Custom AI agents, built for monday.com.
We design, build, and host Claude-powered AI agents that run on your monday.com boards. Webhook triggers, monday GraphQL writes, prompt and model under your control. PM Pack live today; sales-ops, legal-ops, staffing, and agency packs on the roadmap.
What we build
AI agent development, native to monday.com.
Every business running on monday.com eventually wants AI doing work alongside the team. Triage new items. Draft replies. Parse inbound email into structured records. Summarize a day's activity. Look at a photo and write down what it shows.
Some of that work fits into pre-built tools. The deeper, more shaped work needs a custom agent: a prompt written for your domain, a model chosen for your latency budget, triggers wired into the exact monday automations your team already runs.
LaunchEngine Agents is the runtime and dashboard that lets us ship those custom agents quickly. Webhook in, Claude out, monday GraphQL write back. Prompts and models live in the dashboard so they can be tuned without redeploying. Cost telemetry on every run.
Property management was the proving ground. The PM Pack runs today on real PMOS customer workloads. Sales-ops, legal-ops, staffing, and agency packs ship next as we see demand.
Why LaunchEngine Agents
Native. Custom. Already running in production.
Built for monday.com. Triggers, writes, and data live on your boards.
Agents fire from monday automations via webhook, read context from your boards through the monday GraphQL API, and write results back to the same items. No Zapier middleware. No Make scenarios. No glue layer to maintain. The same direct integration pattern as our RentEngine monday Code app.
Your prompt. Your model. Your trigger logic.
Every agent is a row in our runtime with its own system prompt, model selection, and trigger config. Edit from the dashboard, no code deploy. Pick Sonnet 4.6 for reasoning work, Haiku 4.5 for cheap classification, or route between them on the same agent fleet. When the agent misbehaves, you debug the prompt instead of switching products.
Per-run cost telemetry, always on.
Every fire logs input tokens, output tokens, dollar cost, latency, and status. The dashboard at launchengine.com/app shows spend per agent and per run. You know what a workflow costs before the bill arrives, and you can tune the prompt or swap to a cheaper model when something runs hot.
Proven by the PM Pack. Six agents live on real workloads.
Pilot for monday update inbox handling. Echo for inbound maintenance email parsing. Penny for outbound email drafting. Maxine for new-item triage. Iris for photo analysis. Recap for the daily PM digest. All running today for paying LaunchEngine PMOS customers. Real load, real customers, real cost telemetry.
Six years building on monday.com.
Same team that built the PM Operating System, DocRunner, and Bill Runner. We know monday's automation surface, GraphQL quirks, and rate-limit shapes by heart. When you hire LaunchEngine for monday agent development, you get a partner who has already shipped the patterns you need.
What it does
The capabilities, plainly.
Webhook-triggered runs
Trigger from monday automations. Webhook hits the runtime, the agent loads the configured prompt, Claude returns, the runtime writes back via monday GraphQL. Gmail and cron triggers also supported.
Configurable prompt and model per agent
Each agent has its own row in Postgres with prompt, model, and trigger config. Edit from the dashboard. Route cheap classification work to Haiku 4.5 and reasoning work to Sonnet 4.6 on the same agent fleet.
Per-run cost telemetry
Every fire logs input tokens, output tokens, dollar cost, latency, and status. Dashboards break the numbers down by agent and by run, so you can tune prompts before costs surprise you.
Vertical packs
PM Pack live today (six agents in production). Sales-ops, legal-ops, staffing, and agency packs on the roadmap. Packs are pre-built prompt bundles shaped for a vertical's workflows. The runtime itself is vertical-agnostic.
Roadmap packs
PM is live. Four more packs on deck.
Vertical packs are pre-built agent prompt bundles shaped for a specific workflow. The runtime is vertical-agnostic. Build order follows waitlist signal. If you want a pack in your vertical, say so.
PM Pack
LiveSix agents in production. Pilot, Echo, Penny, Maxine, Iris, Recap.
Sales-Ops Pack
RoadmapDeal-stage recap, MSA gap auditor, renewal nudge drafter.
Legal-Ops Pack
RoadmapContract intake classifier, clause flagger, amendment summarizer.
Staffing Pack
RoadmapCandidate intake triage, offer-letter drafter, ghosting detector.
Agency Pack
RoadmapScope-change detector, weekly account digest, retainer renewal nudger.
Pricing
Scoped to your workflow.
Pricing depends on agent count, call volume, and whether you are already a LaunchEngine PMOS customer. Bundled into PMOS for property management customers today. Standalone pricing for non-PMOS customers is set during a scoping call.
Talk to us about pricingFrequently Asked Questions
What LaunchEngine Agents is, how it integrates with monday.com, and how to get started.
Looking for monday.com AI agent development?
PMOS customers get the PM Pack on day one. Non-PMOS prospects can request a pack or build their own agent from the dashboard. Pricing tailored after a 30-minute call.