt, get output, and then manually handle every next step. AI helps you search faster or drafts a message, but you initiate and complete every step of the process.

Common tools at this level: ChatGPT for message drafting, LinkedIn's basic search filters, keyword matching in legacy ATS platforms.

Time savings: Marginal. Measured in minutes per task, not hours per search.

The tell: You are always the one initiating the AI action. Nothing happens unless you start it.

Where most small firms are: Here, or halfway into Level 2.

---

Level 2 — Copilot: AI Suggests, You Decide

What it looks like: AI proactively offers recommendations — candidate match scores, suggested outreach sequences, ranked shortlists — but a human reviews every item before it goes anywhere. You make decisions faster because the AI already did the first sort. But you're still touching every candidate.

Common tools at this level: Bullhorn's AI matching, Avionté Fusion with AI-scored candidates, Loxo's AI-powered sourcing, LinkedIn Recruiter AI. Most modern ATS platforms with AI features enabled are Level 2 tools.

Time savings: Meaningful. Firms at Level 2 typically cut initial candidate screening time by 40–60%. Shortlist quality improves because you're working from a sorted list, not a raw pool.

The tell: You review every AI suggestion before acting. The AI recommends; you approve.

Compliance note: If your AI scores or ranks candidates before you see them, Colorado's AI Act (effective June 30, 2026) and Illinois HB 3773 (effective January 1, 2026) may require disclosure to candidates and documentation of your AI decision-making process. This applies at Level 2 and above. See the compliance section below.

---

Level 3 — Semi-Agentic: AI Runs the Workflow, You Handle Exceptions

What it looks like: AI proactively manages multi-step workflows without you initiating each step. It monitors your open requisitions, moves qualified candidates through defined pipeline stages, triggers outreach sequences, schedules interviews, and sends reminders — then flags the exceptions (unexpected drops, borderline matches, declined offers) for a recruiter to handle.

The recruiter's job shifts: instead of running the workflow step by step, they're managing the exceptions the AI can't resolve.

Common tools at this level: Advanced Bullhorn workflows with automation rules, Avionté with AI-driven pipeline management, Atlas. At mid-to-large staffing firms, Eightfold AI is the platform seeing the most Level 3 adoption in 2026. For smaller agencies, some Loxo configurations can get close to Level 3 for sourcing workflows.

Time savings: Dramatic. The 2026 Bullhorn GRID data shows that firms at this level — those with semi-agentic AI across the recruiting cycle — are 4x more likely to report revenue growth and 56% of top performers achieve placements in under 10 days.

The tell: Things happen in your ATS without you triggering them. Candidates move through stages, outreach goes out, follow-ups land — and you're only pulled in when something needs judgment.

The gap: Most small staffing agencies are at Level 1 or Level 2. The competitive firms are here.

---

Level 4 — Fully Autonomous: AI Manages End-to-End

What it looks like: AI manages most of the recruiting workflow — including candidate selection, outreach, scheduling, and documentation — with minimal human intervention. Humans set parameters and review final placement decisions.

Where this exists in 2026: Limited. Large enterprise staffing platforms are beginning to test fully autonomous workflows for high-volume, standardized roles (light industrial, call center, administrative). Most firms in our readership are not at this level, and the compliance, accuracy, and relationship risks make it premature for most placements.

For our readers: Level 4 is a 2027–2028 story. Focus on Level 3 first.

---

Where Your Firm Is Now: The Self-Assessment

Answer these three questions honestly:

1. Do things happen in your ATS without you triggering them?

• Yes, regularly → You're at Level 3.
• Sometimes, for specific automations I set up → You're at Level 2.
• No, I initiate everything → You're at Level 1.

2. When a new candidate applies to an open search, what happens next?

• AI scores them, moves them to a shortlist, triggers a customized outreach sequence, and alerts me if the candidate responds or has a flag → Level 3.
• AI scores them and adds them to a ranked list I review daily → Level 2.
• I log in and review the application myself → Level 1.

3. How much time does a recruiter spend on any given search before the first qualified candidate phone call?

• Under 2 hours → Likely Level 3.
• 2–5 hours → Likely Level 2.
• 5+ hours → Level 1.

---

Moving from Level 1 to Level 2

The fastest path doesn't involve buying anything new.

Most staffing agencies in 2026 are paying for ATS platforms that include AI scoring, candidate matching, and automated outreach — but haven't turned those features on. The AI is there. It's dormant.

This week:

1. Log into your ATS (Bullhorn, Avionté, Loxo, Manatal, or equivalent)
2. Go to Settings → AI features or Automation → look for "candidate matching," "AI scoring," or "automated outreach"
3. Enable the feature for one active search
4. Run it on your next requisition and compare the shortlist to what you'd have built manually

You may be a settings change away from Level 2.

If your current ATS doesn't have AI scoring (typically older legacy systems), the most cost-effective upgrade for a 5–20 person firm is Loxo (starts at approximately $165/user/month with AI sourcing included) or Manatal (from $15/user/month, with AI ranking). Both get you to Level 2 without enterprise pricing or a sales call.

---

Moving from Level 2 to Level 3

This is the harder transition — and the more valuable one. It requires three things:

1. Defined workflow stages in your ATS. Level 3 semi-agentic AI needs a clearly mapped pipeline: applied → scored → shortlisted → outreach sent → response received → interview scheduled → offer → placed. If your pipeline stages are informal or inconsistently used, the automation has nothing to trigger on. The first move is formalizing your workflow in your ATS — even if it's just renaming the stages and committing to using them consistently.

2. Decision rules for each transition. For the AI to move a candidate from "shortlisted" to "outreach triggered," it needs a rule: what score threshold, what criteria, what action follows. Level 3 requires you to write down the decisions you currently make intuitively and turn them into rules the system can apply.

3. Exception discipline. The risk at Level 3 is micromanagement creep — recruiters who get alerts and then manually re-run the same steps the AI already completed. The discipline is trusting the defined workflow and only engaging when the AI flags something it can't resolve. This is a culture shift, not just a technical one.

Platform path for small firms: Bullhorn and Avionté both have Level 3 capabilities available on their standard plans — the barrier is configuration, not licensing. If your firm is already on either platform, contact your customer success rep and ask specifically about "workflow automation" and "AI-triggered pipeline stages." Most firms don't know what they're already paying for.

For firms not on enterprise ATS platforms: Loxo's higher-tier plans include workflow automation that can approximate Level 3 for sourcing-heavy searches.

---

The Compliance Issue You Can't Skip

If your firm is at Level 2 or above — meaning AI influences which candidates you present to clients — two new legal frameworks apply in 2026:

Colorado SB24-205 (effective June 30, 2026): Staffing agencies using AI to score, rank, or select candidates for employment decisions are classified as "deployers" of high-risk AI. Requirements include: notifying candidates that AI was used in their evaluation, documenting the AI system used and the decision criteria, and allowing candidates to appeal the AI's assessment. Penalties: up to $20,000 per violation, counted per candidate or transaction.

Illinois HB 3773 (effective January 1, 2026): Employers and employment agencies using AI tools that "substantially assist or replace" human decision-making in employment decisions must disclose this to candidates. If you're placing candidates for Illinois clients using AI-assisted screening, this law likely applies.

The three-item compliance checklist for staffing agencies using AI:

1. Add an AI disclosure to your candidate intake form (one sentence: "We use AI-assisted tools in our candidate evaluation process")
2. Document which AI tools influence candidate decisions — name the platform, the feature, what it scores
3. Establish a process for candidates to request human review of their AI-generated score

This is not a reason to stop using AI. It's a reason to use it with documentation. A staffing firm that's at Level 3 with proper compliance is significantly better positioned than one at Level 1 with no documentation.

---

Your Action Item This Week

The self-assessment question that tells you your next move:

Log into your ATS and answer honestly: In the last week, how many candidate actions happened automatically — without you initiating them?

• Zero: You're at Level 1. Your move is to turn on AI scoring for one active search this week. Start with the ATS you already have.
• Some: You're at Level 2. Your move is to identify one pipeline stage transition you still trigger manually and write the automation rule that would handle it automatically. If you can define the rule, you can automate it.
• Most: You're at Level 3. Your move is compliance: verify your candidate intake form includes an AI disclosure and that you've documented the AI tools involved in candidate scoring.

The competitive gap in staffing right now is not primarily about which tools firms have — it's about which firms have configured the tools they already own to actually run workflows. Most small agencies are three settings changes away from Level 2. Most Level 2 firms are three workflow rules away from Level 3.

The firms that win in 2026 are the ones who close that gap before their competitors do.

---

The Crossing Report publishes weekly intelligence for professional services firm owners navigating the AI transition. Subscribe here.