A web app to build AI agents using simple prompts, You can think of it as Lovable but for Agents! Build automations and agents with no effort: 🧑💻 generate agent code 🔄 see agent's flow 🌐 one click to publish your agent 🎨 templates to get started easily
Okibi
Hey Product Hunt community 👋
I am Mahyad one of the co-founders of Okibi, a web app to create agents using simple prompts, and we are here today to showcase our product to you!
❌ The problem and why we care to solve it
The two of us have been building software for over 10 years and working together for the past 7. Actually, This is also our second YC startup together. At our previous company, SigmaOS (S21), where we built the first browser with built-in agents, we ran into a ton of repetitive tasks and had to build many agents for our own product. Automating those tasks was slow and complicated. We often ended up doing things manually or relying on partial solutions like Zapier, Lindy, custom scripts, or hacked-together LLM tools. The agents we did build for our product also eventually became messy and hard to maintain.
Creating AI agents typically requires complex, time-consuming setup, including tool calls, LLM syntax management, multi-agent orchestration, browser use, human in the loop, evals, and hosting deployments.
So, when building agents two primary challenges always come up:
- Evals
- Architecture and Infrastructure
✅ How Okibi solves it for you
Our web app provides a chat interface and toolkit to easily create AI agents. Just describe your agent in natural language, similar to vibe coding, and our app automatically generates your agent's tool calls, human in the loop, browser use, and runs an initial eval on your agent.
Whether you're building an internal workflow automation to remove repetitive or time consuming tasks, or launching a product with agents, Okibi can help you build it.
We are currently working with 15 YC companies from the current and previous batches to automate tasks like:
- Pre-qualify companies with the right person at those companies to sell your product to
- Generate invoices and update their invoice trackers based on email and contract
- Pre-meeting prep for client or sales lead call
- Generate pricing and proposal based on meeting notes and existing contracts
🤖 Ready to start automating with agents?
Head over to ijustwannabuild.ai, and use one of our templates to get started quickly and automate your repetitive tasks!
@mahyad Tried it just now, this is unbelievably fast...I would rather name it as BOLT for agents instead of Lovable (ofcourse, its likeable as well due to the fast output) 😉
Okibi
@shashwat_ghosh_gtm 🙏
@mahyad Interesting.
I have questions though:
how are you going to go with the pricing? Any plan to make it reasonable for small business owners and startups; or even educational use?
can I see your roadmap?
Bardeen
Pretty sick. I build everything in Zapier but we've pushing for n8n internally and with clients, so curious how this will replace that. How does it compare? And does this work well with @Notion and/or @Attio ?
Okibi
@temirlan Attio and Notion are top of the list, will come in the next couple days.
In terms of comparison with n8n, it is about making it super easy building the integrations and automations with just natural language instead of node manipulation, but you can also plug it into products that need agentic flows!
Nice! It's unclear to me what you mean when you answer "prompts" to "by what property do you want agents categorized". What's happening there?
Okibi
@steve_androulakis are you using one of the templates?
@mahyad This is purely from watching the video
April
So useful I see you have X Oauth in your templates - Signed up - will build agent to do cold outreach on X.
Okibi
@vedhsaka amazing, let me know how you get along!
Nia
SigmaOS Browser
@arlanrakh Thanks for the kind words 💪
Okibi
@arlanrakh thanks for all the early feedbacks!
Really interesting spin on the Lovable AI agent trend, Okibi seems to lower the barrier for creating custom agents with just prompts. Curious how it handles more complex workflows or multi-step reasoning. Definitely watching this space. How intuitive was the process of building your first agent on Okibi, and what kind of tasks did you have it perform?
Okibi
@ayanakamura22 the process was straight forward for me, the main agent I use, is a posthog even analyzer
It built it and eventually hit me - as a person testing first agent should not be treated to this - which now puts me off.