# The decisions behind the products.

Case studies in agentic AI, verification, enterprise permissions and product discovery. The decisions, constraints and outcomes behind the work.

## Portfolio Engine

A system of record for private equity where every number traces to its source document. Quarterly close went from weeks to days.

Decision: Make source provenance part of the record, not an audit feature.

Outcome: Quarterly close moved from weeks to days.

[Read the case study](https://thamanmuthappamn.vercel.app/work/portfolio-engine)

## Deal Intelligence

An agentic layer that breaks an investment thesis into linked claims, risks and initiatives, and reasons across every deal a fund holds.

Decision: Structure the thesis first, then let an agent reason over it.

Outcome: Thousands of initiatives proposed and amended.

[Read the case study](https://thamanmuthappamn.vercel.app/work/deal-intelligence)

## Conversational exit modeling

Ask “what gets this fund to 2.0x?” in plain language. 1,000+ scenarios run, none of them able to touch the ledger.

Decision: Separate hypothetical changes from financial ground truth.

Outcome: 1,000+ scenarios run in natural language.

[Read the case study](https://thamanmuthappamn.vercel.app/work/exit-modeling)

## Verification

An AI claim-and-citation checker shipped 0→1. Document review went from weeks to hours, with 1,000+ issues flagged.

Decision: Check the claim and its citation together.

Outcome: Document review moved from weeks to hours.

[Read the case study](https://thamanmuthappamn.vercel.app/work/verification)

## Permissions that fit the work

Permissions rebuilt as role-based access control around the 5 personas who use the platform, so enterprise teams could onboard without hand-holding.

Decision: Model access around the people using the platform.

Outcome: Enterprise teams could share work without hand-holding.

[Read the case study](https://thamanmuthappamn.vercel.app/work/rbac)

## Documents over Tables

A prioritisation call at VectorShift. Usage data said Documents, so I deprioritised Tables and focused the roadmap.

Decision: Stop what the evidence does not support, even when it looks like the obvious bet.

Outcome: Roadmap focus moved to the product customers used.

[Read the case study](https://thamanmuthappamn.vercel.app/work/documents-over-tables)

## Photon Studio

From store-manager interviews through specs, launch and feedback cycles at Infra.Market, done end to end as a software engineer.

Decision: Stay with the problem from the first interview through to use.

Outcome: Store-manager feedback shaped the roadmap after launch.

[Read the case study](https://thamanmuthappamn.vercel.app/work/photon-studio)

## Thal.E

Consumer 0→1 as founding product lead. Food ordering and queue management across 3 campus canteens: 1,000+ DAU and 2,000+ monthly orders.

Decision: Fix both sides of the counter, on one campus first.

Outcome: 1,000+ daily active users, 2,000+ monthly orders.

[Read the case study](https://thamanmuthappamn.vercel.app/work/thale)

## Also shipped

- Diligence Tracker (VectorShift): 0 → 1. Lets private equity clients automate due-diligence workflows.
- Kitchen visualiser (Infra.Market): 4–5 → 1–2 client visits. Real-time kitchen visualisation in Unity, Go and React. Shortened the sales loop.
- Photon (Infra.Market): 30+ retail stores. Operations platform. Owned development and prioritised workflow gaps with store managers.
- Rise (Infra.Market): 80%+ dealer orders automated. Removed manual processing from dealer ordering.

Last updated: 2026-09-13