Written for Operations, technology, data and risk leaders, and the sponsors who have to fund a first integration of AI into systems that already run.
What's in this playbook
This playbook is the operational document behind AsymiLink Meta's AI Integration consulting page. It is a method for designing a first integration of AI into systems that already run: an inventory of what is already in production, a named source of truth, agreed definitions, fourteen design tests, and a bounded first case that can be stood behind. It is complementary to four documents already in this library — the data-strategy guide, the AI Readiness Assessment Framework, the AI Vendor Evaluation Scorecard, and the RAG architecture guide — and is not a replacement for any of them.
It will not promise a twelve-week go-live, will not publish a gallery of named-vendor architectures, and will not claim a certified connector to a named platform. Systems are named as you run them. A program to "connect everything," or an unsupervised write-back across systems, is not a first case this playbook will recommend.
How the tests and the method are meant to be used
A first case is how the path is raised, not a demonstration that every family of work is ready at once. The honest default is a read-and-draft assist: extraction or a recommended next step, from a named system, for a named person. A write into a system of record is a later, different design, with a named signer, a trail, and tests of the failure cases, not only of the happy path. A person sits on every write and every customer-facing message. Review is defined in writing; a prompt is not a gate. Hosting is decided before the case is funded. Where data cannot leave, AsymiLink AI runs on your own hardware, with an optional air gap. Isolation is not coverage of a design that was never finished.
The financial method is The Complete Guide to AI Automation ROI: a named process, a baseline agreed before work starts, and a calculation you can defend. This playbook will not invent an integration-sector return or a twelve-week payback. The 3.8x average first-year ROI across client engagements is an average, not a forecast or a commitment for any individual project. ROI is measured across client engagements against a baseline agreed before work starts. It is an average, not a forecast for any integration project, and is not used as a sector benchmark. Built to SOC 2 standards is the exact wording we use; it is not a certification of the reader's system.
An illustrative mid-size company, introduced in this playbook and labeled hypothetical, sits at 18, Emerging, on the eight-dimension readiness model. Infrastructure and Strategy are relative strengths; Data Foundation is the constraint. Its honest first work is the inventory and a bounded draft, with a next-step target of 24, Developing — a next step, not a timeline, and not a promise. A companion workbook, asymilink-ai-integration-playbook.xlsx, holds a family register, the fourteen tests, questions for counsel, a baseline sheet, and the illustrative company.
Six families of integration work
each with an honest first case, a case that is too wide, and a gate. Each item below is the shape of the map; the full wording, honest first cases and “too wide” warnings are in the PDF.
F1
A bounded extract and draft
F2
A recommend-and-review assist at a named interface
F3
A write-back into a system of record (later)
F4
Cross-system orchestration (later, bounded, with a person on every irreversible step)
F5
A customer- or partner-facing draft
F6
Hosting, decided before the case is funded
The PDF gives, for each family, an honest first case, a case that is too wide, and a defined gate. The summaries above are the shape of the map, not a substitute for that wording.
Fourteen design tests (excerpt)
Fourteen design tests are applied to every case. The answers will differ. The standard does not. The full wording is in the PDF and the companion workbook.
T1–early
inventory and source of truth; proportionate governance; a defined review; counsel asked in time; agreed definitions and evidence.
Mid-tests
hosting; write versus read; an audit trail; a named operator and a stop; training and reuse.
Later tests
irreversible and customer-facing acts; honest notice; a bounded first case; and a signed baseline, without a reporting rhythm invented as a company commitment.
A completed test sheet is not a certificate, not a compliance score, and not a finding that a system is safe or aligned with a named regime.
Get the guide and workbook
Enter your name, work email and company. The files appear on this page immediately. A backup download link is also emailed when email delivery is available.