My turns any API into an AI agent in 60 seconds
A developer named Saya KH has built Orqis, a platform that converts any OpenAPI specification into a working AI agent in under 60 seconds. The tool eliminates the repetitive glue code that developers normally write when connecting large language models to external APIs, and it is...
Saya KH, writing for Indie Hackers on April 10, 2026, describes building Orqis out of direct frustration while working on internal tools for a client. The problem was familiar to anyone who has tried to connect an LLM to a real-world API: you end up writing the same boilerplate every single time. Define the tools, handle authentication, parse the spec, wire up the chat loop, and repeat. Saya built Orqis to end that cycle, and the resulting product is one of the more practically useful agent-building tools to surface on Hacker News this month.
Why This Matters
The AI agent space has a serious last-mile problem, and it is not model quality. It is the hours of integration work required before a model can actually do anything useful. Orqis attacks that problem directly by automating the translation layer between an OpenAPI spec and a functional conversational agent. If the 60-second claim holds up at scale, this tool could reduce the time-to-working-prototype for API-connected agents from a half-day engineering task to something a solo developer can do before lunch. That matters for a developer ecosystem where speed of iteration is the primary competitive advantage.
Daily briefing from 50+ sources. Free, 5-minute read.
The Full Story
Saya KH was not trying to build a product. The origin story here is genuinely practical: a freelance client engagement, a recurring integration headache, and a decision to stop solving the same problem manually. Every time Saya needed to connect an LLM to a client's API, the process was the same four-step grind. Define tools, handle auth, parse the spec, wire the chat loop. None of that work was intellectually interesting. All of it was necessary.
The solution Saya built, Orqis, works by accepting an OpenAPI spec URL as its starting point. From there it reads the full specification, generates typed tool definitions for every endpoint it finds, and produces a conversational agent capable of calling that API using plain natural language. The user never writes a line of code. There is no prompt engineering required. There is no LangChain boilerplate to configure or maintain.
What you actually get out of the other end is a complete agent environment. Orqis generates a visual graph showing all the tools derived from your API endpoints, a sandbox playground where you can test the agent before it touches anything real, and a dashboard for running actual tasks. Authentication tokens and API headers are stored encrypted at rest. Critically, for any endpoint flagged as dangerous, such as DELETE operations, the platform requires human confirmation before the agent proceeds. That last feature alone separates Orqis from a lot of the quick-and-dirty agent wrappers floating around right now.
Saya is also transparent about where the real engineering effort went. The AI integration was not the hard part. Making the agent behave reliably was. Specifically: getting the agent to send complete payloads rather than partial ones, surfacing actual API errors rather than hallucinating a successful response, and handling cases where an API returns something unexpected or incomplete. According to Saya, most of the iteration during development was spent on the agent runtime, not on the user interface. That is an honest and important distinction that most demos in this space conveniently skip over.
The roadmap Saya has outlined includes multi-step workflows that chain agents together, webhook triggers for event-driven automation, and shareable agent templates. None of those are available today, but they represent a logical progression from a single-agent tool toward something closer to an orchestration platform.
Key Details
- Orqis was posted to Indie Hackers by Saya KH on April 10, 2026.
- The platform converts an OpenAPI spec URL into a working agent in under 60 seconds.
- It generates typed tools for every endpoint found in the provided specification.
- Authentication headers and API tokens are stored encrypted at rest.
- Human-in-the-loop confirmation is required for destructive operations, specifically DELETE endpoints.
- The platform includes 4 core components: a visual tool graph, a test playground, a task dashboard, and encrypted header storage.
- Planned features include 3 additions: multi-step workflow chaining, webhook triggers, and shareable templates.
- The live product is accessible at orqisai.com.
What's Next
Saya has signaled that multi-step workflow chaining is the next major development priority, which would move Orqis from a single-agent tool into genuine multi-agent territory. Watch for webhook trigger support as the feature that will determine whether this can compete with automation platforms targeting enterprise workflows. If Orqis ships shareable agent templates, it gains a community and distribution layer that could drive adoption far faster than any paid marketing campaign.
How This Compares
The tool most directly comparable to Orqis is what you get when you combine LangChain's tool-calling infrastructure with something like Zapier's API connectivity layer, and then remove the configuration overhead from both. LangChain gives developers enormous flexibility but demands significant setup time for every new integration. Zapier connects APIs but lacks genuine agentic reasoning. Orqis is attempting to occupy the space between those two products, and on paper that is a real gap worth filling. The question is whether the agent runtime reliability Saya describes actually holds across the messy, inconsistently documented APIs that real-world developers work with.
Compare this to platforms like Relevance AI, which also offers no-code agent building but requires users to manually configure tools and data sources rather than ingesting a spec automatically. Relevance is more mature and has a larger feature set, but it also carries more configuration overhead. Orqis is betting that automatic spec ingestion is worth more to developers than a deeper feature list, and for the initial prototype use case, that bet is probably correct.
It is also worth placing Orqis in the context of what OpenAI has done with its function-calling and tool-use APIs since late 2023. OpenAI made agent-building dramatically simpler by giving models a clean interface for invoking external functions. Orqis builds on that foundation but solves a problem OpenAI did not address: where does the tool definition actually come from? Generating those definitions automatically from an existing spec is where Orqis adds its value. For developers already familiar with AI tools and platforms in this ecosystem, Orqis sits one abstraction level above the raw model APIs, which is exactly where a tool like this needs to live to be useful.
FAQ
Q: What is an OpenAPI spec and do I need to understand it? A: An OpenAPI spec is a standardized document that describes how an API works, including its endpoints, expected inputs, and authentication requirements. Most public and commercial APIs provide one automatically. You do not need to read or understand it yourself. With Orqis, you simply paste the URL where the spec lives and the platform handles everything from there.
Q: How is Orqis different from just using LangChain to connect to an API? A: LangChain requires you to manually define each tool, write the integration code, and configure the agent yourself, which can take hours per API. Orqis reads your OpenAPI spec and generates all of those tool definitions automatically in under 60 seconds. You skip the boilerplate entirely and go directly to testing a working agent.
Q: Is Orqis safe to use with APIs that have delete or write operations? A: Orqis includes a human-in-the-loop confirmation step specifically for destructive operations like DELETE endpoints. The agent will not execute those calls without explicit user approval. Authentication credentials are also encrypted at rest rather than stored in plain text.
Orqis is the kind of tool that gets built by someone who got tired of solving the same problem twice, which historically tends to produce more useful software than tools built by committee. Keep an eye on whether the runtime reliability Saya describes survives contact with production APIs over the coming months. Subscribe to the AI Agents Daily weekly newsletter for daily updates on AI agents, tools, and automation.
Get stories like this daily
Free briefing. Curated from 50+ sources. 5-minute read every morning.
