Runable AI is a general-purpose agent that builds the thing and then tries to grow it. It makes websites, decks, videos and apps, and it also runs paid campaigns, cold email, direct messages and phone calls. The build half is easy to judge. The growth half is where a solo marketer needs to be careful, and the reason is not capability.
Founders Umesh Kumar and Saksham Sarda started the company in 2025 and raised a $21 million Series A in August with a team of fifteen. Kumar’s framing of the problem is the clearest statement of what the product is for: “Software creation got automated. Everything after it didn’t.”
The build side covers websites, mobile and internal apps, pitch decks, slides, images, video, reports, spreadsheets, market research and prospect lists, with databases, authentication, payments and deployment attached. It connects to more than 3,000 apps. Three modes exist: Agent, Chat and Plan, and Plan Mode produces a build plan for your approval before it touches code.
The growth side is longer: paid campaigns on ChatGPT Ads, Meta, Google, LinkedIn and TikTok, social scheduling, cold email, direct messages, voice calls through a phone number, competitor monitoring, brand sentiment, SEO and answer engine optimisation, and customer support. Apps exist for iPhone, iPad, macOS and Android, with Windows listed as coming soon.
The plan ladder is simple. Free gives 1,500 credits a day, capped at 15,000 a month. Basic is $5 a month or $60 a year for 4,000 monthly credits. Pro is $20 a month or $240 a year for 25,000. Max is $100 a month or $1,200 a year for 150,000. Paid plans keep the 1,500 daily credits on top of the monthly pool, and unused monthly credits do not roll over.
What is missing is the conversion rate. The credit model charges every AI action and scales with intensity: video is billed per second, video with AI-generated audio costs roughly twice as much as muted video, and Chat Mode is cheaper than Agent Mode. What it does not say is what one website, one deck or one campaign costs. So the honest answer to “will 25,000 credits cover my month” is that nobody can tell you, including the company, until you have run your own work through it for a few weeks.
Set that against how you already buy AI for content work: a per-seat subscription is predictable and a credit pool is not, which is a budgeting difference rather than a quality one.
The dividing line is not how good the agent is. It is whether you can take the action back. A deck, a website draft or a set of prospect research is reversible: it sits there until you look at it, and a bad one costs you the time to read it. A cold email is sent. An ad campaign is spending. A voice call to a customer has happened.
This is the same question the platforms raised when agentic advertising arrived on Meta, TikTok and X in early October, and the answer there applies here. Before you hand Runable AI, or any other agent, a budget or your sending domain, find out what the spending cap is, what the approval step is and whether there is a rollback. Runable publishes a great deal about what its growth features do and very little about those three things.
There is a second problem specific to outbound. Cold email and direct messages at volume put your domain reputation and platform accounts at risk, and those are shared assets. An agent that sends well for three weeks and badly for one can cost more than it saved.
The homepage carries results like bookings up 35 percent in the first month, support requests down 50 percent in two weeks, and podcast edit time falling from six hours to 40 minutes an episode. None of them carries a methodology, a sample size or a date. Treat them as illustrations of what the company wants to be known for, not as a forecast.
The traction numbers deserve the same handling. Zero to $2 million in annual recurring revenue in three weeks is the company’s own statement. Registered users were about 1.5 million in late August and more than 1.7 million by the end of September. Registered is not active, and neither figure has been audited.
One detail is worth checking yourself before you decide on model quality. On 14 October the pricing page listed its available models as including Claude Opus 5 and Claude Sonnet 5.0, while Anthropic shipped Opus 5.5 on 22 September and Sonnet 5.5 on 28 September. Model lists on aggregator platforms lag, which is normal, but it means the model names in the marketing are not a reliable guide to what you get. The same churn hit the assistants directly when Google changed its free plan this month.
The free tier is genuinely usable for this, which makes the evaluation cheap. You can try Runable here and work through the following in one sitting. To see the interface before you sign up, there is a walkthrough of the 2.0 release below, and more on the company’s own YouTube channel. Treat the rest of the video coverage carefully: Runable runs an affiliate programme, so a share of the walkthroughs and reviews are made by people earning a commission on sign-ups, and the description is where you find out.
There is a real free tier with 1,500 credits a day, capped at 15,000 a month, and it covers the main build features. The limits that bite are the monthly cap and the “Made with Runable” badge on free websites. Paid plans start at $5 a month and buy a larger credit pool rather than extra features.
Not on the evidence published so far. It covers an unusually wide range of jobs for one product, but breadth across building, advertising, outbound and support is a claim about scope, not about quality in any one of them. Test it against the two or three jobs you most want off your plate rather than switching wholesale.
Monthly credits do not roll over, and they cannot be transferred between accounts. The daily allowance resets daily. That makes an uneven workload more expensive than it looks, because a quiet month subsidises nothing and a heavy month can run you out.
Not until you know the spending cap, the approval step and the rollback, and those are not published. Ads spend real money and outbound puts your domain and account reputation at stake. Both are reasonable to automate eventually and unreasonable to automate on day one.
Runable AI is a serious product with real funding and a real free tier, and the build half is worth an afternoon of anyone’s time. The growth half carries the bigger downside, because those actions cannot be taken back and the controls around them are not published. Test the reversible things first, measure what a task actually costs you in credits, and keep your ad accounts and sending domain out of it until the rest has earned the trust. Tools like this are part of how a small team can compete with bigger brands, as long as the first automated mistake is a cheap one.
This is the kind of thing we work on with clients at Social Lady. If you want a second opinion on whether an agent platform belongs in your stack, get in touch.
