Vendor network data says agentic traffic is compounding quarterly. The expensive part is not volume-it is still treating every bot like the same bot.
Via Business Wire (DataDome): DataDome Report: AI Agent Traffic Surged 45 percent in Q2, Driving a Wave of Agent Trust Policy Adoption
DataDome, a bot and agent trust management vendor, released The AI Traffic Report Q2 2026 via Business Wire on July 16, 2026 (also syndicated by outlets including Yahoo Finance and CIO Influence). The vendor-reported headline: AI agent traffic on DataDomes network surged 45 percent quarter over quarter, reaching 17.7 billion AI agent requests in Q2 2026, up from 12.2 billion in Q1.
That is not a trivia statistic for marketing sites and API-heavy SMEs. It is a cost, abuse, and data-leakage surface that grows whether or not your team has a label for agent traffic in the analytics dashboard. DataDomes own framing: agents are becoming more purposeful, more varied, and harder to manage with a one-size-fits-all allow-or-block rule.
According to the release, Metas agents took a dominant share of AI agent traffic on DataDomes network: Meta-ExternalAgent grew +74 percent QoQ and Meta-WebIndexer grew +163 percent QoQ, together accounting for the majority of AI agent traffic-a dominance the vendor says did not exist in Q1.
Crawl volume and referral value are diverging. ChatGPT-User, described as the top agent in Q1, declined 6 percent QoQ in absolute request volume, yet ChatGPT remained the overwhelming leader in referral traffic at 80-88 percent of all AI referrals every month, growing +17 percent QoQ. MCP traffic is now a measurable network signal, the release says, with peaks approaching 500,000 requests per day and clear daily usage cycles.
On policy: 54 percent of DataDome customers have already implemented agent trust policies-vendor-reported-on the premise that distinguishing agents that serve business objectives from those that merely consume infrastructure is no longer optional. Jerome Segura, VP of Threat Research at DataDome, is quoted: not all agents are created equal, and organizations building policy around those distinctions are gaining an edge.
Read the composition shift as an operating problem, not a brand war. If Meta agents dominate raw request volume while ChatGPT dominates referrals, a blunt block AI bots rule can cut revenue-bearing traffic and still leave you paying for high-volume indexing that returns nothing. Tiered policy-identify, score intent, allow/limit/block by endpoint-is the adult version of robots.txt wishful thinking.
You do not need DataDomes 5-trillion-signals marketing line to recognize the failure mode: logging that says bot without saying which agent, why it is there, or whether it returns referral value. Allowlisting by user-agent string alone is how spoofed crawlers walk in the front door. Blocking everything that looks automated is how you cut off useful assistants and partner integrations by accident.
Owner-led firms feel this as unexplained bandwidth bills, scraped pricing pages, poisoned lead forms, and our chatbot partner needs API access requests with no governance owner.
DataDomes network processed 17.7 billion AI agent requests in Q2 2026-up 45 percent from 12.2 billion in Q1-as agent trust policy adoption accelerated.
AgentsROI.ai does not sell bot mitigation hardware. I help owner-led firms invent the missing operating layer: inventory, policy, and accountability. A Shadow-AI Risk Assessment discovers which external agents-and which internal staff agents-already touch your site, CRM, or knowledge base. A Fractional AI Officer writes the tiered access rules: who may crawl, who may act, who needs human approval. Managed AI Operations keeps those policies current as agent identities and volumes shift quarter to quarter-the exact pattern DataDomes Q2 numbers illustrate.
Vendor note: DataDome is both the data source and a seller of agent-trust products. Treat the 54 percent customer-adoption figure as vendor-reported among its own customer base, not as a global market census. The volume trend still stands as a directional warning even if you never buy their stack. Direct fetch of the Business Wire URL timed out in this drafting environment; facts above follow the same Business Wire release as syndicated by Yahoo Finance and CIO Influence.
A practical starter inventory for SMEs: list every integration that can read your site or API; classify each as human browser, verified partner bot, unknown crawler, or AI agent; assign an owner; and decide allow, rate-limit, or block per endpoint class. Without that inventory, agent trust is a slogan taped to a WAF.
DataDomes vendor report is a loud quarterly reminder: agentic traffic is compounding, composition is shifting, and bots is no longer a useful noun. Treat agent identity and intent as first-class controls, or keep paying for someone elses training corpus and infrastructure habit.
Start with a Shadow-AI Risk Assessment if you cannot list which agents hit your systems last month. Book a no-pressure assessment.
For owner-led firms tracking this story, the operational takeaway is straightforward: treat the source facts as constraints on how you buy, govern, and measure AI this quarter-not as inspiration for another unowned pilot. AgentsROI.ai Tier-1 starting points-Workflow ROI Audit, Shadow-AI Risk Assessment, and Fractional AI Officer-turn coverage like this into a decision, an owner, and a metric.
This article summarizes publicly reported information and is for general informational purposes only. It does not constitute legal, tax, financial, investment, security, or compliance advice. AgentsROI.ai is not a law firm, accounting firm, or registered investment adviser. Facts, pricing, statistics, and product capabilities cited here reflect the sources listed at the time of writing and may change. Readers should verify current information independently and consult qualified professionals regarding obligations specific to their industry, jurisdiction, and circumstances-including applicable New York State and New York City requirements. AgentsROI.ai may have commercial relationships with vendors mentioned; where material, such relationships are disclosed. Nothing in this article is an endorsement of any specific AI product, model, or provider.
