Dominic Barrios of Bots at Work has a customer-facing Offer Ladder workflow. Given an initial input, several AI agents generate low-, mid-, and high-ticket offers. A person can choose one tier within the workflow, which then generates a marketing roadmap (customer case study[1]). The person has not been removed; their task has changed from devising low-, mid-, and high-ticket offers to choosing a tier. On the same page, he also says MindPal saves him 400 hours a week. The measurement method is undisclosed, so I do not take that as a finding.
That workflow runs on MindPal. As of 2026-11-18, its website is direct: a no-code platform for coaches, consultants, agencies, and course creators, turning their existing documents, courses, and SOPs into AI agents and multi-agent workflows, then delivering them to customers through links, embedding, or their own domains (MindPal website[2]). The process has three steps: upload existing content, build agents on top of it and connect them into workflows, then publish. What it sells is not a model, but “your methodology can now run itself.”
Image: From the official product demonstration[2], obtained on 2026-11-18; official presentation retained as shown, for product illustration only.
Interestingly, the same page’s FAQ supplies its own limits. It first says a workflow can connect multiple agents and other steps, adding branches, loops, and human input as needed, with both step-by-step and end-to-end testing before publishing. It then acknowledges that model outputs can be wrong even when grounded in your own knowledge. It recommends testing representative cases, stating boundaries, reviewing outputs, and adding human review to consequential decisions. A repeatable workflow does not guarantee correct answers (same source). Together, those passages turn the “24/7 AI employee” into something more like instructions for an intern who needs supervision.
What her previous project taught her
Founder Tham (Sylvia) Nguyen said in an Indie Hackers interview that she and her co-founder began MindPal during their university summer vacation in August 2023. The first version was a single-page app built over a weekend: upload any file and chat with it, with a payment gateway attached (interview[3]).
They could build it so quickly partly because it was simple, and partly because of the mindset and technical knowledge gained from a previous failed project. The two had spent nearly 1.5 years on another startup. Her explanation for its failure was direct: they were completely disconnected from potential customers. She summed up the lesson plainly in the interview: without customers to serve, nothing else matters. Their second approach was to finish over a weekend and launch directly on Product Hunt. She credited that launch for roughly 100 paying customers over the following 2–3 weeks. They subsequently stopped attending university to work on MindPal full-time. A few months later, a US VC wrote a small pre-seed check, which she said mainly funded subsequent infrastructure and operations (same source). Those are the founder’s own causal accounts, not independently verified in the interview.
The product line did change: from “upload any file and chat” to a no-code platform for business owners, entrepreneurs, marketers, creators, coaches, and consultants to build agents and multi-agent systems and automate business processes (same source).
There is no fixed price for a run
The self-service page has four tiers. Three representative ones: Pro is $49/month on monthly billing or a $39/month equivalent billed annually ($468/year); Advanced is $179/month monthly or $149/month annually ($1788/year); Ultra is $449/month monthly or $374/month annually ($4488/year), making it the most expensive self-service tier. There is also Free at $0. Prices are in USD before tax; annual billing collects the full year upfront (pricing page[4]). As of 2026-11-18, the homepage displayed Pro at $15.60/month, an annual-billing equivalent after applying BLACKFRIDAY2026. The page did not state when that discount expires. The prices used here are the regular ones in the pricing documentation.
Credits are what really determine the bill. Paid plans include monthly AI credits, issued monthly even on annual billing. Unused credits do not roll over. How many credits a workflow run consumes is not fixed: it depends on the model chosen, tools called, and steps executed (same source). The “smarter” a multi-agent workflow and the more tools it calls, the harder its bill is to estimate. Additional credits can be bought: Stripe credit packs cost $9 per 1,000 credits, purchased once without expiry. Older Lemon Squeezy credit top-ups are recurring subscriptions (same source).
Bringing your own API key (BYOK) does not solve that uncertainty, and is available only with an active paid plan. You pay the model provider directly, but platform limits do not disappear, and built-in tools and features may still consume MindPal credits (same source). BYOK changes who receives payment for model calls; it does not guarantee lower model-call costs. Platform limits and possible credit consumption remain.
$500 once, or $1,499 to $2,000 a month
Another business line says more about the shape of the company: MindPal Managed. Bought per project, it costs a one-time $500 for a complete automation, delivered in 1–3 business days with a week of revisions after delivery. That tier also requires a separate platform subscription to run the automation (Managed page[5]). The monthly version costs $2,000/month for unlimited automation builds, one at a time, with the platform subscription included. The page displays an annual option and advertises 3 months free (same source).
But the documentation site gives the same service a different price: $1,499/month, including the platform’s highest TEAM Plan (MindPal Managed documentation[6]). Neither page states an effective date. The mindpal.io Managed page describes the audiences—per-project for those starting automation, monthly for those wanting an ongoing pipeline—but the public pages do not explain which monthly figure, $2,000 or $1,499, is the current checkout price. What can be established is that both sales formats exist: a one-off project and an ongoing monthly relationship.
Both share a promise: a full refund if a working AI automation is not delivered within 14 days, no setup fees, and cancellation at any time (mindpal.io page). The page does not say when the 14 days begin or who determines what “working” means, and actual refund execution cannot be verified.
The delivery boundaries are candid. The team supports popular tools such as Zapier, Make, and MCP. It does not support internal systems it cannot access or custom features outside the core automation (same source). Maintenance is also its job: every workflow is tested before delivery, but edge cases appear in real use, and the team fixes breakages (same source). Customers own what is built and can continue to use, modify, and run it after canceling or returning to a self-service plan (same source).
A company selling no-code, self-service building also publishes prices for doing it for you. That is not necessarily a contradiction; at least it shows the company has made implementation a separate service. Structurally, these are different purchases: self-service buys the ability to build it yourself; Managed buys someone building the workflow, getting it working, and maintaining it.
For the same task—putting a small team’s knowledge into an agent, then connecting it into a multistep workflow—Relevance AI’s official documentation also provides a self-service path. Upload files, sync knowledge from Google Drive, SharePoint, or Notion, equip agents with tools, connect multiple agents on a visual canvas, and configure escalations, approvals, and triggers before deployment (Relevance AI documentation[7]). Both self-service paths still require users to build, with human checks available at key points. MindPal has an additional, explicitly priced route: buy Managed and have its team implement and maintain it. This establishes where their service entry points sit, not which runs more smoothly.
The revenue line
The founder reported ARR of $500k when interviewed. That is a historical figure, not live 2026 ARR (interview[3]). The interview does not disclose the reporting cutoff, MRR, profit or cost structure. ARR is not profit; the public material does not allow us to calculate whether this company makes money.
MindPal’s most honest material is not the homepage hero copy but the FAQ and price list. It sells “AI employees that never rest” while spelling out that models can be wrong and consequential decisions need human review. It sells self-service building while making its most expensive offering a human-delivered service. For now, the highest-priced option is still a team building and maintaining workflows for customers, with software behind them.