The Migration Playbook

I built my business on ClickFunnels. Before that, Infusionsoft. Same story every time: 250 features, most of them 70% of what I actually wanted, all wrapped in bloat I was paying for every month.

Then I'd prune the stack to save money and the tools got worse. Custom fixes? A pain in the ass — someone else's developer builds someone else's roadmap on their timeline, not yours. And if the feature you asked for six months ago finally ships, so does the same feature for every competitor in your market.

With agents, you build the exact three-to-five features you actually need. No filler. No bloat. No waiting.

That's what this playbook is. Not "switch SaaS." Not "DIY your stack." Build the tools custom-fit to your business — all killer, no filler — the ones that make you a motherfucking juggernaut of competitive efficiency.

As of July 8, 2026, we're entirely on a custom stack from top-of-funnel through middle-of-funnel. Only sales calls and live events are humans now. Human being, not human doing. That's the point.

Born in the Optimus Mastermind. This kit came straight out of a room full of $5-50M owners done paying platform prices for a fraction of platform usage. The full call — and the room where these get built live, alongside other founder-architects — lives inside the Optimus Mastermind. Apply to join at buildwithoptimus.com →

Member edition. Three kill candidates. Three side-by-side migrations. Three triggers to pull the plug.

This playbook picks up where the free scorecard leaves off. The scorecard tells you what to replace. This tells you how — the lean agent that does the job, the two-to-four week overlap, and the specific condition that makes it safe to cancel.

It's written for a $5–50M founder-architect with a CFO asking hard questions and a team that can't afford a botched cutover. Every replacement here is a concrete, side-by-side plan — the lean agent, the overlap window, the exact cancel trigger — not a whiteboard sketch.

The three kill candidates

The same three categories almost always sit at the top of the kill list:

1. The bloated PM tool. You bought it for tasks. It grew a wiki, a docs system, a chat, a whiteboard, a form builder, and a lite CRM. Your team uses tasks. You pay for the other seven products in the bundle.
2. The bloated CRM. You inherited it from a growth phase two acquisitions ago. It has sixteen custom objects, four automations nobody remembers writing, and a per-seat price that scales linearly with sales headcount.
3. The bloated dashboard. A BI platform with five hundred dashboards, ten of which anyone actually opens, and a query cost that grew forty percent last year. Half the dashboards were built by a consultant who left.

Each of these has the same failure mode: you're paying platform prices for a fraction of platform usage. And each has a lean agent replacement that does the four features you actually use — no more, no less.

Migration one — Replace the bloated PM tool

What you're replacing

The all-in-one project-management platform. Fourteen features across seven product areas. Your team uses:

• Task creation and status
• Assignee and due date
• A weekly view of what's due
• Notifications when something's late

That's it. You're paying for wikis, docs, whiteboards, forms, dashboards, and chat you already have somewhere else.

The lean agent replacement

A task-tracker agent with three ingredients:

• A structured table (Postgres, Airtable, or a Google Sheet with the API on). Columns: task, owner, due, status, project, notes.
• A form endpoint on top of the table. Anyone on the team can drop in a task without seeing the raw table.
• A daily-briefing agent that reads the table each morning, produces a per-owner digest of what's due today and what's overdue, and posts to your team channel.

Total moving parts: three. Total cost to run: under fifty dollars a month for a team under a hundred people. Total build time: an afternoon in Optimus.

Side-by-side migration — four weeks

Week 1 — mirror. Stand up the agent stack. Import last quarter's active tasks from the incumbent tool. Both systems run in parallel; the team keeps using the incumbent.

Week 2 — shadow. New tasks get created in both systems. Owners get the agent's daily briefing in parallel with the incumbent's notifications. You're checking for drift — anything the agent misses that the incumbent surfaces.

Week 3 — cutover primary. New tasks land in the agent stack only. The incumbent is read-only, kept live for reference. Team is coached: "the agent is the source of truth now."

Week 4 — freeze. Nobody writes to the incumbent. It exists purely as a historical record.

Pull-the-plug trigger

Cancel when all three are true:

1. Two consecutive weeks with zero writes to the incumbent tool.
2. The daily-briefing agent has caught at least three overdue items nobody flagged manually. (Proves the agent surfaces the same signals as the incumbent's notifications.)
3. You've exported the last thirty days of history from the incumbent to cold storage.

Then cancel at the next renewal date or the next monthly-billing cycle, whichever is sooner.

Estimated savings

A team of forty on a mid-tier PM platform runs $18K–$30K a year. The agent stack runs $600–$1,200 a year all-in. Net savings: $17K–$29K, recurring.

Migration two — Replace the bloated CRM

What you're replacing

The enterprise CRM you inherited. Sixteen custom objects, four legacy automations, sales seats scaling linearly with headcount. Your team actually uses:

• Contact records with company, title, and last-touch date
• Deal records with stage, value, and expected close
• Email history against each contact
• A pipeline view of open deals by stage

That's four features. You're paying for lead scoring, marketing automation, sequencing, service tickets, and a partner portal you don't run.

The lean agent replacement

A CRM-lite agent with four ingredients:

• A structured contacts + deals table (Postgres, or a well-normalized Airtable base).
• An email-log agent that watches your team's inbox (via Gmail or Microsoft Graph API) and files threads against the right contact record.
• A pipeline-view agent that renders the current state of open deals by stage on demand, and produces a Friday roll-up for the sales lead.
• A stale-deal agent that flags any deal untouched for more than fourteen days.

Total cost to run: under two hundred dollars a month for a sales team under thirty. Total build time: two to three days in Optimus.

Side-by-side migration — eight weeks

CRM migrations are the highest-integration migration in this playbook. Take longer.

Weeks 1–2 — mirror. Export contacts, companies, deals, and email history from the incumbent. Load into the agent stack. The incumbent is still the source of truth.

Weeks 3–4 — shadow writes. New contacts and deals get created in both. The email-log agent starts filing threads in parallel. Reps continue to work in the incumbent. You're checking that the agent's pipeline view matches the incumbent's within a two-percent margin.

Weeks 5–6 — cutover primary. Reps switch to the agent's pipeline view. The incumbent becomes read-only. Any rep who prefers the incumbent's interface gets a two-week grace period, then loses the seat.

Weeks 7–8 — freeze. Nobody writes to the incumbent. All new email logs flow through the agent. The stale-deal agent is running, and the sales lead is getting the Friday roll-up.

Pull-the-plug trigger

Cancel when all four are true:

1. Two consecutive weeks with zero new writes to the incumbent.
2. The Friday roll-up has been accepted by the sales lead for four consecutive weeks as the primary pipeline artifact.
3. No open deal has moved more than one stage without a corresponding record in the agent stack. (Catches quiet incumbent use.)
4. Contact and company data has been exported to a warehouse table you control, not the incumbent's cloud.

Cancel at the next renewal — CRMs have brutal cancellation terms mid-contract, so time this precisely.

Estimated savings

A twenty-seat enterprise CRM contract runs $60K–$180K a year. The agent stack runs $2,400–$6,000 a year all-in. Net savings: $54K–$174K, recurring. This is usually the biggest single line item on the audit.

Migration three — Replace the bloated dashboard stack

What you're replacing

The BI platform with five hundred dashboards and rising query costs. Your team actually looks at:

• A weekly revenue view
• A weekly acquisition funnel
• A weekly retention cohort
• A monthly board-metrics one-pager

Four dashboards. You're paying platform prices to store the other four hundred and ninety-six, plus the query engine cost to keep them refreshable.

The lean agent replacement

A reporting-agent stack with three ingredients:

• A materialized-view layer in your warehouse (Snowflake, BigQuery, Postgres) that pre-computes the four views you actually read.
• A rendering agent (using a lightweight tool like Metabase Cloud, Evidence.dev, or a custom Notion-integrated agent) that produces the four views on demand.
• A narrative agent that reads the underlying data and drafts a two-paragraph "what changed this week" summary, delivered by email each Monday.

Total cost to run: warehouse compute (which you pay anyway) plus $50–$300/month for the rendering tool. Total build time: one to two weeks in Optimus.

Side-by-side migration — two weeks

Dashboards migrate fastest because there are no writes to worry about. Reads only.

Week 1 — build the four. Model the four materialized views. Render them in the new stack. Send the Monday narrative alongside the existing dashboard access.

Week 2 — cutover. Direct all leadership viewers to the new stack. Turn off the incumbent's SSO for everyone except the analytics team, who need read access to move any remaining ad-hoc queries.

Pull-the-plug trigger

Cancel when all three are true:

1. Two consecutive weeks where the incumbent's dashboards receive fewer than five unique-user views organization-wide.
2. The Monday narrative has been referenced (forwarded, quoted, cited) in a leadership meeting.
3. The analytics team confirms every ad-hoc query worth saving has been migrated to the warehouse-native SQL layer.

Cancel at the next renewal or at the end of the current billing month, whichever is sooner. BI platforms often let you downgrade to a bare-bones tier for the analytics team during a transitional quarter — do that instead of full cancel if it saves you the audit hassle of one-off SQL migrations.

Estimated savings

A mid-market BI platform with fifty viewer seats runs $50K–$140K a year. Warehouse-plus-rendering runs $10K–$25K a year. Net savings: $40K–$115K, recurring.

The rollup — running all three at once

Most founder-architects run these three migrations in sequence, not parallel. Reasoning:

• The PM migration is the easiest and builds team confidence. Start here.
• The dashboard migration is the fastest and produces the most visible weekly artifact. Do this second.
• The CRM migration is the biggest savings and the highest risk. Do it last, with the team already fluent in the side-by-side pattern from the first two.

Total elapsed time: about a quarter. Total recurring savings for a typical $5–50M founder-architect running all three: $110K–$320K a year, net of migration and agent-runtime costs.

The pattern

Every migration in this playbook follows the same shape:

1. Identify the four features you actually use. Ignore the other eighty percent of the platform.
2. Build a lean agent that does exactly those four features. No more.
3. Run agent and incumbent side-by-side long enough to prove the swap is safe. Two to eight weeks depending on integration depth.
4. Define the exact trigger conditions that mean you can cancel. Not "when it feels right." Specific, verifiable conditions.
5. Cancel at renewal, not mid-contract, unless the math justifies eating the cancellation fee.

Follow the pattern and you can compress your SaaS spend by sixty to eighty percent without slowing the team down. Skip the side-by-side and you'll re-buy the incumbent three months later at a worse price.

What comes next

Once you've killed or replaced the top three, run the scorecard again on the remaining stack. There's almost always a second wave — usually the marketing-automation platform, the customer-support desk, and the identity/access tool. The same pattern applies.

Not in Optimus yet? Apply to join the Optimus Mastermind → — it's where these migrations get built live, alongside other founder-architects.

Already a member? Bring your last renewal invoice and active-seat count to the weekly help call on Wednesdays and we'll build the agent replacement live.