Cesurhan Uygun

Awesome to see someone tackling the commission problem for restaurants via WhatsApp. We're building something adjacent at TalkBuildr, AI chatbots that restaurants embed on their website for reservations and menu questions. How do you handle multi-language menus? Our restaurant clients in Amsterdam keep asking for that.

Great angle targeting trades specifically. We're building something similar for web agencies (AI chatbots they can white-label for their clients) and the biggest challenge has been making the AI sound like it actually knows the business, not just generic responses. How are you handling the knowledge training part, do clients upload docs or is it more structured setup?

The 'dead landing page' problem is real, we dealt with the same thing building TalkBuildr. Our chat widget is also a single script tag embed so the distribution model is similar. I would like to know how you handle the cold start, what do the notifications show before you have any real signups?

Smart angle, turning audits into client-ready pitches instead of just reports. We build TalkBuildr for the same audience (agencies) but on the support side, AI chatbots they deploy on client sites. Would like to know how agencies respond when the pitch is AI-generated vs hand-built, do they care or just want the close faster?

The commission math is brutal, $3.78M/year for a 50-location brand is insane. We see the same thing from the support side building TalkBuildr, restaurants spending staff time answering the same menu and hours questions that a chatbot could handle for €24/month. How are you handling the customer support layer once orders come through your platform, do restaurants still field calls directly?

Next.js + Supabase gang. We built TalkBuildr on the same stack for AI chatbots as a service. The client portal being included on every plan is smart, we learned the hard way that agencies will pay for the tool but churn if the client-facing piece costs extra. About the AI layer you mentioned building last, what made you sequence it that way?

Really cool to see this, we ran into the exact same scaling problem building TalkBuildr (AI chatbots as a service for agencies). The white-label portal + billing combo is where it gets tricky. How are you handling the credential handoff when clients need to swap their own API keys mid-deployment?

Planning by vibe is a fun angle. I'm working with a travel community that wants something similar but powered by local insider knowledge instead of generic AI. Where does your travel data come from, scraped reviews or something more curated?

The voice-first approach is smart, especially the part about tradespeople with gloves on. I'm building an AI chatbot platform for freelancers and agencies and we ran into the same insight, these folks don't have time for complicated UIs. How do you handle the AI accuracy when someone speaks in mixed English/Spanish mid-sentence?

Building in the same space (AI chatbots connected to custom knowledge bases). Curious about your embedding approach, do you support multi-site deployment where the same agent lives on multiple client websites? That's been one of the trickier parts for us to get right.

The 'upload SOPs and get instant answers' flow is basically what we built for website chatbots, same RAG pattern but different vertical. How do you handle conflicting info across different manuals, like when an old SOP says one thing and a newer revision says another?

The no-rigid-scripts angle is smart, most widget builders force you into decision trees that break the moment someone asks something unexpected. how are you handling knowledge base freshness when the site content changes?

Interesting angle on AIO. we build embeddable AI chatbots for websites (knowledge base backed) and the 'how do you show up in AI conversations' problem is something we think about too. Have. you seen any data on whether users actually click these widgets vs just going straight to ChatGPT themselves?

This is cool, the single-link flow is exactly what freelancers need. we're building something adjacent (AI chatbots that freelancers/agencies embed on client sites) and the booking friction you described is real. Do you see agencies using this for their own clients too, or mostly solo creatives?

Cool idea, the multi-language auto-detect is smart. I'm building something similar for website chatbots (AI that answers from a knowledge base) and the tone-matching piece is tricky to get right. How are you handling cases where a negative review is actually constructive feedback vs just someone venting? Feels like those need very different reply styles.

Cool concept, we build tools for the same audience (freelancers and agencies). The revision rounds with deadlines feature is smart, that's usually where feedback loops break down in agency work. Do clients actually use the pinning or do most of them still just send screenshots over email?

Love the agency-first angle here. We're building in a similar space (AI chatbots that agencies white-label for restaurant clients) and the OpenTable lock-in problem is real. How do you handle onboarding for restaurant owners who aren't super technical, do they self-serve or does the agency set everything up?

Nice to see someone tackling the restaurant ops mess. We've been building AI chatbots for restaurants too (different angle, more on the customer-facing Q&A side) and the multi-language piece is brutal to get right. How are you handling menu item translations, are they AI-generated or does the restaurant provide them?

Great approach to the lead discovery problem. We're building TalkBuildr (AI chatbots for agencies) and the 'hunting vs building' split you describe is exactly what our users deal with too. How do you handle lead quality scoring, do clients set filters upfront or does the AI learn from which leads they actually pursue?

Really cool to see RAG done right on a niche vertical like fragrance. Building something similar (AI chatbots for business verticals) and the 120k+ fragrance database is impressive. How are you handling hallucination when users ask about fragrances outside your dataset?