The LinkedIn Ads Bottleneck Isn’t Creative — It’s the Plumbing
If you run social for a B2B brand, you already know the dirty secret of the creator economy’s least glamorous channel: LinkedIn Ads is where good content strategy goes to die in a spreadsheet. You spend a week building a campaign brief, another three days waiting on design, then two full afternoons wrestling Campaign Manager’s targeting UI, and by the time the ad set is live, the offer has changed. The creative was never the hard part. The plumbing was. That’s the gap ZenABM is trying to wedge itself into with its second Product Hunt launch — an MCP server and AI agent called Zena that writes, builds, uploads, and optimizes LinkedIn campaigns from inside Claude, ChatGPT, or Perplexity. Whether you’re a solo founder running your own demand gen or a social lead at a SaaS company, the mechanics here are worth understanding, because the “AI that reads your ads” era is ending and the “AI that writes into your ad account” era is starting.
What Zena Actually Does (And Why the “Write” Part Matters)
Let me be precise about the claim, because the marketing language around AI ad tools is usually mush. According to maker Emilia Korczyńska’s launch post, ZenABM started from her own experience as a VP of marketing running a seven-figure ABM program on LinkedIn Ads — a process she describes as “like eating glass,” where each campaign took days or weeks from brief to upload. Her framing of the competitive landscape is the sharpest part of the pitch: most AI marketing tools, she argues, can only read your performance data. They’ll tell you a campaign is underperforming. They won’t build the replacement.
Zena is positioned as the write-side counterpart. From a plain-language brief inside Claude, ChatGPT, or the ZenABM agent itself, it:
- creates image, document, video, and text ads with copy written for you
- builds the campaign, ad set, targeting, bidding, and budget
- pushes everything into Campaign Manager as a draft — nothing goes live without human approval
- then keeps optimizing: flagging and pausing weak ads, adjusting bids, excluding bad-fit companies, and attributing pipeline
- generates AI reports down to the individual company level, framed as prospecting fuel for AEs and SDRs
The technical substrate is an MCP server with, per the maker, “15 expert skills and 96 read & write tools across LinkedIn Ads, ABM and CRM data, trained on knowledge from 30+ LinkedIn Ads experts.” That’s a specific, checkable claim — and the word “write” is doing a lot of work there. My take: the read/write distinction is the entire product thesis, and it’s the right one to bet on. Every B2B operator I know has a dashboard graveyard. Nobody needs another dashboard.
The draft-first design is the trust unlock, not a limitation
The detail that keeps surfacing in the launch comments — from hunter Rohan Chaubey, from reviewer S.S. Rahman, from Amrita P — is that campaigns land in Campaign Manager as drafts. On the surface this sounds like a hedge. In practice it’s the only sane architecture for 2025. LinkedIn’s ad review process is unforgiving, brand safety incidents in B2B are expensive, and no serious marketing org is going to hand an LLM live spend authority on day one. Korczyńska’s own comment nails the posture: “a human in the loop is still a must, in my opinion, but you can still ask Zena to launch the campaigns and ads for you after you have reviewed them.” That’s the correct amount of automation for where the technology actually is.
How It Stacks Up Against the Incumbents You’re Already Paying For
Here’s where I want to be honest, because the launch page is understandably light on head-to-head comparisons and heavy on enthusiasm. Let me do the comparison work the makers didn’t.
Against Factors.AI and Fibbler: These are the two names that come up repeatedly in the comments, and Korczyńska’s own answer is refreshingly non-defensive. She concedes Factors is “a very sophisticated and expensive analytics and revenue attribution tool, primarily,” and positions ZenABM as “more of a budget tool for campaign management and creation, plus analytics and revenue attribution, and AI-first.” A reviewer named Renata Zinnatullina, an agency owner, explicitly says ZenABM is “cheaper and better than most alternatives.” I’d treat that as a customer opinion, not a benchmark — but the directional claim (analytics-first tools are expensive; this is cheaper and does more of the build work) is the one to test against your own stack.
Against the Buffer/Hootsuite/Later tier of scheduling tools: Different sport entirely, and worth saying out loud so nobody wastes a trial. Those tools publish organic posts to Instagram, TikTok, X, LinkedIn, Threads, and Pinterest. They do not touch paid LinkedIn ad accounts, they don’t build ad sets, and they don’t do revenue attribution. If you’re a creator managing organic distribution, Zena is not for you. If you’re running paid B2B acquisition, your scheduling stack is irrelevant here.
Against native LinkedIn Campaign Manager: The honest answer is that Campaign Manager is the system of record and will remain so. Zena is a layer on top of it, not a replacement. The value is entirely in the time between “brief” and “draft sitting in Campaign Manager.”
Why TikTok creators should care more than LinkedIn ones
Counterintuitive, but I’ll defend it. If you’re a TikTok-first creator, you’re probably thinking this whole essay is irrelevant to you. It mostly is — except for the pattern. Zena is the first credible instance I’ve seen of an AI agent that doesn’t just analyze a platform’s ad data but writes into the platform’s ad account via an official-ish API path. That’s the shape of the next two years of creator tooling. The equivalent for TikTok would be an agent that takes a brief, generates the hook variations, builds the Spark Ads campaign, uploads the creative, and reports back on watch time and completion rate — all without you opening Ads Manager. Nobody has shipped that convincingly yet. But the architecture Zena demonstrates (MCP server + read/write tool surface + draft-first approval) is the template. Watch this space, and watch whether Meta and TikTok open the same doors.
Where the math breaks
I want to flag a structural risk that the launch page doesn’t address. MCP-based agents are only as good as the underlying tool schema, and reviewer Salman Anwar Awan — who hooked Zena up via MCP rather than the native connector — makes exactly this point: he wants “more information regarding MCP tool schema documentation” because he’s currently “relying on assumptions regarding certain parameter behaviors.” He also asks how Zena handles budget conflicts between concurrently running campaigns, and why the optimizer pauses a campaign rather than tagging it. Neither question appears to get a direct answer in the thread.
That’s not a knock on the product — it’s a knock on the documentation maturity of every MCP-first tool in 2025. If you’re an operator evaluating this, the schema docs and the optimizer’s decision logic are the two things I’d pressure-test before you trust it with a real budget. An agent that pauses a campaign you didn’t want paused is worse than no agent at all.
What Social Teams Can Borrow From This, Even If They Never Buy It
Three transferable lessons, and I mean these as operating principles, not product endorsements.
1. Separate the read layer from the write layer in your own workflow. Most social teams have a beautiful analytics stack and a chaotic execution process. Zena’s thesis — that the bottleneck is execution, not insight — applies whether or not you use an AI agent. If you’re spending more time building campaigns than deciding what they should say, your process is inverted. Fix the build step first.
2. Draft-first is the correct default for any AI-assisted publishing. Whether it’s an ad campaign, a LinkedIn post, or a TikTok caption, route AI output through a human approval gate before it hits the platform. Korczyńska’s framing — “human in the loop is still a must” — is the right posture, and it’s the posture you should enforce on your own team’s AI usage, not just on vendor tools.
3. Company-level attribution is underrated for organic too. The feature Korczyńska highlights — reporting “down to individual company level” for AE and SDR prospecting — is a paid-ads feature, but the underlying idea is portable. If you’re running organic LinkedIn content for a B2B brand, are you tracking which companies engaged, or just aggregate impressions? The former is a sales pipeline input. The latter is vanity. Most social teams I audit are doing the latter.
The pricing reality check
Plans start at $59/month, with unlimited MCP and Zena access from $159/month, per the launch post. There’s a 30% launch discount code (“PH30”) running through a stated end date, and a free 37-day trial at app.zenabm.com/signup. For a solo founder or small agency, $59 is a trivially cheap experiment. For a mid-market team, $159 with unlimited agent access is cheaper than a single freelance LinkedIn Ads contractor hour in most markets. I’d bet the pricing is deliberately aggressive to land logos, and I’d expect it to move. Not disclosed: whether the unlimited tier has fair-use caps on MCP calls, or what happens to your data when you connect CRM. Ask before you sign.
Where My Judgment Says It Falls Short
I’ll be direct, because the launch page is a launch page and you deserve the counterarguments.
The “96 tools” number is impressive and meaningless without context. Tool count is the new “10x your reach.” What matters is whether the tools cover the specific operations you run — lead gen forms vs. conversation ads vs. document ads, matched audiences, predictive audiences, etc. The maker doesn’t break this down in the source. Ask for the tool inventory before you commit.
The optimizer’s transparency is an open question. Multiple commenters (Anwar Awan, Amrita P) ask variants of the same thing: when Zena pauses an ad or changes a bid, does it explain why? The makers don’t give a clear yes in the thread. For an agency running client accounts, unexplained optimization is a liability. For a solo operator, it’s annoying. Either way, it’s the feature I’d demand next.
It’s LinkedIn-only, which is both the moat and the ceiling. The multichannel attribution page (linked by Korczyńska as zenabm.com/multichannel-attribution) shows Meta and Google ad interactions in company journeys, but the build capability is LinkedIn-only. If your acquisition mix is 60% Meta, 30% Google, 10% LinkedIn, you’re buying a tool that automates your smallest channel. That may still be worth it — LinkedIn is the most labor-intensive to build in — but be honest about the coverage.
Who this is NOT for: organic-first creators, consumer brands with no B2B motion, teams with no LinkedIn Ads spend at all, and anyone who needs the tool to work inside a platform other than LinkedIn on day one. Also — and I mean this kindly — anyone who hasn’t yet learned LinkedIn Ads fundamentals. Zena will happily build you a well-structured campaign against a bad strategy. The tool amplifies your judgment; it doesn’t replace it.
What I’d Watch / Test Next
If you’re an operator curious about this category, here’s what I’d actually do this week — not next quarter.
First, run the free trial against a real but small campaign. The 37-day trial and the $59 entry tier make this a low-stakes test. Build one campaign you’d have built manually, and time both paths. The delta is the entire value proposition.
Second, pressure-test the MCP schema before you connect CRM. Ask the makers directly for the tool schema docs — Anwar Awan’s complaint is a fair one, and their response to it will tell you a lot about how seriously they take operator trust.
Third, watch the optimizer’s decision log for a week. If you can’t see why it paused an ad or changed a bid, don’t give it write access to a live budget. Draft-only until you trust the reasoning.
Fourth, benchmark the company-level reporting against whatever you’re using now. If you’re on Factors or a similar attribution tool, this is the comparison that matters. If you’re on a spreadsheet, this is the upgrade that matters.
Finally, watch the category, not just the product. The MCP-plus-ad-account pattern is going to show up for Meta, TikTok, and Google within 18 months, I’d bet. ZenABM is early. Early is good. Early is also unfinished. Treat it accordingly — as a promising bet on a workflow that’s about to become standard, not a finished replacement for the humans who currently do the work.
