Follow-up leakage
Proposal follow-ups are inconsistent even though one miss can stall a high-value opportunity.
Example sample
What an initial consultation output looks like.
This example shows the kind of map an initial consultation can produce before any implementation work begins.
Workflow map
The work loop before AI.
Referral or inbound lead to discovery call, notes, proposal, follow-up, kickoff, delivery, and weekly client check-ins.
Context search
Meeting prep requires searching across Gmail, Calendar, Drive, Notion, and past notes.
Mental load
Client commitments and next actions live partly in Maya's head.
Opportunity scoring
What the consultation recommends first.
The first build should combine a Work Radar, Proposal Follow-Up Radar, and Meeting Prep Brief. Higher-risk automation waits.
| Opportunity | Score | Recommendation | Why |
|---|---|---|---|
| Weekly Work Radar | 20 | Build first | Surfaces leads, proposals, client commitments, and weekly priorities. |
| Proposal Follow-Up Radar | 19 | Build first | Flags stale proposals and drafts review-ready follow-ups. |
| Meeting Prep Brief | 18 | Include in first pilot | Assembles context before client calls from approved notes and documents. |
| Client Knowledge Search | 14 | Phase 2 | Useful after source-of-truth boundaries are cleaner. |
| Auto-Send Follow-Ups | 11 | Do not build early | Too much action risk before approval queues and evidence displays exist. |
Readiness and risk
Ready, but only with review.
Maya's workflow has clear value and approval ownership. The right first mode is limited, read-only, and draft-first.
5/5
Business Value
Missed proposals and prep time have direct revenue and time impact.
4/5
Workflow Clarity
Consulting pipeline is straightforward enough to map.
3/5
Data Readiness
Useful data exists, but it is scattered across tools.
5/5
Human Approval
Maya approves all client-facing outputs.
3/5
Operational Maturity
Solo practice, simple review loop possible.
Private-AI boundaries
Decisions reserved for people.
The safest early system makes work visible and drafts next actions. It does not mutate client-facing records or act externally without approval.
Starting boundary
- Start with uploaded non-sensitive documents and manual exports.
- Keep Gmail, Drive, Notion, and pipeline data read-only at first.
- Draft follow-up emails, but do not send anything automatically.
- Show evidence next to every recommendation.
- Exclude pricing changes, commitments, and client-facing document updates from early automation.
Suggested next step
Move from consultation to an operational audit.
The paid deep-dive would go further: confirm the tools and data boundaries, agree what the system may do on its own, and produce a concrete pilot plan.