Pi's KMS and Skills: Memory and Playbook for Ecommerce Teams

The uncomfortable truth about most AI tools is that they forget you. Each conversation starts fresh. Every question requires you to re-explain your business, your priorities, and your constraints. What worked yesterday teaches nothing about what should happen today.

The cost of that gap adds up fast. A senior marketplace manager might spend twenty minutes explaining pricing floors, priority ASINs, competitor exceptions, and MAP guardrails to an AI. Next week, all of it has to be explained again. The knowledge that lives in your team's heads never gets encoded anywhere Pi can use.

KMS and Skills are the parts of Pattern Intelligence designed to fix that. KMS gives Pi a memory of your brand. Skills turn the workflows that memory enables into commands your whole team can run.

What does KMS actually remember about your brand?

KMS is Pi's Knowledge Management System. It's a brand-controlled library where your team loads its own guidelines, tone, and preferences directly into Pi, so Pi doesn't have to relearn your brand every conversation. You control what goes in, which means you control how Pi acts on your behalf.

The kinds of context KMS holds include:

• Brand guidelines: voice, positioning rules, category strategy, and anything else that shapes how decisions should be made on your behalf

• Priority ASINs: which SKUs deserve heavier attention, tighter guardrails, and faster response times

• Pricing floors and MAP rules: the guardrails competitive pricing must respect, including exceptions by channel or geography

• Terminology: the product names, category language, and internal shorthand your team uses, so Pi speaks about your brand the way your brand does

• Past corrections: moments where a Pi recommendation was overridden by a team member, along with the reason

• Approved decisions: strategic calls leadership has made that should inform future recommendations without needing to be re-litigated

Every piece of this context makes the next answer better. When you ask Chat-to-Data a question, Pi filters through what it knows about your brand before it responds. When Sensors detect a condition, Pi checks whether your KMS holds a rule that applies. When Approvals surface a decision, the context Pi presents already includes your priorities.

The important word is controlled. KMS doesn't guess at your preferences from behavior and hope it got them right. It's a library your team fills deliberately, reviews, and updates. If your MAP policy changes or you reprioritize a product line, you update KMS and every downstream Pi action reflects the change from that point on. The brand stays in charge of the rules the machine operates under.

Why does memory compound faster than intelligence?

A common mistake with AI tools is measuring them by how smart they seem in the first conversation. That's the wrong test. The right test is how much better they get after fifty conversations, after five hundred, after a year of daily use.

Systems that remember get better with use. Systems that forget stay the same forever.

KMS is engineered around that reality. Every override, every correction, every strategic decision your team makes gets encoded. Six months in, Pi has absorbed a picture of your brand that would take a new hire two years to develop, and that picture keeps sharpening every week.

Consider what that looks like concretely. In the first month, Pi might surface a repricing recommendation that bumps against a seasonal MAP exception your team knows about but hasn't yet taught it. Your team corrects it once. That correction lands in KMS. The exception is now part of how Pi reasons about pricing on that product, permanently. Multiply that across hundreds of small calibrations, and the system a brand runs after half a year is meaningfully more aligned than the one it started with, without anyone having to re-explain the basics.

This is what separates an operational partner from a generic AI tool. A generic tool answers today's question with today's data. Pi answers today's question with everything you've told it matters, everything it has watched happen, and every decision your team has already made.

What are Skills, and how do they fit with KMS?

If KMS is Pi's memory, Skills are its playbook. A Skill is a saved, structured workflow that turns a recurring task into a single command your team can trigger by name. Pattern describes them as pre-built automations for ecommerce workflows, each one informed by your brand's knowledge base so every action stays on-brand.

Ecommerce operations are full of workflows that would benefit from this. Weekly revenue summaries. Monday competitive briefs. Monthly ad spend breakdowns. Featured offer audits. Quarterly listing content refreshes. New product launch checklists. Each of these has a defined shape: a set of steps, a specific data query, an expected output format. Historically, they've lived in someone's head or in a Notion doc that no one updates.

Skills give those workflows a home inside Pi. Once a Skill is defined, anyone on your team can run it by typing a slash command, and Pi handles the rest. The Skill pulls the relevant data, applies the logic your team has agreed on, and produces the output in the format you've standardized. Skills can run on demand when someone needs them, or on a schedule so the output arrives automatically, the same way every time.

How does a Skill actually work in practice?

Imagine your team runs a weekly Buy Box review. Today, that means an analyst pulls the top 20 ASINs, checks Buy Box percentage across the last seven days, flags any losses that exceeded a threshold, and formats the findings into a Monday morning digest. It's an hour of work that happens the same way every week.

As a Skill, that entire workflow becomes a single command. When a manager runs it, Pi pulls the same data, applies the same threshold, formats the same digest, and delivers it. The analyst who used to spend Monday morning building the report can spend it analyzing top of the report instead.

Now layer KMS on top. Because Pi knows which ASINs your team treats as priority, the digest leads with those. Because Pi knows your MAP guardrails, it flags a Buy Box loss driven by a competitor breaching MAP differently from one driven by your own price position. The Skill defines the shape of the work. KMS makes the output specific to how your brand thinks. A generic weekly report tells you what happened. This one tells you what happened in the terms your team already cares about.

The advantage compounds as your team defines more Skills. A brand running twenty saved Skills across pricing, content, advertising, and inventory is running a codified playbook of its best operational practices. New team members can execute those workflows from day one. Senior team members stop being bottlenecks for routine work. And because each Skill draws on the same knowledge base, the whole set stays consistent with how the brand operates, even as the team running them changes.

Why does the KMS-plus-Skills combination matter?

KMS and Skills work together because memory without action is just archived context, and action without memory is just automation.

A Skill running against a fresh AI would produce a generic output. A Skill running through Pi produces an output shaped by your brand's guidelines, priorities, and past decisions. The Skill defines what the workflow does. KMS defines how it should be done for your specific brand.

This pairing also changes who can do the work. When the playbook lives in Skills, and the brand context lives in KMS, expertise stops being trapped in individual people. The analyst who always ran the competitive brief a particular way has encoded that method into a Skill anyone can run. The strategic guardrails leadership set are in KMS, applied automatically. Institutional knowledge that used to walk out the door when someone changed roles now stays in the system.

The result is that Pi becomes more useful the longer you use it. Your team's institutional knowledge gets captured in KMS. Your team's operational playbook gets captured in Skills. Together, they turn Pi from an AI tool into an extension of how your ecommerce operation actually thinks and works.

How do KMS and Skills connect to the rest of Pi?

KMS and Skills aren't standalone features. They're the layer that makes the rest of Pi specific to your brand. Pi runs active sensors across featured offers, advertising, content, pricing, and inventory, and when a sensor fires, automated action loops respond. KMS is what tells those loops how your brand wants them to respond.

The same knowledge base informs the Daily Brief, so your rolling performance summary is framed around the priorities you've set. It informs Chat-to-Data, so plain-English answers reflect your guardrails rather than generic benchmarks. It informs Approvals, so when Pi surfaces a decision that needs human judgment, the context already accounts for what your brand cares about. Skills, meanwhile, let your team reach into all of that activity and run the recurring work on command.

That connection is the point. A brand's memory and its playbook shouldn't sit in a separate tool from the systems doing the daily execution. In Pi, they're the same system, which is why the actions Pi takes on your behalf stay on-brand whether a Skill ran them, an action loop triggered them, or a person approved them.

How Pattern helps

Pattern accelerates brands on global ecommerce marketplaces leveraging proprietary technology and AI. Our brand partners grow 2x faster than the Amazon category average, and Pi's memory and Skills layers are part of how that acceleration compounds over time. If your team's institutional knowledge is trapped in Slack threads and Notion docs, let's talk about what a Pattern partnership could look like for your brand.

‍

FAQ

What is Pi's KMS?

KMS stands for Knowledge Management System. It's a brand-controlled library inside Pi where your team loads its own guidelines, tone, pricing floors, MAP rules, terminology, priority ASINs, past corrections, and approved decisions. Pi reads against all of it before acting, so its answers and recommendations reflect your specific brand rather than generic best practices. You control what goes in, which means you control how Pi acts on your behalf.

What are Skills in Pi?

Skills are pre-built automations for recurring ecommerce workflows, each one informed by your brand's knowledge base so every action stays on-brand. A weekly revenue summary, a Monday competitive brief, a monthly ad spend breakdown, a Buy Box review, a launch checklist: any workflow with a defined shape can be encoded as a Skill and run with a slash command, either on demand or on a schedule.

How do KMS and Skills work together?

KMS is Pi's memory. Skills are Pi's playbook. When a Skill runs, it pulls the workflow logic from the Skill definition and combines it with the brand-specific context in KMS. The Skill defines what the workflow does. KMS defines how it should be done for your brand. Together, they turn Pi from a general AI tool into a system that reflects how your team specifically operates.

Does Pi remember overrides and corrections?

Yes. Every time a team member overrides a Pi recommendation, the correction and the reason behind it become part of KMS. This is how Pi gets sharper over time. Six months into a partnership, Pi's understanding of your brand reflects hundreds of small calibrations beyond what was documented at kickoff.

Can Skills run automatically, or only when I ask?

Both. A Skill can run on demand when a team member types its slash command, or on a schedule so the output arrives automatically at a set time, formatted the same way every run. A Monday competitive brief can land in your inbox every Monday without anyone triggering it, while an ad-hoc analysis runs the moment someone needs it.

Who defines the Skills for our brand?

Skills can be defined by your team, by your Pattern account team, or through collaboration between both. Common Skills across ecommerce operations tend to emerge from the workflows your team already runs manually. The exercise of encoding them into Skills often surfaces process improvements at the same time.

How is this different from a Notion doc or internal playbook?

A Notion doc describes a workflow. A Skill executes one. When a team member runs a Skill, Pi pulls the actual data, applies the actual logic, and produces the actual output. The playbook goes from documentation to operation, with no manual steps in between, and because it draws on KMS, the output is specific to your brand rather than generic.

Do I need to be a Pattern partner to use KMS and Skills?

Yes. KMS and Skills are part of Pattern Intelligence, which is included in every Pattern partnership. Both features are built to work with Pattern's proprietary data about your specific brand, so they require an active partnership to generate value. Pi is available to Pattern brand partners at pi.pattern.com, and prospective partners can learn more at pattern.com/pi.

‍