Hundres.
An agentic growth-automation platform. AI agents plan, preview and run marketing work across ads, social and email, and a person approves every action that spends money.
Status: pausedHundres set out to automate the day-to-day of growth marketing: deciding what to run, preparing it, launching it and learning from the result. The hard part wasn't getting agents to act. It was making them safe and legible enough that a non-technical owner would trust them with a budget.
Agents that ask before they spend
Paid actions on Meta and Google ads run through a sequential execution state machine. Each step shows a preview first and waits for a human to confirm it. A channel-balance guardrail stops budget piling into one channel, and the Meta ads throttle logic was reworked to remove a permanent deadlock.
A layer that learns
A DSPy-inspired knowledge layer scores the outcome of every action and feeds the patterns that worked back into future planning prompts, so the plans improve as results come in.
Integrations as MCP servers
- LinkedIn, email, Instagram and analytics MCP servers
- A Buffer API-based social media MCP covering 11+ platforms
- Approval gating exposed as MCP tools
Logs a non-engineer can read
The activity log was redesigned around plain-English feeds and pre-action previews, so an owner can see what the agent did, what it plans to do next and why, without opening a dashboard.
Where it stands
Hundres is paused. Proving its decisions properly needs live ad-account data at scale, and that comes next. The architecture behind it — gated agents, outcome scoring and MCP integrations — is what we now bring to client work.
Ask us about it ↗