Erin Stanley, Evoked - Engagement Window
The synthetic-content marking obligation lands December 2. This window closes December 31.
Three engagements. Public pricing. Available through December 31, 2026. Then the chapter ends and a different one begins.
Your agents shipped. The audit room is asking a different question than the product room asked. Not what the agents can do. What they should refuse, who decides, and where that decision is written down. Most teams cannot answer that yet. You have weeks, not quarters, to find out whether you can.
I have a bounded 2026 window for three kinds of work. Public rates. No discovery dance. If the offer matches what you need, the math is already on the page.
Before you choose
The engagements below deliver two things in one. The technical shortcut to where you need to be by December 2 - the audit, the spec, the retainer. And a shift in how your team holds the governance question itself.
If you arrived ready for both, the engagement is right. If you arrived ready only for the deliverable, the writing surface or the products page may be the better service for you right now.
The Three Offers
Sovereignty Architecture Audit
Know your standing before your auditors do.
Eight weeks. Four working sessions. Written architecture deliverable.
The 47-point assessment applied to your agent system across seven domains over an extended contract engagement - identity, memory, restraint, voice, drift, refusal architecture, and governance surface. We go deeper than the score. The eight weeks let us walk your build team through the findings as they land, run architecture recommendations against your real constraints, and produce a remediation roadmap your team can execute against with effort sizing on every line. You learn where your agents hold and where they break before your auditors do - and you have time to fix what matters most before the next compliance window opens.
- Written sovereignty architecture audit scored against 47 properties
- Architecture recommendations sized against your current build state
- Prioritized remediation roadmap with effort estimates per finding
- Four working sessions across the eight weeks: one executive readout, one build team kickoff, one mid-engagement checkpoint, one closing alignment
- Async review of refusal specs, system prompts, and architecture artifacts throughout the engagement
- Thirty-day follow-up window for clarifications after final delivery
Agent Architecture Spec
From demo to deployment.
Two to three weeks. One working session. Written spec.
A specification document your autonomous agents can execute against over long time horizons. Covers identity boundaries, memory architecture, refusal rights, voice discipline, and drift detection. Written in language a build team can ship from and an audit team can verify against. The difference between a demo and a deployment.
- Written agent architecture specification
- Refusal specification authored against your operational risk surface
- One working session to align the build team on the spec
Governance Advisor Retainer
A governance partner at your ship cadence.
Sixteen hours per month. Through December 31. Renewable.
Standing governance counsel for teams whose agents are shipping faster than their guardrails. Weekly office hours, async review of build artifacts, executive briefings, and direct authorship on policy documents when the situation calls for it. Two retainer slots available through year-end.
- Weekly thirty-minute office hours with the build or governance lead
- Async review of refusal specs, system prompts, audit artifacts
- One executive briefing per month, on request
- Renewable January 1, 2027 at a different rate and cadence
Why December 31
A 2026 governance window is not abstract. The EU AI Act use-based high-risk deadline just moved to December 2027 under the May 2026 Digital Omnibus (provisional political agreement; consolidated text pending Official Journal publication). The synthetic-content marking obligation lands in December 2026. Article 5 prohibitions and General-Purpose AI rules are live today. The Anthropic-Blackstone joint venture launched a $1.5 billion Claude services build in May. The mid-market Claude consulting space is being capitalized in real time. The teams that get architected this year are the teams that help shape the standards. The teams that wait inherit them.
Regulatory facts verified 2026-06-06. EU AI Act in active amendment; see canonical analysis.
The window is also mine. The work in January 2027 looks different than the work in November 2026 - new EU deadlines move from the calendar into the codebase, and what teams need from me changes accordingly. The December 31 date is not a marketing hook. It is when this offer page comes down.
Why Me
Start the Conversation
If one of the three offers matches what you need, the next step is a thirty-minute scoping call. Public rates apply. No proposal cycle.
Two retainer slots. Limited audit slots. Architecture spec capacity scoped per week. Slots fill in the order they are booked.
Smaller engagements and self-serve tools - from $49 framework kits to $5,000 governance audits - live at /consulting.
The agents are not going to stop shipping. The audit room is not going to stop asking. The window is what it is. If this is the year you make your agents legible to the people who will judge them, we should talk.
Erin Stanley, Evoked