Clidey Built an Open-Source Org Modeler and Then Pivoted to Documentation That Writes Itself

The Macro: Every Company Is Drowning in Its Own Data

There is a particular kind of suffering that happens inside growing companies. The data exists. The relationships between teams, systems, customers, and decisions are all captured somewhere. But “somewhere” is the problem. It is in Notion pages that nobody updates. It is in Confluence wikis that are six months stale. It is in Slack threads that disappear into the void. It is in the heads of three people who have been at the company long enough to know where everything connects.

This problem splits into two related challenges. The first is organizational modeling: understanding how the pieces of your company fit together, what depends on what, and how decisions ripple through the system. The second is documentation: keeping an accurate, current record of how your codebase works so that new engineers can actually onboard without a two-month guided tour.

On the organizational modeling side, the market has tried a few approaches. Notion and Coda offer flexible databases. Airtable lets you model relationships between entities. Palantir’s Foundry does this at enterprise scale for governments and large corporations. But none of these tools were designed specifically for the problem of “model your entire organization as a living, queryable system.” They are general-purpose tools that people bend into shape.

On the documentation side, the situation is worse. Every engineering team knows their docs are out of date. Every engineering team says they will fix it next quarter. Nobody does. The tools that exist, Swimm, Mintlify, ReadMe, are good but require ongoing human effort to keep current. The moment someone refactors a module and forgets to update the docs, entropy wins.

AI is making both of these problems more tractable. Language models can read code and generate descriptions. They can infer relationships from data. The question is whether anyone can build a product that does this reliably enough to replace the manual processes that teams have grudgingly accepted.

The Micro: Two Products, One Vision

Clidey is a four-person team in San Francisco, part of YC’s Spring 2025 batch. The company was founded by Hemang Kandwal and Anguel Hristozov.

The founding team brings relevant background. Hemang previously worked at Palantir and major financial institutions, where he built over twenty proof-of-concepts recognized by BlackRock, J.P. Morgan, and other firms. That Palantir experience is directly relevant to the organizational modeling problem. Palantir’s entire business is built on connecting data, modeling entities, and making decisions on top of complex data structures. Hemang has seen what this looks like at the highest level and is now building a more accessible version. Anguel brings financial software development experience and community mentorship background.

The company has two products. WhoDB is the original vision: an open-source platform for modeling organizations and running decisions on top of their data. You define core entities, relationships, rules, and objectives. The system maintains a real-time model that you can query and act on. Think of it as a live org chart that extends beyond people to include every entity that matters to your business.

The second product, Docucod, feels like it came from watching what AI can actually do well right now. It automatically generates and maintains complete documentation from your codebase. No manual writing. No stale pages. The documentation stays current because it is regenerated from the source of truth: the code itself.

Running two products from a four-person team is either ambitious or unfocused. I lean toward ambitious, because the products share underlying technology. Both are about extracting structured understanding from unstructured or semi-structured data. The organizational modeler reads your business data and builds a model. The documentation tool reads your code and builds docs. The AI pipeline is similar even if the applications are different.

The open-source approach for WhoDB is a smart distribution play. Developer tools with open-source cores tend to build communities faster and face less resistance during adoption. It also gives potential enterprise customers a way to evaluate the tool before committing budget.

The Verdict

I think Clidey is working on real problems with a team that has the right background to solve them. The Palantir lineage gives Hemang a credible perspective on organizational data modeling that most founders in this space lack. The open-source strategy for WhoDB gives them a distribution advantage.

My concern is focus. Two products at four people is tight. WhoDB is a complex platform play that needs integration depth and enterprise sales motion. Docucod is a developer tool that needs seamless IDE integration and repository support. These are different go-to-market motions, different buyer personas, and different support requirements. At some point, they will need to pick a lane or grow fast enough to run both.

At 30 days, I would want to see GitHub stars and contributor activity on WhoDB. Open-source traction is the leading indicator for whether the organizational modeling concept resonates with developers. At 60 days, the question for Docucod is whether the generated documentation is good enough that teams actually use it as their primary reference, or whether it sits alongside the existing stale docs as yet another source. At 90 days, I would want to know whether any paying customer is using both products together, because that cross-sell is the real thesis for running them as a combined company.

The problems are real. The team is credible. The execution risk is focus. If they can thread the needle, both products could find meaningful traction in a market that is clearly underserved.