May 2, 2026 · 1h 0m · latent-space
⚡️ Competing with ChatGPT and Sierra, building a $10M ARR company — Yasser Elsaid, Founder, Chatbase
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Chatbase founder Yasser Elsaid shares how he bootstrapped an early RAG side project into a $10M ARR AI customer support platform, detailing his playbooks for harness engineering, product-led enterprise growth, and competing at scale without venture funding.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The hosts as informed peer 4.4 Guest teaching 4.4 Guest disagreement 1.5 The hosts pushing back 1.9
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 15:13 Dismissing zero-cost model switching claims
Yasser forcefully refutes the common market assumption that switching LLMs is frictionless, explaining that 95% of performance is tied to custom harness scaffolds.
Hardest push from the hosts ▶ 41:50 Host pushes back on outcome-based pricing skepticismAlessio explicitly rejects Yasser's skepticism of outcome-based pricing, pointing out that enterprise call centers already budget on per-resolution metrics.
Biggest teaching moment ▶ 58:05 Explaining real equity value in bootstrapped vs VC-backed startupsYasser educates the hosts on recruiting dynamics, breaking down why profitable bootstrap equity is less risky than options tied to inflated VC valuation hurdles.
The host holds their own ▶ 37:47 Host dissects AEO as repackaged SEOAlessio brings industry domain knowledge to cut through marketing buzzwords, noting that AI Engine Optimization fundamentally relies on the same retrieval mechanisms as SEO.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Launching the Prototype and Early Guardrails | 4 | 3 | 1 | 1 | The hosts reference early LangChain, Pinecone, and guardrail limitations, prompting Yasser to explain how pre-processing and system instructions addressed hallucination in early prototypes. | |
| Leveraging LinkedIn Virality for B2B Growth | 4 | 4 | 1 | 1 | Host brings up influencer playbooks like TikTok for consumer apps, but Yasser explains how organic LinkedIn influencer posts drove immediate B2B SaaS conversion and Stripe MRR spikes. | |
| The Decision to Bootstrap and Early Scaling Pains | 3 | 4 | 1 | 1 | The hosts explore Yasser's decision not to raise venture capital despite explosive growth, with Yasser sharing operational anecdotes about maxing out personal credit cards on OpenAI API fees. | |
| Token Volume and LLM Provider Distribution | 5 | 4 | 1 | 1 | Hosts ask specific technical questions about token throughput across providers, prompting Yasser to break down Chatbase's model market share between OpenAI, Anthropic, and Google. | |
| Harness Engineering and Model Switching Costs | 5 | 6 | 2 | 1 | Yasser dispels the notion that model switching costs are zero, explaining that 95% of customer support performance relies on custom harness scaffolds rather than raw foundation model intelligence. | |
| Product-Led Growth and Moving Upmarket | 4 | 4 | 1 | 1 | Hosts examine the self-serve pricing strategy, while Yasser explains how starting with a strict PLG model forced intuitive product design that now benefits enterprise upmarket moves. | |
| Surprising Enterprise Use Cases and Global Customer Stories | 3 | 5 | 1 | 1 | Yasser shares how organic self-serve adoption uncovered unexpected customers like Chuck E. Cheese, theme parks, and the UN handling crisis communication in war zones. | |
| The B2B Go-To-Market Manifesto | 4 | 5 | 2 | 1 | Yasser outlines his B2B go-to-market manifesto, advising founders to avoid consumer marketing distractions like TikTok UGC and instead focus on foundational B2B demos, outbound, and narrative content. | |
| Employee-Generated Content and Scaling Conviction | 4 | 4 | 2 | 2 | Host questions whether employees without social media habits can be forced into employee-generated content, with Yasser outlining how team culture and conviction drive posting rhythms. | |
| Warm Outbound and AI Engine Optimization (AEO) | 6 | 4 | 2 | 4 | Host challenges the novelty of AI Engine Optimization (AEO), arguing cynically that it is identical to traditional SEO, while the co-host and Yasser discuss prompt injection and LLM retrieval nuances. | |
| Evaluating Outcome-Based vs. Usage-Based Pricing | 6 | 5 | 3 | 6 | Host directly challenges Yasser's skepticism toward outcome-based pricing by contrasting self-serve startups with legacy call centers served by competitors like Sierra. | |
| Roadmap: Chief Customer Officer and Replacing Zendesk | 5 | 4 | 1 | 2 | Host explores positioning risks with legacy incumbents like Zendesk, and Yasser explains how Chatbase acts as a transitional AI layer before fully displacing legacy support suites. | |
| Relocating to San Francisco and Operating at VC Scale | 5 | 4 | 2 | 3 | Host contrasts Yasser's relocation to SF with the typical nomad lifestyle of bootstrapped founders, prompting Yasser to explain why he operates his profitable company with VC-scale aggressiveness. | |
| Hiring for Outcomes and Modern AI Development Tools | 4 | 5 | 1 | 2 | Yasser explains why bootstrapped startup equity is often more meaningful than bloated VC valuations, while sharing his developer workflow across Claude Code and OpenAI Codex. |
Statements from this episode (26)
No commercial products augmented GPT models with custom data pre-ChatGPT
“Like I saw some people doing demos, but like in like a CLI or something like that, but there was no product doing like this model, but with additional data on top of it.”
Building viable AI startups still requires traditional sales and user calls
“I think a lot of people were saying, oh, like, AI is going to change how you build a business, but I think what helped me actually, like, make this a business is I did all the, you know, like, the normal business stuff, you know, like, talking to users, gettin…”
Chatbase scaled from zero to $1 million ARR in 117 days
“I think doing that helped a lot into, Getting me from like zero, basically to one million AR in a 117 days.”
Single LinkedIn posts from large AI pages drove $3,000 MRR spikes
“Basically like I would have like this maybe big AI page post about chat base, and I would see Stripe like maybe a few hours later go up by like 3000 MRR from just that one post.”
Early Chatbase users primarily wanted external customer-facing AI bots
“Most of the people that were interested were mainly interested in the external use case because they were used to like the, you know, like the drag and drop canvas AI chatbot, like a flow builder, which was very limited.”
Chatbase skipped VC fundraising because rapid traction left no time to pitch
“I thought I was going to do that, but because the product went so, like, it got so much traction so quickly, the reason I didn't do this is I didn't have the time.”
Chatbase customer model usage is split 50% OpenAI, 50% Anthropic and Google
“Maybe 50% is still on Okunai. Yeah. Yeah. And then 50 on everything else. Yeah. But everything else is like mainly Anthropic and Google.”
Yasser Elsaid: 95% of AI customer service limitations stem from harness architecture
“I would say like, 95% of the limitation is not from the model. It's from the harness. So like, it's my job to fix, but I think the intelligence is there and like, I think has been there for a while, especially for like, if you're only thinking about customer s…”
Proper AI harnesses enable 80% to 90% customer issue resolution rates
“If the harness is good enough, I would say, of course, it depends on the industry, but on average, you can get to 80, 90% of resolutions for, like, customer product issues, but to get there, it's the job of the harness”
Open-source AI demand spiked on hype before reverting to frontier labs
“Like all the open source models, I think what happened was they got like very hyped and people were very interested in using them. But I think like over time, like there was a spike in usage for these models. And then it goes back to open AI, Anthropic and Goo…”
AI model switching costs remain non-zero due to fragile harness optimizations
“People keep saying like the cost of switching between models is zero. It's cheap, but it's not zero because sometimes you like spend, you know, three, four months, like fine tuning exactly how the model should be, or like how the, Product should be. And then o…”
Chatbase currently serves 10,000 active enterprise and mid-market customers
“We have 10,000 active customers.”
The United Nations uses Chatbase on Facebook Messenger for regional crisis support
“Like, the UN is using us right now on their Facebook messenger. So when people reach out to them, specifically for, like, specific regions, it gets drafted to Chatbase, and Chatbase helps them.”
Self-serve GTM reveals lucrative niche verticals that sales-led assumptions often miss
“When you're not doing self-serve, when you're doing sales-led, you make a lot of assumptions. You assume, oh, like, of course, like, I'm gonna go into banks, or I'm gonna go into like startups or something like that, which is probably a good idea, but you woul…”
Employee-generated content is replacing ineffective influencer marketing strategies
“I think EGC is very underrated. I think that's the new influencer marketing. Influencer marketing, I feel like, doesn't work anymore. So I think that's, yeah, that's the new thing.”
Pure self-serve PLG fails for enterprise adoption without human outbound assistance
“I think a big mistake that PLG companies do is that, like, we're a PLG company, like, here's the pricing page, just, like, do your thing and pay us every month. But I think, one, like, even if your product is extremely intuitive and, like, everyone is able to …”
Offering custom demos to free signups functions as highly converting warm outbound
“You do content, you have the PLG motion, and then you put people in an email flow and in a calling flow where if they sign up, even if they don't subscribe or if they Like one of your posts, you reach out to them and say, hey, like, we want to build you a demo…”
AI Engine Optimization is fundamentally no different from traditional SEO
“So I don't think it's very different, honestly, no, because What the AI does is, is it searches the web and like finds answers and then it gives it to you. So it is very similar.”
Most Chatbase enterprise customers still prefer usage-based pricing over outcome-based pricing
“We started to like give some bigger customers the option to do that. And I think most of them still choose like the usage based, but some do choose the outcome based.”
Defining what qualifies as a resolution is the flaw in outcome-based pricing
“I think the only bigger, big problem with the outcome based is the definition of the outcome. Like you can call anything a resolution.”
Chatbase is actively replacing Zendesk as customers migrate to AI-native solutions
“Yeah, I think then, yeah, we are replacing Zendesk. So a lot of people are moving off of Zendesk to using it.”
Chatbase's sustained profitability enables aggressive VC-style hiring and scaling
“We're now at a point that we can function as a VCVAC company because we've been profitable for so long that it's as if we raised already, and, like, we can, now I have the mindset of being as aggressive with things like hiring and scaling,”
Outcome-driven engineers will thrive; craft-focused coders will struggle in the AI era
“The people that like coding because of what, how powerful it is, those people will thrive in, in this era. And I think people who like coding for the craft of coding will Probably have a bad time because that's not gonna be like the biggest thing in coding any…”
Yasser Elsaid uses Claude Code directly despite his team using Cursor
“No cursor, no. Like, the team uses cursor, but I feel like I'm one step removed from, like, the code that I just use cloud code directly.”