Where AI helps, and where a person decides.

Each blueprint names what the agent reads, what it drafts, what it may not do and who approves it. None is deployed yet.

AI by team

Conceptual blueprint

Agent blueprints

Sales and CS

Account brief assistant

Trigger
A meeting is scheduled
Reads
Authorised CRM notes, invitation, public pages
Produces
Draft brief, every fact sourced
May not
Send, edit records, fill gaps
Reviewer
The account owner
Open the shared workflow
Customer Success

Health change investigator

Trigger
A health input crosses a threshold
Reads
Usage, support, relationship signals
Produces
Likely causes, each linked
May not
Contact the customer, change scores
Reviewer
The assigned CSM
Customer health workflow
Sales and RevOps

Pipeline review preparer

Trigger
Before the weekly review
Reads
Opportunity changes since last review
Produces
Changes and risks, with links
May not
Edit stages, amounts, forecast
Reviewer
The sales manager
Pipeline review workflow

The Mossbridge Demo brief shows the target output, hand-written from invented records with no agent involved: download the synthetic example. Every blueprint stops at read, draft and suggest, and a named person approves the rest: how we review.

Questions

Does wayhatch run AI agents?

No. wayhatch publishes conceptual blueprints and research, plus one hand-written synthetic example. No deployable agent or live integration exists, and no business results are reported yet.

What does a blueprint contain?

The trigger, what the agent reads, what it produces, what it may not do and who reviews the output. Each blueprint links to the shared workflow it would support.

Who approves an agent output?

A named person: the account owner, the assigned CSM or the sales manager. Anything a customer sees or any change to a system of record needs that approval.