In a September 2024 interview, Daniel Nguyen described a result that did not look much like a growth story: BoltAI's price had gone from $19 all the way up to $79, monthly revenue was slightly lower than at the $39 tier, and the customer count was down to about half. Refunds were down, support inquiries were down, and unsuitable customers were down. His own conclusion was blunt: he was happier. A solo founder also has to optimize for the founder's happiness. [1]
That sentence is worth putting up front because it brings onto the table a cost that is rarely counted in pricing: a solo founder's support time. BoltAI is an AI client on the Mac, and the work it does on people's behalf starts with a copy and paste.
On a Mac, reading a PDF, looking at an error message, or staring at a screenshot a colleague sent follows a standard motion: select, copy, switch to the browser, paste into the chat box, then switch back. BoltAI wants to cut out those middle steps—a global shortcut brings up the window, a screenshot turns directly into a question, scanned PDFs go through OCR to extract text, and dictation and in-place rewriting are completed in the original input field. The company calls it a native Mac AI workspace, one that can switch among OpenAI, Anthropic, Google, Mistral, Azure, and Amazon Bedrock, and can also run local models with Ollama/LM Studio, with a stated 300+ models. [2] · [3]
Image: BoltAI official display[2], accessed 2026-11-14, original interface retained.
By default, it stores chat history locally, and chats are not used for training; prompts are sent directly to the model provider you choose; Cloud Sync is optional, and data is put in the cloud only if it is turned on; API keys can be encrypted with a custom passphrase. [4]
Put plainly, it is not selling a model; it is moving the model next to the action you are doing right now.
Daniel's starting point for this was not a dignified one. In September 2024 he was 35, lived in Ho Chi Minh City, and called himself a solopreneur. His previous product, KTool, had been going for a year and still had not reached ramen profitability; his savings were only enough to last another year; and he had been depressed over KTool's growth. To find freelance work, he went to learn generative AI and the OpenAI API, and while learning he ran into the product idea for BoltAI—a better way to use AI on macOS. BoltAI launched in May 2023, and he says that at first it was mainly to learn SwiftUI and OpenAI app development. By September 2024, it had become his highest-revenue product: 7,000+ customers and 100+ updates. [1]
The channel path was not a straight line either. Early on he mainly built in public on Twitter, sharing new features every week; later he tried AI directories, paid PPC, free tools, and SEO help articles, and attributed one of Pieter Levels's retweets to a viral growth round; he also pulled growth back by chatting with his “most active customers” and adding features in a targeted way. He positioned ShotSolve as a lead magnet (doing marketing the way he does engineering), and as of that interview it had brought BoltAI 1,700 visitors. These channel results are all his own account, with no channel-level conversion data. [1]
As of the September 2024 interview, BoltAI averaged about $15K/month, from two sources—license sales on the official site and Setapp subscriptions—so it was mixed revenue, not pure MRR. After joining Setapp, subscription revenue increased, and the Setapp team also put considerable effort into promoting it; the side effect was that many customers turned around and bought a license directly outside Setapp. This revenue figure did not come with full costs, so no net profit can be calculated from it. [1]
On the pricing page read on November 14, 2026: BFCM2026 promotional prices were Pro $59 (regular $99), Pro+ $119 (regular $199), and Lifetime $239 (regular $399), priced in US dollars, taxes additional, with a 30-day no-reason refund; Pro and Pro+ are one-time licenses that include one year of updates, and Lifetime includes lifetime software updates. [5]
The license terms are stated very plainly: after the one-year update period ends, local features remain available; Cloud Sync and mobile access are suspended and made read-only for up to 24 months, after which the cloud copy is deleted and local data is unaffected; renewals are optional, manual by default, at 60% of the original undiscounted price, and auto-renewal is off by default. [6]
These terms effectively write “not renewing” as a normal path rather than a failure state. A one-time buyout, bring-your-own API key, and chat history staying on your own machine—stacked together, these three things sell “this thing belongs to me,” not “I am continuously being served.” But what “belongs to me” means is the software and local data: BoltAI is BYOK by default, and beyond the license fee, calls to cloud models are still paid by the user to the model provider; Cloud Sync also does not include a model usage allowance.
This buyout-plus-BYOK model has buyers elsewhere too: Tony Dinh of TypingMind says that from launch through the Product Hunt release he had cumulative revenue of $22K, with 99% of sales coming from his existing Twitter audience. [7] · Tony Dinh[8] BoltAI's license terms, by contrast, spell out the post-update-period state in more detail—local functions keep working, cloud sync pauses, and renewal is optional.
The pricing page also says BoltAI has been on the market for 3+ years, has released 220+ public builds since 2023, has 20,000+ users, and has 0 investors and is self-funded. The 20,000+ figure is users, not paying customers. [5]
Back to that $79. Price increases are usually told as “filtering for high-value customers”; the bill Daniel presents is more plain: customers halved, revenue fell slightly, and trouble fell by a good chunk. The interview did not say why the people who left did so; it only gave the result—refunds, support inquiries, and unsuitable customers were all down. Whether this trade was worth it depends on whether you count the founder's energy as a cost or not.