HOW TO ENGAGE ASI

Briefing guide

This page is a simple template for briefing ASI. You don’t need a 40-page strategy deck — just enough structure that we can understand your world, your constraints, and what “good” looks like.

1. Context

A short snapshot of where you are today. This doesn’t need to be perfect; it just anchors the conversation.

Helpful details to include:

• What your organisation does in one paragraph.

• Where AI is already in use (if anywhere).

• The teams or roles most affected by AI decisions.

• Any regulatory, compliance, or safety constraints.

Example (compressed):

“We’re a B2B SaaS company handling sensitive customer data. Support, sales and product teams all touch the same information, but in different tools. We currently use LLMs for drafting messages and summaries; nothing is automated end-to-end yet.”

2. Constraints

Constraints are not a problem — they are where good architecture starts. Be explicit about what we cannot do.

Typical constraint themes:

• Data residency, privacy, or PII rules.

• Systems that cannot be changed (legacy ERP, core banking, medical systems).

• Risk appetite: what absolutely must not fail silently.

• Budget / time horizons for an initial deployment.

Example wording:

“We cannot send raw customer data outside our region. Any AI layer must explain its decisions. Failure modes around billing, legal commitments, or clinical advice are unacceptable.”

3. Desired outcomes

We focus on outcomes, not features. Tell us what success looks like in human terms — we’ll map that to agents, reasoning layers, and invariants.

Useful outcome frames:

• “Reduce X by Y%” (e.g. manual triage time, noise, rework).

• “Increase clarity for Z team” (e.g. fewer meetings, clearer summaries).

• “Make this decision safer / more transparent.”

• “Turn this messy process into a stable workflow.”

Example outcomes:

“Reduce time spent triaging inbound tickets by 40% without losing edge cases.”
“Give our leadership team one reliable weekly AI-generated briefing they can actually trust.”

4. Engagement patterns

We typically work in small, focused phases. You don’t need to commit to a multi-year programme to get value.

Typical path:

1. Architecture / risk briefing (remote session).

2. Draft system sketch: cognition base, reasoning loops, agents, invariants.

3. Prototype phase around one high-value workflow.

4. Evaluate behaviour under pressure; adjust invariants and guardrails.

5. Decide whether to scale, pause, or refine.

5. Sending your briefing

You can use this page as a checklist. A simple email with a few clear sections is better than a glossy deck.

Suggested email structure:

1. Context: 3–6 sentences.

2. Constraints: what must be respected.

3. Desired outcomes:

4. Timeline / urgency:

Send your briefing to asicai@protonmail.com and we’ll respond with a simple architecture sketch and recommended next step.