Every request, one shape
Forms, emails, and chat asks get read and normalized against your intake schema — whatever the requester typed, however incomplete.
Service 1 of 4 — heals: intake
Messy brief in. Clean project out.
AI intake triage reads every request — form, email, or "quick ask" in chat — classifies it, picks the right project template, maps the fields, assigns the owner, and sets the priority. Vague requests bounce back with the questions already asked.
Book a teardown
Bring your intake form. We'll map it live.
Re: before / after
The same Tuesday, twice.
Before — manual
09:00 AM request lands in shared inbox
10:40 AM producer notices, opens PM tool
11:15 AM re-keys 14 fields from the email
11:30 AM wrong template. starts over
02:10 PM emails requester: "quick question…"
NEXT DAY project finally exists
After — triaged Automated
09:00 AM request received → parsed
09:00 AM classified: packaging refresh
09:00 AM template matched, 14 fields mapped
09:01 AM owner assigned, priority set
09:01 AM requester gets scope questions
09:02 AM project exists. nobody typed.
FIG. 1 — Same request, minus the day of latency.
Re: how it works
Step 1 — Parse
Step 2 — Classify
Step 3 — Bounce or build
Three steps, all logged.
Right template, right owner
The request is matched to a project template in your PM tool, fields mapped, owner and priority set by your routing rules.
Vague in, questions out
Complete requests become projects in about a minute. Incomplete ones bounce back to the requester with the missing questions already written.
Re: proof
Running now, at a national health system.
11:03 AM creative request detected → 3 assets
11:03 AM 3 projects built, templates matched
11:04 AM designers + writers assigned. 0 re-keyed.
1 form → done
One request becomes a project per asset, on the template its own metadata called for, staffed with the designers and writers that metadata names. Failures log to a queue; the critical ones page Slack.
Re: questions you'd ask
FAQ
What tools does it work with?
Workfront, Asana, Monday, Wrike, and most PM tools with an API. The triage layer sits in Make, n8n, or Workfront Fusion — wherever your automations already live.
What happens to requests it can't classify?
They route to a human queue with the AI's best guess attached. Nothing is silently dropped; every decision is logged.
How long is the build?
Two weeks from teardown to running system, tested against your real intake backlog — not demo data.
Re: next step
Put your intake form out of its misery.
Book a teardown
30 minutes. We map your intake and price the waste.