Neosantara & Vestara: From Idea to Validated AI Product
We build AI infrastructure for Indonesian developers and builders. The goal is simple: make it easier to access and integrate modern AI models without juggling multiple provider accounts, separate billing setups, or foreign-currency payments.
Here is the uncomfortable part. Technology alone does not solve the hardest part of building a company. In March 2026, CB Insights analyzed 431 venture-backed startups that shut down since 2023 and found 43% failed from poor product-market fit. They built something nobody wanted (CB Insights, 2026). A developer can ship an impressive AI product in a weekend and still learn that nobody needs it.
This is why our partnership with Vestara is practical.
Key Takeaways
- 43% of startups that shut down since 2023 failed from poor product-market fit, per CB Insights' March 2026 analysis of 431 post-mortems.
- Neosantara handles the technical setup (how to build with AI), while Vestara helps founders test whether an idea is worth building.
- Vestara's three-step method (Assess, Act, Advance) tests assumptions with real users before founders spend serious money, time, and code.
- Together, the goal is to help Indonesian developers ship validated products that compete locally and globally.
Why is AI infrastructure still hard for Indonesian developers?
Over 84% of developers now use or plan to use AI coding tools (Stack Overflow 2025 Survey, 2025), yet access stays uneven. For Indonesian developers the friction is practical: foreign cards, fragmented billing, per-provider quotas, and latency to overseas APIs. Neosantara exists to remove that layer so builders spend time on the product, not the plumbing.
The global AI ecosystem moves fast. New models, APIs, tools, and features appear weekly. That creates opportunity, but it also creates complexity.
Different providers mean different APIs, pricing models, billing systems, and rate limits. Neosantara provides a single layer that makes this easier. Instead of managing each underlying provider directly, developers can focus on building their product. If you are sizing throughput for production, our guide to understanding rate limits explains how RPM, input-token, and output-token quotas work per tier.
This matters in Indonesia. There is a growing group of developers here who can use AI to create software for both local users and international markets. And they can plug in through familiar tooling: Neosantara works as a native provider in any-llm and LiteLLM, so switching models rarely means rewriting integration code.
Why does building faster make failure faster?
AI-generated code jumped from 28% to 54% of developers' output in a single year (State of AI 2026, 2026). Speed is real. So is the trap it sets. Developers now spend 11.4 hours a week reviewing AI-generated code versus 9.8 hours writing new code (Digital Applied, Q1 2026). When prototypes cost almost nothing, what stops a team from building the wrong thing at record pace?
Making AI development easier has an unintended consequence. When software becomes cheaper and faster to prototype, more people build. That is helpful, but it also means teams can ship bad ideas faster than ever.
The question is no longer only whether you can build something. The question is whether you should build it.
For developers becoming founders, this distinction matters. A technically sound product does not automatically solve a meaningful problem. Before spending months writing code, founders need to understand their users, test their assumptions, and gather evidence that the problem is real. In our experience, the teams that skip this step pay for it later, with interest.
This is where Vestara brings a complementary perspective.
What does Vestara actually do?
Vestara helps early-stage and first-time founders turn a raw idea into an evidence-backed startup. The method runs in three stages, Assess, Act, and Advance, each designed to kill bad assumptions early while they are still cheap to kill.
Fig: Vestara, the validation partner behind the Assess, Act, Advance framework. (Source: Vestara, 2026)
| Stage | What happens | Question it answers |
|---|---|---|
| Assess | Map where the startup stands today, list the key assumptions, and decide what must be proven first. | What exactly are we betting on? |
| Act | Run focused actions that test those assumptions with real people. | Do real users confirm or reject the bet? |
| Advance | Use the resulting evidence to decide the next step: adjust the idea, test further, or move into product development. | What is the cheapest next step with the highest learning value? |
As the Vestara team puts it:
"You've got a startup idea, and you're ready to move. Maybe that means building the product, hiring a developer, finding a co-founder, starting marketing or looking for funding.
Vestara helps first-time founders strengthen the idea before the startup makes that move, because after can get very costly. Sometimes even hard to recover from."
This approach addresses a pattern we see constantly among technical founders: the gap between a working prototype and a viable business. A prototype proves you can build. It says nothing about whether anyone will pay. If you are already past that gap and designing systems, our walkthrough of LLM architecture evolution picks up where validation ends.
Comment from the Vestara team
CB Insights' 2026 analysis adds a detail worth sitting with: two-thirds of product-market-fit failures were early-stage teams that never found a market at all, exactly the stage Vestara works at (CB Insights, 2026). Here is what that work looked like from their side.
Throughout our work with Vestara, we focused on testing ideas directly and turning feedback into practical changes.
Simon from the Vestara team shared this comment on our work together:
"What has impressed us most about the founder of Neosantara is not simply his ambition, but his ability to absorb new ideas, challenge his own thinking, and then translate what he learns into action remarkably quickly. Throughout our work together at Vestara, we've watched him take approaches, methods and feedback, make them his own, and apply them with real discipline, and the results are already beginning to speak for themselves.
That combination of curiosity, humility, speed of learning and relentless execution is rare in a young founder. It's one of the reasons we're so excited about what Neosantara is building. The progress so far has been exceptional, but more importantly, it gives us enormous confidence in what lies ahead. We believe the future for this company is very bright indeed."
Simon also noted:
"Great founders have an unusual ability to turn insight into action. When you see someone learning, adapting and executing at this pace, you don't just get excited about the progress they've made, you get excited about how far they could go."
We publish this unedited because it captures the mechanism both sides care about: learn fast, change fast, show evidence.
Why the partnership makes sense
The logic fits on one line. Neosantara answers how to build with AI, while Vestara answers what to build and what evidence shows it is worth building. CB Insights' 2026 data backs this design: "ran out of capital" shows up in 70% of post-mortems, but the report calls it the final symptom, not the root cause (CB Insights, 2026). The loop attacks the causes upstream, before the cash runs out.
These are distinct problems, but they belong in the same loop:
Problem -> Validation (Vestara) -> Prototype -> AI Infrastructure (Neosantara) -> Feedback -> IterationThis works better than the common trap:
Idea -> Code -> More Code -> Launch -> Nobody Uses ItWhen developers validate first, they write code for real demand. And when they need to build, Neosantara gives them regional routing, standard OpenAI compatibility, and Rupiah billing so infrastructure does not slow them down. Teams wiring this into existing stacks can start from our LiteLLM provider guide or ship agents with Agno on top of the same gateway.
How do you build from Indonesia for a global market?
Start local, ship global. Indonesia's developer base keeps growing, and AI has collapsed the cost of reaching international users, but only for teams whose infrastructure does not fight them. Our job is to keep that infrastructure boring: local payments, transparent token pricing, low regional latency, and documentation you can actually follow.
Our focus at Neosantara is local. We build for developers in Indonesia.
That means addressing the practical constraints developers face here: access to international AI providers, local payment methods, transparent token pricing, low regional latency, and clear documentation.
At the same time, we do not expect Indonesian developers to stay limited to the domestic market. AI lowers the technical barriers to shipping software globally. Our job is to keep the infrastructure accessible so developers can spend their time solving real problems.
Building good products requires both execution and validation. Developers need dependable infrastructure, and ideas need evidence. Neosantara and Vestara address both sides of that work.
You can try the gateway at neosantara.xyz and learn more about founder validation at vestara.ai.
Frequently Asked Questions
Vestara helps early-stage and first-time founders validate startup ideas before committing serious money or code. Its three stages, Assess, Act, and Advance, map assumptions, test them with real users, and convert the evidence into a go, adjust, or retest decision.
Neosantara is a regional AI gateway for Indonesian developers: one OpenAI-compatible endpoint across many models, with regional routing, transparent token pricing, and Rupiah billing. It removes multi-provider accounts, foreign cards, and fragmented quotas from the build process.
Because the failure data points at validation, not tooling: 43% of startups that shut down since 2023 failed from poor product-market fit (CB Insights, March 2026). Pairing Vestara's validation loop with Neosantara's infrastructure covers both halves, what to build and how to build it.
Talk to prospective users before writing production code, then prototype the smallest thing that tests your riskiest assumption. Vestara's framework structures the first half; Neosantara's gateway, with guides for rate limits, LiteLLM, and any-llm, covers the second.
Neosantara signup includes free starting credit, so a first prototype costs nothing but your time. Details on Vestara's programs live at vestara.ai. Validate the idea first, then scale spend with evidence.
Source References
- [1] CB Insights, "Why Startups Fail: Top 9 Reasons," March 2026. URL: https://www.cbinsights.com/research/report/startup-failure-reasons-top/ (retrieved September 11, 2026)
- [2] Stack Overflow, "2025 Developer Survey: AI," 2025. URL: https://survey.stackoverflow.co/2025/ai/ (retrieved September 11, 2026)
- [3] Digital Applied, "AI Coding Tool Adoption 2026: Developer Survey Results," April 2026. URL: https://www.digitalapplied.com/blog/ai-coding-tool-adoption-2026-developer-survey (retrieved September 11, 2026)
- [4] State of AI, "State of AI 2026," 2026. URL: https://2026.stateofai.dev/en-US (retrieved September 11, 2026)
Validate first, then build on Neosantara
Test your riskiest assumption with Vestara's framework, then prototype on Neosantara's regional AI gateway with free starting credit, OpenAI-compatible APIs, and Rupiah billing.
