What's new at adextract: ad intelligence updates for July 2026

July has been a busy month at adextract. We shipped several updates that change how performance marketers and agencies use AI agents for competitive ad intelligence. If you have been tracking the ad intelligence space, you know the biggest shift in 2026 is the move from dashboard-based tools to agent-native workflows. Our July updates double down on that bet.

Here is a quick rundown of what changed and why it matters for your competitive research workflow.

What changed this month

We focused on three things in July: making the MCP server faster and more reliable, expanding our ad library coverage, and tightening the agent connection flow. Each of these matters on its own, but together they change how quickly a team can go from asking a question to getting a useful competitive signal.

The core idea behind adextract has not changed: give AI agents direct access to competitor ad data across Meta, Google, TikTok, and LinkedIn. What changed is how fast and how deep those agents can now go.

If you are new here: adextract is an MCP server that connects to every major ad library. Your AI agent (Claude, Cursor, or any MCP-compatible client) can search, read, and compare competitor ads without leaving the conversation. No dashboards, no exports, no switching tabs.

MCP server: ad intelligence inside your AI workspace

The biggest July update is to our MCP server at mcp.adextract.co. We rebuilt the search pipeline to be faster and more consistent across ad libraries. Queries that previously took 8 to 12 seconds now resolve in under 4 seconds on average. That may not sound like much, but when your agent is running a multi-step research workflow, those seconds add up.

We also made the OAuth connection flow smoother. Claude users can now connect with a single click: paste the MCP endpoint URL, sign in with Google, and the agent has full access to ad library search. No API keys, no config files, no manual token management.

For teams using platforms beyond Claude, we published updated connector docs for Cursor, Windsurf, and any editor that speaks MCP. The setup is the same: one URL, one OAuth flow, done. You can read the full setup guide on our blog

StackAdapt recently launched their own MCP server for campaign intelligence inside Claude. The Model Context Protocol is becoming the standard way ad platforms connect to AI agents. Our July updates keep adextract at the front of that shift for competitive research use cases.

Multi-library ad search: all platforms, one query

adextract always searched across Meta, Google, TikTok, and LinkedIn. What changed this month is how unified that search feels. Instead of running four separate queries and stitching results together, your agent now gets a single ranked feed with results from every library, deduplicated and sorted by relevance.

This matters for real workflows. If you are researching how a competitor like Monday.com runs ads, you want to see their Meta creative, their Google search copy, and their LinkedIn positioning in one view. That cross-platform picture is what tells you whether their strategy is coherent or scattered. Our July update makes that picture available in a single agent response.

We also improved coverage for the TikTok ad library, which has been a common request from agencies tracking DTC brands. TikTok's ad transparency tools are less mature than Meta's, which means you need a reliable third-party layer to get consistent results. Our July update closes that gap.

Agent-first design: why it matters for competitive research

Most ad intelligence tools are built as dashboards. You log in, filter by platform, type a competitor name, scroll through results, and manually note what matters. That workflow made sense in 2022. In 2026, it is the slow path.

Agent-first tools work differently. Instead of you navigating a dashboard, you ask a question: "Show me every new ad Nike launched on Meta this week and compare the creative angles to Adidas." Your agent calls adextract, gets the data, and answers in seconds. No clicking, no filtering, no exporting to a spreadsheet.

This is not just about saving time. It changes what questions you can afford to ask. When a competitive search takes 20 minutes, you do it twice a month. When it takes 20 seconds, you do it every morning. That shift from periodic analysis to continuous monitoring is where the real strategic advantage lives.

We wrote about how AI agents find competitor ads in more detail in an earlier post. The July updates make every workflow described there faster and more reliable.

What's coming next

We are working on a few things for August and September that are worth flagging:

Ad change detection. The most common follow-up question after "what ads is this competitor running" is "when did they change them." We are building a change detection layer that tracks creative swaps, copy updates, and landing page shifts over time. Your agent will be able to answer "what did this brand's ad strategy look like in Q1 vs Q3" without a manual audit.

Competitive spend estimates. We are exploring ways to surface estimated spend ranges for competitor campaigns, using impression frequency and placement data as proxy signals. This is technically harder than creative tracking, but it is the feature most requested by performance marketers who manage budgets.

Multi-agent workflows. A single agent calling adextract is powerful. Multiple agents collaborating on competitive research is the next step. We are designing workflows where one agent monitors ad libraries, a second tracks landing pages, and a third compiles weekly competitive briefs. This multi-agent pattern is something we explored in our guide to building multi-agent ad intelligence workflows.

Getting started with adextract

If you have not tried adextract yet, here is the quickest path:

1. Go to adextract.co and copy the MCP endpoint URL.

2. Open Claude (or any MCP-compatible editor) and add a custom connector.

3. Sign in with Google. No API key needed.

4. Ask your agent to search for a competitor's ads. Example: "Search Meta ads for Nike running in the last 30 days."

The first 50 searches are free. No credit card, no trial expiration. If you are a solo founder or a small agency team, that is probably enough for a full competitive audit across your top 5 competitors.

We maintain a growing library of guides and tutorials on the blog. If you want to go deeper on a specific platform or workflow, check out our posts on competitor ad creative analysis for Meta and Google and how to track competitor ads without burning your budget.

Frequently asked questions