Archive’s Archie Vets Creators—but Humans Still Own the Approval

Archive’s August 31, 2026 announcement introduced Archie, a Creator Search agent that turns a campaign brief into proposed screening criteria and returns a reasoned creator shortlist through Slack. The product is in early access through a waitlist for customers on Archive’s 2026 Startup, Growth and Enterprise plans.

Archie automates the first-pass work of sourcing and assessing creators, but it does not make the final selection. Marketers can review its criteria, approve or reject results and overrule an exclusion, keeping the recommendation separate from the decision to advance a creator.

Archie converts an open brief into reviewable rules

The workflow begins with a natural-language campaign brief. Archie can work from a detailed request, a rough idea, a PDF added to a conversation or an existing discussion, then writes back the criteria it intends to apply before searching.

Those proposed rules can include follower range, campaign niche, content signals and disqualifying conditions. This gives the marketer an opportunity to catch a misunderstood category, an incomplete competitor list or another specification problem before it removes candidates from consideration.

Net Influencer’s September 1 report confirms that marketers can brief Archie through Slack or Microsoft Teams, receive a shortlist with reasoning attached and feed approvals or rejections into the next round. Preserving both the original brief and Archie’s proposed criteria creates a record of what the team requested and how the agent translated that request into screening rules.

The agent replaces the profile-by-profile first pass

Once the criteria are accepted, Archie sources candidates and evaluates them individually. Its checks cover whether an account remains active, whether published posts reveal competitor conflicts, whether the creator’s actual content fits the brief and whether videos or comments contain patterns associated with bots or fake audiences.

Results arrive in batches with reasons attached to recommendations and exclusions. If an earlier competitor post causes Archie to reject a creator, the workflow can expose that specific post for inspection, allowing the marketer to decide whether the conflict is current, relevant and genuinely disqualifying.

This is the manual screening Archie is designed to replace: opening profiles, checking recent activity, reading content and comments, and repeatedly applying campaign-specific exclusions. Human attention moves to disputed cases and consequential exceptions rather than disappearing from the process.

Explainable recommendations still carry evidence gaps

A visible reason makes a decision inspectable; it does not make the conclusion certain. IVRIS Tech’s launch analysis identifies several details that remain undisclosed, including calibrated confidence levels, the handling of incomplete creator histories and the conditions under which Archie might abstain because evidence is ambiguous.

Those gaps matter because each screening category depends on context. A sponsorship may sit outside the history available to the agent, an old competitor relationship may no longer violate the brief, and unusual but genuine audience behavior may resemble manipulation. The explanation helps a reviewer locate the contested judgment, but it cannot prove that the available record is complete.

An auditable review therefore needs the evidence behind negative decisions to remain connected to the decision. A competitor exclusion can retain the triggering post; an inactivity finding needs a defined review window; and a fake-audience flag should remain a signal for investigation rather than become an unsupported accusation about the creator.

Approval remains a distinct human decision

Archie’s output is a recommendation queue, not a completed booking list. Marketers can approve and reject candidates as batches arrive, and they can reverse an exclusion after examining the supporting material. Feedback can influence later batches without transferring final authority to the agent.

The clearest audit trail preserves four separate records: the original campaign brief, Archie’s proposed criteria, the evidence and reasoning attached to each inclusion or exclusion, and the marketer’s final disposition. An override should retain the agent’s original reason alongside the reviewer’s rationale, rather than erasing the disagreement.

The published workflow ends with a reasoned shortlist. It does not establish that Archie independently contacts creators, negotiates terms or books them, and the brand retains responsibility for the creator relationship after selection.

For now, Archie remains a waitlisted early-access product rather than a generally available release. Its launch documentation establishes the brief-to-criteria workflow, automated screening, evidence-backed explanations and human override, while leaving precise data coverage and confidence behavior unresolved. Wider access and operational evidence will determine how consistently marketing teams can audit its judgments without recreating the entire manual review.

Also read:

• AI Is Scaling Video Ads—But Humans Still Own the Claims
• Venting to AI Can Ease Distress—but the Evidence Is Still Short-Term
• AI Data Automation Speeds Routine Work—Humans Still Own the Exceptions