Banu launched SiteGPT in March 2023 after a two-week build sprint. He already had 100 customers from his prior SaaS product, Feather (a Notion-based blogging tool). He initially conceived SiteGPT as a tool for those existing Feather customers — everyone had a blog, so a chatbot that could answer questions about their blog content was a natural fit. He then decided to spin it out as a standalone product rather than limiting it to Feather's user base. The launch itself drove enormous early traction — he reached approximately 10,000 MRR within the first month. The transcript does not specify the exact launch channel (e.g., Product Hunt, Twitter, etc.), but the rapid uptake suggests his existing audience from Feather and/or social media presence played a significant role. This early success was so overwhelming that he sold Feather for $250,000 to focus entirely on SiteGPT.
SiteGPT
AI chatbot builder that answers questions using your website's content
7 moves, in order
- Pre launch / IdeaExisting customer base
Identified SiteGPT's core use case by asking what he could build for his 100 existing Feather customers. Everyone had a blog, so he built a chatbot that could answer questions about blog content — then broadened it into a standalone product.
Product scoped and validated against a known customer segment before writing a line of code - Launch (March 2023)Product launch organic
Went from idea to launched product in two weeks. Leveraged existing audience from Feather SaaS and social presence to drive initial signups.
~$10,000 MRR within the first month of launchMRR $10k - Early GrowthAsset sale to double down
Sold prior SaaS product Feather for $250,000 to free up 100% of his time to focus on growing SiteGPT, which had already exceeded Feather's MRR.
$250,000 exit; full focus redirected to SiteGPT - Growth — SEO Free Tools StrategySEO — free tools
Built ~50 free AI micro-tools (e.g., PDF-to-Markdown converter, chatbot name generator) each targeting a specific low-competition keyword. Each tool has a contextual CTA linking back to SiteGPT. Tools are built with AI (Cursor/Claude) in under 5 minutes by prompting off existing tool templates.
~1 million total website visitors; 50,000 clicks/month from Google; ~90% of Google traffic from free toolsMRR $13k Users 130 users - Growth — Keyword Research ProcessSEO keyword research
Uses Ahrefs Keywords Explorer with a blank search to surface all keywords, then filters by: (1) relevant include keyword (e.g., 'AI', 'generator'), (2) keyword difficulty under 10, (3) minimum 1,000 monthly search volume. Lists results in Notion and prioritizes by high volume + low KD + easy to build + high relevance to SiteGPT.
Systematic pipeline of rankable tool ideas; avoids wasting build effort on high-competition terms - Growth — CTA Design per ToolConversion optimization
For every free tool, designs a specific CTA that bridges the tool's use case to SiteGPT's value prop. Example: a 'chat with PDF' tool CTA reads 'You've tried chatting with one PDF — what if you could create a chatbot for all your business content? Try SiteGPT.'
~200 leads/month from 50,000 visitors; 60 trial conversions/month; 25–40% trial-to-paid rateMRR $13k Users 130 users - Ongoing — Tool Scaling with AIAi assisted building
Uses Cursor/Claude Code to spin up new free tools in under 5 minutes by referencing existing tools as templates and providing the target keyword. This keeps the marginal cost of each new tool near zero.
50 free tools built to date; ongoing expansion of SEO surface area with minimal time investment
Banu had an existing base of 100 paying customers from his prior SaaS (Feather) who became the seed audience for SiteGPT. He also had prior experience scaling a SaaS to $6K MRR and selling it, giving him product-building and monetization know-how that most first-time founders lack.
seo_free_tools (engineering as marketing — ~50 free AI tools built to rank for low-difficulty, high-volume keywords)
Banu explicitly avoided paid marketing entirely ($0 spend). He does not call out any specific failed channel, but the implied contrast is that traditional marketing tactics (cold email, content writing, paid ads) were not used — by design, because he disliked them.