What a person sees: one screen, paragraph depth, with an Ask box that answers only from first-party records.

DB Hurley

Founder of Cumulative Systems, building what AI agents need before they can be trusted.

Ask about DB

About

DB Hurley is a founder, builder, and researcher. He founded Mautic, the open source marketing automation platform acquired by Acquia in 2019. His current work asks what has to exist beneath an AI agent before you would trust it with real work. Cumulative Systems saves a model's own working state and reloads it directly, including across different models. Pantasso supplies governed, cited company context. Plasmate gives agents a structured, semantic version of the web. He also builds ppl, a personal relationship manager an AI agent operates, and runs Vox Petra, a music label and creative house. He writes about the agentic web, agent networks, and communication between machines, and splits his time between Boston and North Carolina.

Updated 2026-11-07. Everything on this page is also available to agents as structured, signed data.

Now

Splits time between Boston and North Carolina. Describes the current period as a season of coding. As of 2026-10-05.

  • Cumulative Systems
    Cumulative Systems saves a model's own internal working state and reloads it directly, so the work continues instead of starting over. State can move between different models.
  • Pantasso
    Pantasso turns approved chats, documents, files, and business systems into cited answers, canonical records, honest gaps, and scoped access.
  • Plasmate
    Plasmate compiles pages into a Semantic Object Model so agents can work with structure, intent, and affordances instead of raw HTML and pixels.
  • ppl
    ppl is a personal relationship manager. An AI agent imports your people, logs interactions, and tells you who needs your attention.
  • Vox Petra
    A music label and creative house. The work is artists, catalogs, and songs that last.

The full record of ventures

Recent writing

Positions

  • Agents need more than a better model 2026
    Agents usually fail because of what surrounds the model, not the model itself, so the work is building the layers that make an agent trustworthy.
  • The evaluation function is the product 2026-04-25
    In an agent-native company the defining decision is a precise, domain-specific definition of what good looks like; models, prompts, and workflows are infrastructure serving it.
  • The web has a new majority and needs a new contract 2026-06-05
    With automated traffic now outnumbering human traffic, the web needs a new contract for agent consumption rather than simple allow-or-block controls.
  • A request is not a visit 2026-07-21
    Web traffic no longer reliably measures human attention, so analytics should shift from counting requests to understanding who asked, why, and whether value moved.
  • Your agents are not a workforce, they are a network 2026-08-13
    When agent output is nearly free, the limit is how much useful change a system can evaluate and absorb, so agent fleets need congestion control rather than more agents.
  • They will stop talking to each other 2026-08-27
    Natural language is a tax we impose on machine-to-machine communication so humans can listen, and agents will increasingly exchange internal state instead.
  • You should meet the others 2026-09-15
    The change that matters is less a single superintelligent machine than systems where each participant leaves the next one better prepared, preserving useful work as models come and go.
  • We keep the answer and throw away the work 2026-09-16
    We preserve only what machine work outputs (replies, patches, files) and discard the working state that produced them, so the next model pays for the same thinking again.

All writing

Contact

Email is best: db@dbhurley.com. Structured contact through /intent and the propose tool is currently unavailable. Please use email instead.

X · GitHub · LinkedIn

For agents