PinnedShyam Desigan·Mar 25Automating Finance SOPs with AI: A Phased Framework for Recovering 1,700+ Hours Per YearFrom Manual Journal Entries to Predictive Finance Infrastructure
PinnedShyam Desigan·Mar 25Metabolic Commerce: Deploying AI Agents as Financial and Operational ReflexesFrom Retrospective Accounting to Predictive Finance Infrastructure
Shyam Desigan·Sep 5The Connector Is Not the ControlI pushed UiPath into a number of our repos this week. A clean path into the Microsoft stack removes a whole class of brittleness. It does…
Shyam Desigan·Sep 4Grade the Close. Then Grade the Board Pack.A CFO OS is not a chatbot. Strata scores the function and the draft on the same rubric. Then it turns the heatmap into a 90-day plan.
Shyam Desigan·Sep 3The Smallest Safe Way to Bring AI Into FinanceStart with one workflow, one control point, and evidence that can be re-performed. Then scale.
Shyam Desigan·Sep 3Governed AI Is Not the Same as Generic AutomationFinance needs identity, an adversary, a lock, and a sampleable trail before it scales AI. MCP does not specify that. CHP does.
Shyam Desigan·Aug 22I had one protocol implemented four times.I have been building agent governance tooling for about a year. The core of it is something I call the Consensus Hardening Protocol — a set…
Shyam Desigan·Jun 7Stop Building Agents. Start Shipping Workflows.The post-mortems are accumulating. A $30K AI agent build that worked in the demo fails in production. Eight LLM calls cascade into a…
Shyam Desigan·Jun 7SEC Earnings Workbench: CHP-Hardened Multi-Agent Research for SEC Filings & Earnings AnalysisEquity research has a provenance problem. A stock rating arrives with a recommendation and a target price. Somewhere upstream, an analyst…
Shyam Desigan·May 27Why I Built My Multi-Agent AI Sandbox on Firecracker MicroVMs — Not DockerThe moment I stopped trusting containers with autonomous agent output