Consulting practice · AI advisory

Ephod Advisory

I help firms and households figure out what to do about AI, then build or set up whatever the answer turns out to be.

Ephod Advisory is my consulting practice. I build software, set up AI tools, train the people who use them, review vendors, and advise on the decisions around all of it. What I’m most careful about is not over-building. A lot of what gets pitched as an AI project is really a paid seat configured well, or a small piece of software. When that’s the case I’ll say so rather than propose something bigger, because I’d rather do the right amount of work and get referred to the next person.

Engagements are hourly or on a short retainer, scoped in writing up front.

What I do

Build software Architecture and implementation for scoped products and internal tools: AI-backed apps, integrations, and operating layers over the systems a team already runs. I keep them small and readable so the client’s own people can maintain them afterward. Recent work includes an operating layer for a family office that connects documents, mail, banking, and accounting data.

Advise Review proposals, join vendor calls, answer questions by email, and think through strategy with whoever is running the AI program. I’m not selling the build, so the opinion is straightforward.

Set up AI tools Get people onto the right tools, connect them to the mail, calendar, and files they already work in, and set up the recurring tasks that make the tool worth opening every day.

Train and enable Hands-on sessions for partners, executives, and the assistants who run an office. We work on their actual tasks rather than slides, and I follow up until people are using the tool on their own.

Review vendors and consultants Put competing proposals on the same criteria, check what’s actually being promised, and run a standard diligence list: data use and retention, security, access controls, IP ownership, confidentiality, term.

Governance and policy Review or draft AI use policies so they cover the tools people actually use, say who approves what, and don’t accidentally prohibit work the firm already does.

Data access and boundaries Decide which accounts and folders an AI tool can reach before it’s connected to anything. Most of the time this is a permissions decision, not a system to build.

Prioritize Turn a long idea list into a scored plan, and separate the real AI use cases from ordinary automation.

How I work

I don’t over-build.: If the cheapest correct answer is a subscription and an afternoon of setup, that’s what I’ll recommend. Building is on the table when it’s the right answer, not when it’s the most billable one.
Access comes first.: Which accounts get connected and what they can reach is settled before any tool touches sensitive material. Approvals stay on for anything that sends, shares, or deletes.
Start small.: One or two tasks that work beat a dozen that get switched off. Widen the scope once the first pieces have earned it.
Straight about scope.: If the ask is bigger than the budget, I say so in the first conversation and we trim or resize. Nobody finds out at the invoice.

Who it’s for

Professional services firms working out what AI should mean for them. Family offices and the people who run a household’s affairs. Small teams that need a specific piece of software built well and handed over.

Most clients come by referral, and I don’t name engagements publicly.

Formats

Build Architecture and implementation, scoped and priced as its own project
Retainer A block of hours for review, calls, email, and strategy, used as needed
Sessions A fixed run of working sessions to get a person or a team using a tool
Review A single deliverable: a policy redline, a vendor comparison, a diligence checklist

Get in touch

For engagements and referrals me@wesleydickens.com

Or book a thirty-minute intro call

If the scheduler doesn’t load, open it on Calendly.