Plan · 31 July 2026

Plan & Task Tracker

The 90-day sequence, who owns what, and a live tracker you can edit directly on this page. Built around one principle from the call: prove it before we build it.

The team

Three people, three different things

The split is unusually clean, which is the strongest early signal this venture has.

EH
Build & AI delivery

Eduardo Haddad

Dripping Springs, Texas

Thirty years in finance and ERP transformation at Deloitte, HP and Google. Now builds production systems hands-on with AI tooling. Owns the product, the front end, the agents and the financial architecture.

BringsThe build capability. Turns an idea into something working in days rather than quarters.
DV
Enterprise & client side

Dirk van der Merwe

Houston, Texas

Managing Director at Accenture, complex ERP transformation programme design and delivery. Previously global SAP digital innovation lead at HCL AXON, solutions architect at HPE, and Controllership PMO at HP. Owns integration strategy and enterprise credibility.

BringsThe integration expertise that is the actual moat, plus C-suite relationships and commercial instinct.
A
Market access & back office

Albert to confirm

To be confirmed

Placeholder — sketched from Dirk's description on the call, to be replaced once Albert has actually said yes. Serves small and mid-sized clients in the $2–3M band. Long relationship with Dirk; they co-own property. Brings the customers and the back-office knowledge neither of the others has at this scale.

BringsClient base and back-office expertise. As Dirk put it: "We need someone who has actually served the small-medium clients."

⚑ First decision: what is this called? "Meridian Austin" was chosen for a one-person Austin practice. It no longer fits a Houston-and-Austin team of three, and the name blocks the domain, the email addresses and anything printed — so it needs settling in week one, not month three. Three routes:

  • Keep Meridian, drop the city. Just Meridian, or Meridian Field Systems / Meridian Operations. Cheapest option — Eduardo's résumé and LinkedIn already carry "Meridian", so nothing gets rewritten. Caution: "Meridian" is one of the most crowded names in US business, and Meridian AI, Inc. filed a bare-word MERIDIAN trademark in Class 42 in February 2026 — so anything of the form "Meridian AI" is off the table.
  • Keep Meridian as Eduardo's practice, name the venture separately. Cleanest legally, and it keeps Eduardo's independent consulting identity intact alongside the three-way venture.
  • New name entirely. Best long-term asset, most work now. Worth it only if we believe this becomes a product company rather than a consultancy.

Recommendation: decide the shape in the first conversation — is this Eduardo's practice with two partners, or a new company all three of you own? The name follows from that, and the answer also drives the equity question (task 22). Don't buy a domain until the shape is agreed.

Why the split works. Eduardo can build it, Dirk can integrate and sell it into serious organisations, and Albert can reach the buyers and knows their back office. Most early ventures have three people who all do the same thing. This one doesn't. The dependency to watch: Albert is the only route to the customer right now, which makes him a single point of failure until we have our own pipeline.

Sequence

The first 90 days

Cheap and reversible first. Nothing here requires an entity, a bank account or a dollar of capital until Phase 3.

Phase 0 — Agree the thing. Week 1.

Albert conversation, then all three sit down. Agree the target segment, who does what, how time is split, and what "in" means for each of us. Dirk checks his Accenture position on outside activity before anything is registered. No money, no entity, no commitments yet.

Phase 1 — Prove the pain. Weeks 2–4.

Albert introduces us to five owners in the target band. We ask what bad job capture actually costs them and whether they can even measure it. Separately and just as important: put a rough prototype in front of five working technicians and watch them use it. If they won't, nothing downstream matters.

Phase 2 — Build the narrow thing. Weeks 4–8.

One trade, one workflow, end to end: capture on a phone → job card → customer signature → invoice into QuickBooks. Not a platform. The narrowest path that produces a paid invoice a day earlier than before. Free tiers only.

Phase 3 — One paying client. Weeks 8–12.

A single friendly client at a real price, on whichever commercial model they prefer. The goal is not revenue — it is a reference story with a number in it, and an honest measurement of how many hours delivery actually took. Form the entity now, not before.

Phase 4 — Decide. Day 90.

Look at the reuse figure. If client two would take a third of the hours of client one, this compounds and we invest. If it would take the same, we have a consultancy — which is a fine business but a different one, and we should say so out loud rather than drift.

Live tracker

Task tracker

Edit owner, status and notes directly below. Changes save as you make them.

# Phase Task Owner 1 Owner 2 Status Note

🔒 Confidential working draft. Please don't share this link outside the three of us.