What Is Octar? AI for Enterprise Software Implementations
Octar is a YC Spring 2025 startup using AI to keep requirements, design and configuration aligned in ServiceNow, Salesforce and SAP rollouts.
For IT, Product Management and Operations

What is Octar?
Octar is an AI platform for enterprise software implementations. When a large company rolls out a system like ServiceNow, Salesforce or SAP, Octar helps keep the business requirements, the design decisions and the final configuration connected from the first discovery conversations through go live. The founders went through Y Combinator's Spring 2025 batch. Their core claim is that these projects often fail because the people involved lose track of what the business asked for, not because the software is hard to configure.
TL;DR
- Octar targets software implementations, the long projects where a company installs and configures a major platform like ServiceNow, Salesforce or SAP.
- The founders first planned to build "configuration agents" for each platform, then changed course after hearing that misaligned requirements and design cause much of the pain before any building starts.
- The key idea is traceability: being able to follow a business need through the design document and into the actual configuration.
- Octar says it keeps that context even after the people who worked on the project leave.
- The founders found the problem by talking to dozens of services leaders, after a friend described a very expensive project run on Word and Excel.
- Their early product grew into a pile of demo-driven features that was hard for new customers to adopt, which they describe candidly.
- If you work near a big rollout, the practical lesson is that writing down and linking decisions early protects the project more than any tool.
What is a software implementation, and why do they go wrong?
A software implementation is the project of making a platform work for a specific company. Buying the license is the easy part. Someone has to find out what each department needs, write it down, decide how the system should behave, configure it, test it and switch it on. Large companies often hire a system integrator (a consulting firm that does this work) or use an internal IT team, and the project can run for months or years.
The Octar founders described one such project: a multi-year ERP transformation for a U.S. state, with a budget in the hundreds of millions of dollars. The friend who told them about it was working at Deloitte, a large global system integrator. By his account, the team was running much of the work through Word documents and Excel sheets.
That detail matters because the failure modes are mostly about information, not code:
| Stage | What gets written down | What can go wrong |
|---|---|---|
| Discovery | Notes from workshops and interviews | Needs are heard differently by different people |
| Requirements | A business requirements document | Requirements are vague or conflict with each other |
| Design | A solution design document | Design drifts from what was actually requested |
| Configuration | Settings, workflows, fields in the platform | The build no longer matches the design |
| Go live and after | Handover and maintenance | The people who knew why things were built that way have left |
Each handoff loses a little context. By go live, nobody can easily say why a given setting exists or which business need it serves.
How does Octar try to fix this?
Octar connects the whole chain. The founders describe the platform as spanning initial discovery, design, configuration and go live, with the same context carried through all of it. They also say it keeps that context after a project ends, so a new team can see where things came from even when the original people have moved on.
The central concept is requirements traceability. In plain terms, it means every piece of configuration can be traced back to a design decision, and every design decision back to a requirement from the business. If a requirement changes, you can see what else is affected. If a setting looks strange, you can see the reason it exists.
One person the founders interviewed, who had worked in the industry for 30 years, told them nobody had solved traceability between requirements, design and configuration. That comment helped convince them the problem was worth building a company around.
Why did Octar start before the configuration step?
This is the most interesting part of the story. The obvious AI product for this market is a tool that does the configuration work. That was the founders' first idea too. After about 30 conversations with people in the industry, one founder wrote down what he thought the models could solve and landed on "config agents" for ServiceNow, SAP and Salesforce.
He then took that idea back to the roughly 50 people he had been talking to. They said it would help, but it did not address the root of their pain. Much of it came from misalignment in requirements and design before the build began. A faster build does not help if you are building the wrong thing, or if nobody can prove what the right thing was.
So the founders started upstream, with requirements and design alignment, which is also the wedge they pitched when applying to YC. According to them, the product has since been pulled downstream into configuration. The order is the lesson. They began where projects tend to go wrong, and the build step followed.
How did the founders find this problem?
They started with a friend's complaints. He was staffed on that large state ERP project and kept describing manual processes and tools like Word and Excel on a project worth millions. The founder who heard this was watching AI models improve around the GPT-3.5 and GPT-4 era and gain real usability in enterprises. His reaction was that it could not be the case that this was still how it was done.
He then spent about two months talking to dozens of services leaders: people at boutique consulting shops, at large system integrators and on internal IT teams. His account of those calls is useful for anyone researching a market:
- The first 20 or so calls were basic questions about how the industry works.
- By the 20th to 30th conversation he was speaking the same language and asking sharper questions.
- He repeated what people said back to them and asked for corrections.
- He wrote the idea in a shared Google doc and iterated on it with the same people.
The change of direction came from that last step. Showing a written idea to people who live with the problem gave them something to push back on.
What happened during the YC batch?
The founders describe a deliberately fast, sales-led batch. They launched early, posting a YC sign photo in about the second week that featured a golden retriever named Summer, whom they called the "chief bark officer." They say the LinkedIn post came close to a million impressions and booked roughly 200 to 300 demos within three days, though they admit many were not qualified. Some large enterprise contracts they hold today, they say, came from people who saw that post. A second, more conventional video launch at the halfway point drew about 400,000 to 500,000 impressions.
Then came the honest part. With hundreds of demos on the calendar, one founder ran the calls while the other two built product, shipping changes each night in response to that day's objections. They raised a seed round and moved from San Francisco to New York. They also say the product at the end of the batch was a "Frankenstein of demo features."
Their example: customers asked for a business requirements document builder and a separate solution design document builder. What they really needed was one document builder with a flexible agent behind it. Too many narrow features meant a steep learning curve for new customers, which made adoption hard.
The transcript cuts off before the founders explain how they fixed this, so what came next is not covered here.
What does this mean for you?
You do not need to work at Octar or a consulting firm to be affected by this. If any of the following describes you, the lessons apply.
If you are on a business team in a rollout. You are often the source of the requirements. Ask to see how your request is recorded and how it will be tied to the final setup. If you cannot answer "where does my requirement live now?", raise it early.
If you are a project or program manager. Check whether anyone can trace a configured feature back to a requirement without asking a specific person. If the answer depends on one person's memory, that is a risk to handle before go live.
If you work at a system integrator or in IT. Expect more tools that draft requirements and design documents, and treat the traceability link as the real value, not the draft itself. Output quality matters less than whether the chain holds.
If you are evaluating a tool like this. Ask how it handles changes mid-project, what happens to context when staff leave, and how quickly a new user can get value. The Octar founders' own story shows that a feature-heavy product can be hard to adopt.
A low-tech starting point works with any tool. Give each requirement an ID, record which design decision answers it, and record which configuration item implements it. Even a simple table makes gaps visible.
Can you build your own requirements tracker?
For a smaller rollout, a team may not need an enterprise platform to get the basics of traceability. A shared tracker with requirements, design decisions, configuration items, owners and status covers a lot. If you would rather have that as a proper app than a spreadsheet that drifts out of date, you can describe it to Remy, which builds and deploys full-stack business apps from a plain-English description. A team could ask for a requirements tracker with linked design decisions, a request form for stakeholders to submit needs, and a status dashboard, with sign-in and a database included. It is a way to own a small internal tool instead of adding another subscription.
Frequently Asked Questions
What does Octar do?
Octar is an AI platform for enterprise software implementations. It aims to keep business requirements, design decisions and configuration connected from discovery through go live, for platforms such as ServiceNow, Salesforce and SAP.
Is Octar a Y Combinator company?
Yes. The founders were in the Spring 2025 batch. They raised a seed round by the end of it and moved from San Francisco to New York.
Does Octar configure the software for you?
That was the founders' original idea, with configuration agents for each platform. They found the larger problem was upstream, in requirements and design alignment, and began there. They say the product has since been pulled downstream into configuration.
What is requirements traceability?
It is the ability to follow a business requirement through the design that answers it and into the configuration that implements it. It lets a team see what a change affects and why a setting exists, even after the original staff have left.
Why do software implementations fail or run over?
According to the Octar founders, a common cause is misalignment in requirements and design at the start, before any build work begins. Context also gets lost across handoffs and when people leave the project.
Topics
How we source, write, and correct our articles. Editorial standards
Related articles

OpenBot vs Grok Bot: Which AI Agent Should You Use?

CLI-Anything: What It Is and How to Use It With AI Agents

What Is Box Architect? How Box's AI Designs Agents
