Manifesto · 2026
Operational intelligence is a missing category.
Every serious operator runs across five to ten disconnected systems of record. The state of the art for understanding what's actually happening across them is still a human reconciling spreadsheets. That's the gap we're building into.
The problem isn't data. It's synthesis.
For thirty years software has been pointed at one half of the problem: capturing more data, in more systems, more cleanly. QuickBooks for the books. Salesforce for the pipeline. Slack for the chatter. Jira for the work. Each one is excellent at its job and indifferent to the others.
The other half — the part that decides whether the business is actually doing well — has been left to humans with spreadsheets. Pull the export. Tab over. Rebuild the chart. Email the partner. Do it again next quarter.
BI tools were supposed to fix this. They didn't. They moved the spreadsheet into a browser and called it a dashboard. The synthesis problem — what is happening, why, and how confident am I in that answer — is still solved with a calendar, a deck, and a week of someone's time.
Why this is finally solvable.
Three things changed in the last eighteen months. Large language models became reliable enough to reason over heterogeneous business data without the brittleness that killed every previous attempt. The Model Context Protocol gave the industry a standardized way to expose tools and data to those models. And agent architectures with persistent state — beliefs, intentions, audit trails — gave us a way to do per-entity reasoning that survives a regulator's audit.
None of those existed at production quality two years ago. Together they make a category of software that wasn't possible before: an always-on intelligence layer that synthesizes every connected system into a single sourced answer, in plain English.
Why we're starting in private equity.
Pick the place where the pain is most concentrated, the buyer is sharpest, and the ROI of a single answered question is quantifiable in dollars. That's private equity portfolio monitoring.
A PE partner owns the outcomes of ten to thirty companies they do not operationally run. Quarterly reviews are the moment of truth. The work of getting from "let's talk about Acme" to a defensible IC-ready answer is still measured in days. ORIS answers it in seconds, with sources cited and confidence scored.
That's the wedge. It's narrow on purpose.
Why it doesn't stay in PE.
The same primitives — connectors, per-entity agents, sourced natural-language queries, immutable audit — apply anywhere a single operator runs across many entities and many systems. Holding companies. Family offices. Multi-brand operators. Franchise networks. Large independents whose CFO owns five businesses on paper and zero in real time.
Operational intelligence isn't a vertical. It's a layer. PE is where it gets sharpened.
What ORIS is.
An operating system, not a dashboard. Always-on agents per entity and per function — finance, sales, ops, comms — that maintain a live, sourced model of what's happening. A query layer that turns questions into answers without anyone touching SQL. An audit trail that logs every belief, every action, every answer, in a form that survives a regulator.
No carousels. No dashboards-as-decoration. No charts no one reads. The output is an answer, sourced, with a confidence level. The interface is a question.
What ORIS is not.
Not BI. We are not competing with Tableau. Not workflow automation — we are not n8n with a chat box. Not a chatbot bolted onto a data warehouse. The reasoning happens in agents with persistent state, not in a stateless prompt.
If this is for you.
We're onboarding a small group of pilot partners while v1 is built. Mostly private-equity firms, a few adjacent operators we think will sharpen the product. If that's you, leave an email and we'll be in touch.