Perpetual Core is a company that builds operating companies. We do not ship single-purpose software. We build the operating substrate underneath the company that uses it — and we build several of those substrates as their own brands, all shipping on the same runtime, all held to the same eval standard.
The portfolio
- Atlas is the AI-native COO for fund-backed companies. The flagship. Substrate, three departments, evals, reporting — installed inside the portco in six to ten weeks. The Operating Partner receives a fund-aggregated view across every install. In market.
- Sentinel is the sister product. DD dossiers on live deals, RFP intelligence, executive vetting. Atlas operates after the close. Sentinel evaluates before. Same substrate. Ships Q3 2026.
- Herald is the production runtime underneath both products — always-on agent infrastructure, the eval framework, the connectors, the Skills Library. Internal. Not a customer-facing brand. The structural advantage that makes the rest of the portfolio possible.
Operators, not vendors
Perpetual Core is run by operators. Lorenzo Daughtry-Chambers has spent the last decade building three things in parallel — institutional capital through DeepFutures, workforce infrastructure through Uplift Communities, and the AI engineering layer that ties them together. Founder of the Institute for Human Advancement. Investing partner of DeepFutures. Achumboro brings the operator seat: the daily reality of running departments, hitting numbers, and renewing on outcome rather than promise. No SDR layer. The founders read every reply themselves.
The brand of trust under Perpetual Core is structural, not stylistic. Every page on every property is footed with the lockup “Atlas is a Perpetual Core company.” Every product ships under the same eval discipline. The portfolio compounds because the substrate is shared and the operating standard is one.
The category window
The category we are competing in was theoretical on May 3, 2026, and a calendar event by May 5. The Anthropic-Blackstone JV announced $1.5B; OpenAI-TPG announced $4B+ within minutes. $5.5B behind a single thesis — AI installed across PE portfolios. The frontier labs and the world's largest PE firms have just chosen. The category window is the next twelve to eighteen months. Standalone AI vendors get filtered into noise; pilots that do not graduate are how the buyer learns to filter faster.
Casetext occupied the equivalent position beneath Thomson Reuters before the $650M acquisition. Atlas occupies the equivalent position beneath Anthropic-Blackstone and OpenAI-TPG. Same eval discipline. Same shape. Same exit math, only this time the window is open before the precedent is set. The operating-company shape buys the trust gradient single-product companies cannot buy. The lighthouse install is how the precedent gets set.