AdKit review for running Meta ads and Google ads through an AI agent

What AdKit is and who it is for

AdKit is an ads toolbox built around one idea: running ads should mean asking an AI agent, not clicking through Ads Manager. It gives agents such as Claude, ChatGPT, Claude Code, Codex and Cursor an MCP server (and a CLI), so the agent can pull live campaign data, draft budget and targeting changes, generate ad creative and research competitor ads from a normal conversation.

Alongside the MCP, there is a web app with a multi-platform ad library, competitor tracking, activity alerts, a swipe file and an AI ads generator. The vendor pitches it at founders running their own ads without an agency, agencies and media buyers handling several ad accounts, growth marketers who already do most of their work with AI agents, and brand managers keeping an eye on competitors. The founder, Nico, says on the site that he was a media buyer before building it.

Key features of the ads MCP and app

Install the MCP in your AI assistants

The getting started docs list setup guides for Claude and Cowork, ChatGPT, Perplexity, Claude Code, Codex, Cursor, Gemini CLI, OpenClaw, Hermes, n8n, Grok, Meta Muse and HubSpot. The install is described as connecting your ad account and copying one line into your agent, and the founder offers help if you're stuck. There is also a doc on moving local images and videos from an agent session into Meta or Google Ads, and one on giving each business its own project.

Draft-first changes

This is the feature that matters most if you are nervous about letting an agent near your ad spend. The vendor says every change your agent makes (campaigns, ad sets, creatives) is saved as a draft in the dashboard until you click approve, so nothing touches your live accounts without you. It also says it is an officially approved Meta and Google tech partner, sends every action through the official API, manages rate limits and validates each request before sending.

Competitor ads, tracking and alerts

The ad library lets you browse and search ads from Meta, Google and LinkedIn in one place, and the vendor says it holds over 1M ads. You can add any competitor, see their full history over time and get an alert when competitors launch new ads. The library page also mentions extra data you don't get natively, such as video transcripts and how many times an asset was used, plus filters by format and audience and a place to save the ads you like.

AI ads generator and cloner

You can generate ads from scratch or take a competitor ad you like and have AI rework it in your own brand. Creative analytics (performance insights across platforms) is listed on the Multiple Projects plan but marked as coming soon.

Supported platforms, from Meta ads to Google ads

The pricing page says every plan can connect Meta, Google, TikTok, Reddit, LinkedIn, X and Microsoft. Most public customer reviews talk about Meta ads, so that looks like the most used part of the tool right now.

AdKit pricing for each business

There are two plans, both with every tool and MCP access. The difference is how many brands (projects) you can run.

There is a 7-day free trial, and no card is charged until the trial ends. Billing is by project rather than per seat or per account, and every plan includes unlimited team members. Paying yearly saves 30% or more, according to the pricing FAQ. The pricing page showed higher crossed-out prices ($49 and $97) next to the current ones, so it is worth checking the live price before you commit. There is no free plan.

What customers say right now

On Trustpilot, the tool had a 4.6 TrustScore on the day I checked, from a small set of reviews (18 were shown on its page). The repeated themes are speed (building Meta campaigns from Claude rather than in Ads Manager) and fast support from the founder over live chat. One reviewer says they came from Pipeboard and saved money. Another says the product has occasional small bugs that the team fixes quickly, and that they had only used it for Meta so far. One review describes a Meta-side issue that needed a workaround until Meta shipped an update. That is a small sample, and many of them praise the founder personally, so I would treat them as a sign of good early support rather than proof of long-term reliability.

Limits and drawbacks

• You need your own ad accounts. The vendor says it cannot create them, because each platform has to verify you and your payment method.
• If you don't already work with an AI agent, much of the value goes away and you are mostly paying for an ad library.
• The Single Project plan connects one account per platform, so agencies will need the $49 plan.
• Creative analytics is not live yet, so performance reporting still happens in the agent or in the native ads managers.
• Draft-first approval reduces risk, but someone still has to review budgets, targeting and copy before approving. An agent that drafts quickly can also draft mistakes quickly.

Alternatives for running ads with an agent

• Pipeboard: another way to connect AI agents to Meta ads, and the tool one reviewer says they moved from.
• Meta Ad Library: Meta's free public library for looking up an advertiser's ads, without tracking or alerts.
• Meta Ads Manager and Google Ads directly: no extra cost, but you do every campaign change by hand.

Verdict

AdKit is one of the more careful takes on letting an AI agent manage ads. The draft-first approval, official API access and flat per-project pricing address the obvious worries, and the competitor research adds value on top of the MCP. I would pick it if you already use Claude or another agent every day and run Meta or Google ads for one or more brands. If you don't use an AI agent, or you only need a swipe file, there are simpler and cheaper options.