Knowledge
You sell software. Your operating cost is still a human assembling context.
Onboarding checklists, implementation tickets, support escalations, and renewal packets still depend on people searching Slack, the CRM, and last quarter's notes. Shipping another feature does not reduce that load.
How I see it
SaaS companies still drown in implementation and support packets
Technology companies often assume they are already good at AI because they build software. The operating workflows can still be analog: a CSM assembling a QBR, an implementation manager chasing configuration, a support lead reconstructing an account.
AI is useful when it gathers the account context, drafts the next action, and writes back to the system of record. An internal copilot that never updates Salesforce is a side task.
The first workflow should be one that grows with customers: onboarding, a class of support cases, or renewal prep. Measure time-to-go-live or time-to-resolution. Do not start with an autonomous engineering fantasy if the COO's problem is implementation capacity.
Keep humans on pricing, exceptions that create contractual exposure, and product judgment. Move the assembly work.
Common mistakes
What teams usually get wrong.
Building an agent platform instead of changing one customer workflow
The company already knows how to build software. The missing piece is usually the operating path.
Giving everyone a chat window and calling it adoption
Usage is not time-to-go-live. Attach the tool to a named artifact.
Starting with engineering agents because that is the identity
Engineering agents can be a later bet. The COO's first win is often onboarding or support.
A useful diagnostic
Five questions before you fund the work.
Which customer workflow grows linearly with logos?
That is the leverage point.How long does a person spend reconstructing an account before they can act?
Reconstruction time is usually addressable.What is the completed result: go-live, resolved ticket, or closed renewal packet?
If the unit is 'be more AI-native,' you do not have a workflow.Will CS or implementation change the definition of done?
If the old deck is still accepted, the new path is optional.Which actions create contractual exposure?
Those stay human.
Economic model
Implementation loadnew customers × hours of assembly and chase to go-live × loaded cost, plus delayed revenue while the packet sits
If delayed revenue is the larger term, cycle time is the case.
Three credible paths
How far should you go?
Do not force one solution. Choose the path the economics, the risk, and the organization can support.
Improve the work you already have
Keep the process mostly intact and use AI on the bottlenecks that create delay, rework, or follow-up.
The workflow is already sound and a few steps create most of the friction.
Gains are usually incremental. The operating economics do not change much.
Redesign the workflow around AI
Question every handoff, queue, and duplicate step, then rebuild the process around what AI can now do.
The process grew over years and coordination now costs more than the work itself.
Requires process change, clearer ownership, and a willingness to retire old steps.
Put an agent on a high-value outcome
Give an agent responsibility for one valuable result across systems, with humans at the control points that require judgment, authority, or risk acceptance.
The workflow is high-value, variable, multi-step, and worth engineering for production.
Needs stronger architecture, evaluation, controls, and monitoring. This is not a prompt project.
When this is the wrong next step
Do not start with a sector program.
- The company wants a platform story for investors more than a changed customer workflow.
- Support and CS will not share real tickets.
- There is no system of record, only Slack.
- The only request is autonomous coding.
A useful next step
Bring one expensive workflow from this industry.
We guide the implementation, go deep on the technical path, and stay hands-on through operations — or tell you when a simpler answer is better.
Discuss an AI opportunity