Skip to main content
AI Stack
Book a fit check
Install process

Four weeks to a working operating rhythm.

AI Stack does not run open-ended AI projects. We map the workflow, connect the tools, set approval rules, test silently and go live with weekly review.

Book a fit check
  • Fixed price, fixed scope
  • Human approval on anything client-facing
  • Silent test before go-live
  • 60-day guarantee
Task log
  1. CheckedNew enquiry from web form
    Checked
  2. UpdatedCRM record created
    Checked
  3. DraftedFollow-up message
    Draft ready
  4. FlaggedMissing phone number
    Needs approval
  5. PreparedDelivery handover note
    Draft ready
  6. WaitingManager approval
    Waiting on manager

Example log

Week 1

Fit and workflow map

We map how work currently moves through the business.

We look at lead sources, CRM use, inbox flow, task handovers, approval rules and current gaps.

Output

Workflow map, agent scope and go-live rules.

Week 2

Build and connect

We connect the CRM, inbox, task tools and team chat.

We configure daily summaries, draft rules, stale lead alerts, handover logic and escalation triggers.

Output

Configured agent in test environment.

Week 3

Silent test

The agent runs without taking live action.

We check whether it finds the right enquiries, matches contacts properly, updates tasks correctly and writes useful summaries.

Output

Test report, fixes and approved workflows.

Week 4

Go live

Approved workflows go live.

The agent begins daily summaries, follow-up drafts, stale lead alerts, handover preparation and weekly reporting.

Output

Live CRM and Workflow Agent.

Weekly review

Weekly review during the first month

The first month after go-live is used to tighten the operating rhythm.

  • Work completed.
  • Draft quality.
  • Missed detections.
  • Approval delays.
  • Tool errors.
  • Manager actions.
  • Scope changes to park for later.
Measurement

How you know it is working

Every month you get a one-page report. Not a dashboard you have to log into and not a data export.

  • Hours returned. Measured against the baseline we agree in week one.
  • Work completed. Enquiries checked, follow-ups drafted, records updated, handovers prepared.
  • Things caught. Stale leads flagged, missing details found, overdue actions surfaced.
  • Where it needed you. Approvals waiting, escalations raised, drafts you rewrote.
  • What we are changing next month. Every correction you make becomes a rule the agent keeps.

The last line matters most. An agent that is not learning from your corrections is a script, and scripts go stale.

Monthly report, example
Hours returnedagainst a 3 hr baseline
11.5 / wk
Work completed
412 items
Things caught
27
Where it needed you
9 approvals

Next month: tighten stale-quote rule from 7 to 5 days.

Example only

What we need from you

The install works best when the client provides one clear internal owner.

  • Business goals and current pain points.
  • List of software used.
  • Access permissions.
  • Approval owner.
  • Actions the agent must never do.
  • Daily and weekly reporting preferences.
  • Feedback during silent testing.

Start with the workflow map.

The fit check shows whether your current lead, CRM and handover process is ready for an agent.

Book a fit check