Skip to content
A clearer record of competitor creative.Meet the method

Competitive intelligence infrastructure

Winning ads,ready for your agents.

Meridic monitors public competitor ads, qualifies likely winners from complete observations, and serves clean evidence to Claude Code, Codex and every MCP-compatible agent.

Public ad activity · Evidence-qualified signals · MCP-ready output

Meridic / Intelligence infrastructureNightly record · 02:00 UTC MCP endpoint live
01
Public sourcesWatched competitors
02
Evidence engineDated observations
03
Winning signalsQualified patterns
04
Meridic MCPClean, typed output
Competitive record / Today

Winning signals, with receipts.

7 / 7 complete
01NorthstarMove lighter
14dobserved+2variantsMediumconfidence
Likely scalingEvidence ready
02EverfieldOne less step
11dobserved+4variantsMediumconfidence
Likely scalingEvidence ready
03CurrentReady when you are
8dobserved+3variantsEmergingconfidence
Likely scalingEvidence ready

Interpretation from public activity. Spend, conversions, CPA and ROAS remain unavailable.

One evidence layer.
Every research surface.

Claude CodeCodexAny MCP client

From signal to next move

Research that gives
the question somewhere to go.

Built for the moment between seeing a competitor’s creative and deciding what your team should explore next.

01

A new angle appears

Notice the move before you imitate the ad.

A launch becomes useful when you can place it beside the advertiser’s existing creative — and see what changed.

02

The family expands

See the pattern, not just the newest variant.

Continued activity plus additional variants creates context for a creative brief. The evidence stays attached.

Find your workflow

Meridic MCP / One endpoint

A living competitor record.
Ready for every agent.

Teach your workflow where the evidence lives once. Meridic keeps the record current and makes the same qualified context available wherever your team works.

Connect your first workspace
01

Live competitor tools

Ask for likely scaling ads, new winners, recorded endings, brand momentum or a direct competitor comparison.

Workspace-fenced data
02

Analysis skills included

Give every connected agent the same versioned research method, evidence rules and creative analysis vocabulary.

Discoverable without a key
03

Clean evidence, honest limits

Every response separates verdict, confidence, evidence and limitations so an agent knows what it can safely conclude.

Typed JSON output
One evidence layerhttps://api.meridic.com/mcp
Claude CodeCodexClaude DesktopCursorAny MCP client
Connected

02 / Evidence before inference

A signal has to
earn its place.

Observation first. Interpretation second. Mature signals require 7 complete observations with no gaps in the evidence window.

An example evidence window7 observations
01
02
03
04
05
06
07
Complete recordQualified interpretation

A missing night holds the conclusion.
It doesn’t get filled in with a guess.

  1. 01

    Observe consistently

    Build a dated record of public ad activity, with coverage you can inspect.

  2. 02

    Compare the changes

    Separate a new creative, an additional variant and a recorded ending.

  3. 03

    Keep the limits visible

    Show what the record supports. Say clearly what it cannot tell you.

A little more clarity

Good questions.

Clear answers, including the things
the record cannot tell you.

What does Meridic actually track?

Public Meta ad activity: what appears, how it changes across observations, what continues and what has a recorded ending. It turns a series of snapshots into a record you can investigate.

Does a scaling signal mean an ad is profitable?

No. Additional variants and continued activity can support an interpretation of likely scaling. They do not reveal spend, conversions, CPA or ROAS.

Why wait for 7 complete observations?

A gap in the record can look like a change in behavior. Meridic keeps first-day facts visible while holding back mature signals until the required evidence window is complete.

Can Meridic connect to Claude Code and Codex?

Yes. Meridic exposes one HTTP MCP endpoint for Claude Code, Codex, Claude Desktop, Cursor and other MCP-compatible clients. Live workspace data requires a read-scoped API key.

What can an agent ask Meridic?

It can retrieve likely scaling ads, new winners, recorded endings, brand momentum and competitor comparisons. Responses keep verdict, confidence, evidence and limitations separate.

Your agents need better context

Stop prompting
from snapshots.

One watched competitor set. One evidence layer. Every MCP client.