mention-intelligence
Maintains: Awareness of every OpenProse mention across monitored platforms, alongside competitor activity, with deltas since the last run.
OpenProse on OpenProse.
We are writing our own company as responsibility contracts, with evals and durable traces. These are experiments in progress, not a live operations floor, and this page says so. The dogfooding story is here, honestly. Alpha and frontier R&D throughout.
The experiment: run OpenProse on OpenProse.
We are writing our marketing, GTM, enrichment, and community operations as Prose programs. The goal is that the same language we ship to customers is the one running our own company: on a schedule, with evals, to Slack. We are not there yet, and this page will say so until we are.
Each responsibility declares a contract (what it requires and what it maintains). Workflows bind the outcomes to cadences and channels. Everything gets an eval.
Because we have to.
We sell service-as-software: outcomes on a schedule, not a tool you operate. If Prose cannot run a company, it cannot run yours. So we are building ours on it first, and the parts that break are the parts that get fixed in the language.
Because contracts compound.
A responsibility framed by what it maintains (not what it does) gets better as models improve. Declarative by default. Every contract is paired with an eval, so we can measure how well it holds before we rely on it.
Because the shape scales.
The same shape holds as the clusters grow. Product-engineering already has two responsibilities and six evals; back-office and customer success follow the same pattern. No reorgs. Just new rows in the same table.
Every responsibility declares what it maintains.
Grouped by the work they cover. Each has a written contract (requires, maintains, errors, strategies) and a paired eval.
Marketing intelligence
What the market is saying, and where to show up.
Four responsibilities for staying aware of the conversations happening around agent frameworks, and for drafting a first move when one of them wants an answer.
competitor-intelligence
Maintains: A structured read on competitor framework moves (releases, funding, messaging changes) filtered to actionable signals.
opportunity-discovery
Maintains: A deduplicated, scored feed of conversations where OpenProse naturally adds value. Never leads with promotion; the job is genuine helpfulness first.
engagement-drafting
Maintains: A response for each opportunity that answers the question first, matches the platform's voice, and holds for human review before posting.
Public indices
The reports we are building for the world.
Two responsibilities cover the public ecosystem maps behind agentsy.live; a third watches traffic on our own site.
agent-ecosystem-index
Maintains: A current map of agent-native communities (activity levels, trends, health indicators) powering the pulse/index on agentsy.live.
saas-readiness-index
Maintains: An empirical leaderboard of SaaS products ranked by how easy they are for AI agents to integrate with.
agents-site-analytics
Maintains: Awareness of who is visiting agents.openprose.ai, where they came from, and how the mix is changing. Sourced from CloudFront logs, not a third-party tracker.
Enrichment
Turn a hint into a cited profile.
Two shared enrichment capabilities that other contracts call into. Pure enrichment, no qualification framing, so downstream consumers can add their own interpretation.
person-enrichment
Maintains: A structured, cited person profile from any identity hint: name, LinkedIn, X, GitHub, bio snippet, or email. Exa-backed.
company-enrichment
Maintains: A structured, cited company profile from any company identifier: name, domain, or GitHub org. Cacheable per company, so downstream calls amortize.
GTM
From a lead to a demo repo.
Three responsibilities designed to compose serially. Lead in, enriched profile and program pair out, and a scaffolded customer repo ready for a pilot. Human-gated at the step that becomes visible to the prospect.
lead-enrichment
Maintains: Composes person- and company-enrichment and adds qualification: fit score, Save/Grow framing, customer slug.
program-designer
Maintains: Two candidate Prose programs (one Save, one Grow) tailored to this specific person and company. Demo-ready contracts, not slideware.
customer-repo-scaffolder
Maintains: A complete customers/prose-{slug}/ directory: responsibilities, functions, workflows, and evals. Ready for the pilot engagement.
Community
Every new star gets a face.
One responsibility today, with more planned. The signal we care about most is not download count. It is who, specifically, found us and why.
stargazer-intake
Maintains: Every new star on openprose/prose becomes a classified, enriched record within 24 hours. Maintains a high-water mark so replay is safe and the delta is the interesting number.
How they compose (the workflows that bind outcomes to cadences and channels, the intelligence-daily and gtm-pipeline worked examples) is covered in the OpenProse manual.