Mark by Airtop - Vibe automation for solo marketers
by•
Give Mark your website and it researches your business, creates a personalized GTM plan, and builds web agents that automate lead gen, enrichment, outbound, SEO, and Google Ads campaigns. Using Mark is like vibe coding, but for sales and marketing campaigns. Built on Airtop's Agent Builder platform, Mark compiles every automation into deterministic code, so its agents run reliably and 10-100X cheaper than LLM-per-step agents. Real marketing automations, built just by typing.
Replies
Interesting idea. Does Mark let you adjust targeting criteria before the agents actually go live?
Airtop
@dhiraj_patel5 yes, of course, every agent is built exactly for your needs. It allows you to take your "craziest" marketing ideas and turn them into repeatable workflows
Airtop
Hey folks! I'm Jordan from Airtop. Feel free to ask me questions about Mark or how the Airtop platform builds compiled agents <3
Looks cool guys, great idea. What are your policies on data handling and encryption, data privacy and training? Is data encrypted at rest and in transit? Would like to review your privacy and data handling procedures.
Airtop
@adam_lavine Check our trust center - https://trust.airtop.ai/
How does it handle sites with heavy bot protection or CAPTCHAs, like LinkedIn or banking portals? Curious if that breaks the "just describe it" promise pretty quickly.
Airtop
@akpaknihat66966 Good question, bot detection is very hard to overcome for all web automation players in the game. At Airtop, we equip the agent with proxies, persisten session storage and captcha solver to mitigate anti-bot measures which really works on many websites.
The compile-agents-into-executable-code approach instead of keeping an LLM in the loop is what makes the reliability and cost claim believable. My setup question is about the accounts Mark logs into: when he posts to social or pulls LinkedIn data, where do those sessions/credentials actually live, on Airtop's hosted browsers server-side, or something scoped per user? And when a site changes its flow and a compiled agent breaks, does Mark detect the failure and auto-recompile, or do I have to trigger a rebuild?
Airtop
@noctis06 We do provide a Vault system as part of Airtop, where you can securely store your credentials to be used on the sites.
Each agent also stores the browser profiles, so that subsequent runs do not require to login each time.
In particular, the Vault and browser profiles are encrypted for your or your team's use only.
If the agent breaks, for any reason, it will notify you, and you will be able to review the failure and adapt the agent accordingly.
Now, depending on the failure, we do provide a self-healing mechanism for some of the actions the agent does, but websites vary in quality and potential changes, so there will be scenarios where you will need to adapt the agent (trigger a rebuild) to correct the failure.
Congrats on the 4th launch :) The line I keep circling is "compiles every automation into deterministic code", it's the real bet, and where I'd push.
Compiled code runs cheap and reliable, agreed, but it freezes an assumption about a page you don't control. LinkedIn reshapes its DOM constantly. An LLM agent re-reads the page and adapts; a compiled one keeps doing exactly what it was compiled to do, right until a redesign makes it fill the wrong field with no error thrown.
So: when a site changes under a compiled agent, how does Mark notice (silent no-ops don't crash), and does it recompile itself or page a human? The happy path isn't the test, the silent drift is.
Congrats again!
Airtop
@keirodev Agents use airtop actions to interact with the web page.
Airtop actions have a degree of self-healing, and they involve a bit of AI to adapt to changes, so we do have room to handle variability up to a point.
If the site change is a bit too drastic, you will need to rebuild the agent.
In the case the agent breaks, we will notify you and provide you with the tools to rebuild the agent, using the failure information to better guide the rebuilt process.
@kibvindas Good to know the self-healing lives inside the actions themselves rather than an LLM re-read per run, that's actually the missing piece of the "deterministic and cheap" pitch. My real worry was never the loud break, it's the LinkedIn field that gets filled with the wrong value and still returns success with no error. Is detection there coming from a schema/assertion check on the output, or does it rely on downstream signals like a failed submit or a human noticing? That's the gap between "notices" and "silently wrong."
The "chatbots that ideate then leave you with all the work" line is exactly why I stopped trying to use AI for marketing planning early on. Getting a strategy back that reads well but has no path to execution is worse than having no help at all, because now I feel obligated to try it before admitting it was theater.
The compile-to-deterministic-code angle is where this gets interesting. Most agent frameworks I've seen have the same flaw as the strategy chatbots, they generate a plan every run, which means the plan drifts even when the task hasn't. Compiling once and re-running the compiled artifact is the pattern I keep wishing existed. Curious how you handle the case where the underlying page/site changes and the compiled agent breaks, do you re-compile on failure, or does Mark flag it and ask?
Also honest question, for someone running solo who already has a rough GTM shape they trust, does Mark benefit them, or is the biggest value for people who need help figuring out the plan itself? Trying to figure out if I'd get more value from the planning side or the execution side.
Airtop
@elias_motionfy Regarding changes in the underlying page/site - the agent detects those changes and will notify you, it is then easy to re-compile it.
I see more value in the execution phase, coming with the GTM plan or strategy is something you can do with Claude, actually automating your GTM in a reliable and cost-effective way is is where Airtop shines.
@amir_ashkenazi @amir_ashkenazi Genuine of you to answer it that way, most makers would have said "both" and hoped I signed up for the wrong reason. The Claude-for-planning, Airtop-for-execution split is the honest positioning, and honestly the more useful one for anyone already in build mode.
The change-detection-then-notify flow makes sense. My follow-up worry would be silent breaks, cases where the underlying page technically renders the same but a downstream field or button ID changed, so the agent runs to completion but produces subtly wrong output. That's the drift class I've seen wreck other automation setups: the loud failures are easy, the silent successes-that-aren't are the ones that erode trust. Does Airtop's compiled agent detect that class, or is it mostly guarded against the visible-break cases?
How does it handle sites with strict bot detection or CAPTCHAs when the agent tries to log in and grab data?
Airtop
@fahri1585951 The honest answer - works on most, does not work on some.
How does it handle sites with heavy bot protection or CAPTCHAs that block most automation tools?
Airtop
@mitmarazrqe That's the core of what Airtop's execution layer handles. Agents run in a managed cloud browser fleet using real browser environments with genuine fingerprints and residential IPs, plus built-in handling for auth, 2FA, and CAPTCHAs. The idea isn't to trick anti-bot systems, it's to actually run as a real browser session so the traffic is legitimate rather than a headless signature that gets flagged.
It's not magic, and heavily fortified sites are always a moving target, but that reliability layer is exactly what we've built and where most of the engineering goes.
How does it handle sites that require two-factor authentication or have aggressive bot detection? Curious how reliable it is in practice on those kinds of tricky logins.
Airtop
@glcanycesupuww Regarding two-factor authentication, it is in our roadmap, but not yet implemented.
Agents have access to a vault where you can safely store your credentials and a browser profile, so that the agent doesn't have to login into the site on every run.
Regarding bot detection, we provide features to minimize it as much as possible, but bot detection is always a moving target.
Our agents use real browsers (not stateless browsers)
Browser actions mimic human actions, trying to do real clicking, typing, etc
We offer automatic captcha solution and a residential proxies.
In case you have any other questions, let us know.