See what Kylon does, map how your business runs, and pick the first job to hand to AI.
Time: about 45 minTake away: a business analysis of your company
LESSON
01
Five lessons · Lesson 1
01Course mapLesson 01 · Make your company AI-native
Five lessons. This is lesson 01
Five lessons, in the order an owner decides: go AI-native, redesign how the company works, give AI employees their roles, capture your data and know-how, then launch your first AI-driven project.
01Make your company AI-nativeTake awayA business analysis, and the step to start with
02Redesign how your company worksTake awayA company design map
03Give AI employees their rolesTake awayAn AI employee on the job
04Capture data and know-howTake awayOne App and one Skill
05Your first AI-driven projectTake awayA weekly workflow, and the evidence to decide on rollout
01Where many companies are todayLesson 01 · Make your company AI-native
Everyone uses AI. The company isn't any faster.
Each person opens their own chat window and moves on. The know-how, files and progress never stay with the company.
01A little historyLesson 01 · Make your company AI-native
Business software has gone throughfour generations in forty years
Each one took a type of work off people's hands.
1980sSpreadsheetsCompanies gotOne person doing the numbers and tables much faster
1990sERP and CRMCompanies gotOne record for orders, customers and stock
2010sCollaboration toolsCompanies gotThe whole company talking and sharing files online
2022Chat AICompanies gotAn assistant everyone can ask and have write
NowAI employees (Agents)Companies gotAI with a role, connected to company systems, getting work done
VisiCalc, the first spreadsheet, came out in 1979. Salesforce moved CRM online in 1999. Slack launched in 2013. ChatGPT was released in November 2022.
01What changed this timeLesson 01 · Make your company AI-native
Old softwarekept records, this generationgets work done.
Software used to need a person to run it. An Agent takes a task, looks things up, does the work and hands it back for a person to confirm.
01CompareLesson 01 · Make your company AI-native
All of them use AI.Three ways to use it,and how they differ
Q&AChat window
How you use itYou ask, it answers
What it can reachWhat you paste in
Where it staysYour personal chat history
Who checks itYou
AssistantAI inside office software
AI
How you use itHelps you write inside one app
What it can reachWhat's in that app
Where it staysPersonal documents
Who checks itYou
AI employeeA Kylon Agent
# sales
Approve
How you use itGive it a role, it gets the work done
What it can reachCompany-approved email, sheets and systems
Where it staysThe company's Rooms and Apps
Who checks itA named person approves
01The tech in three sentencesLesson 01 · Make your company AI-native
Large language models:a new hire who has read a lot
01It continues text
You sayHelp mewritea
It continuesfollow-upemail
Trained on huge amounts of text, it can write, summarize and organize
02It doesn't know you
LLM?
Your company
Tools
People's know-how
SKILL
Skill
DoDon't
Judgment
Past conversations
It knows none of this until you connect it
03It can be wrong
Draft
Confirm
Anything external starts as a draft for a person to confirm
01The tech, one step furtherLesson 01 · Make your company AI-native
An Agent ismade of four things
LLM
Brain: understands and writes
Agent
Tools and connections
Hands: email, sheets, CRM
Memory and knowledge
Company files and know-how
Rules and permissions
What it can see, who approves
Approve
01The tech, last stepLesson 01 · Make your company AI-native
An Agent does a taskin four steps
Step 1
Gets the task
Someone asks, or it's time
ResultA clearly defined task
Step 2
Looks things up
Reads sheets, email and past discussions
ResultContext
Step 3
Does the work
Drafts, organizes, updates
ResultA draft or an update
Step 4
Hands back the result
External actions wait for approval
ResultThe result and a record of the work
People explain the task and sign off on anything external. The rest goes to the Agent.
01CompareLesson 01 · Make your company AI-native
Which work to hand over,and which to keep with people
Hand to an Agent
Repeated work with a set method
Weekly follow-ups and reminders
Questions that take a lot of digging
Organizing, summarizing, drafting
Keep with people
Decisions and accountability
Commitments and quotes to customers
Exceptions nobody has seen before
The final yes
01Remember thisLesson 01 · Make your company AI-native
With AI on board, your company gainsstaff who keep the work moving.
01What Kylon isLesson 01 · Make your company AI-native
Your people and AIworking in one place
01
The workspace is the company, a Room is a department
People and Agents are both members. They hand work along in the same Room, so whoever picks it up sees where it stands.
02
An Agent is an AI colleague with a role
It connects to email, sheets and your usual systems and does its daily work. Anything external becomes a draft and only goes out when a person approves.
03
Apps, Skills and Automations keep the work in the company
Business data lives in Apps, methods are written as Skills, and weekly repeat work goes to Automations.
01Company harnessLesson 01 · Make your company AI-native
What you see is chat.What makes it reliable is below the surface
Above
Chatting with AgentsRooms and discussions
AppResults waiting for approval
What it runs on
Six layers below the surface decide whether you can trust an Agent
01Identity and permissions
Every Agent is a member with a role
02Agent routing
Decides which Agent takes the work, and where
03Connections
Real accounts, keys never in chat
04Runtime
Isolated, runs on a schedule or an event
05Build and launch
Apps from draft to live
06Governance
People approve external actions, every run is logged
Company harness: connecting your company's identity, data, workflows and records to AI.
01For ownersLesson 01 · Make your company AI-native
AI for one person,vs. Kylon for the company
AI for one person
A chat window each
Files and know-how stay with individuals
Everyone asks on their own
A new person starts from zero
vs. Kylon for the company
One shared workspace
Workspace
# sales
# ops
# finance
Know-how and data belong to the company
People and Agents hand off in the same Room
Whoever picks it up sees how far it got
01CompareLesson 01 · Make your company AI-native
Kylon and the toolsyou already use
Team chat
DoesPeople talk online
Can'tGet AI to finish the work
Automation tools
TriggerStepStep
DoesRuns fixed workflows automatically
Can'tHandle exceptions nobody wrote in
Personal AI assistant
DoesMakes one person faster
Can'tKeep the know-how in the company
Kylon
# sales
AppAutomation
DoesPeople and Agents finish the work together
Also keepsKnow-how and data in the company
Drawn from typical use of each kind of tool. Individual products vary.
01A few words firstLesson 01 · Make your company AI-native
How your companyruns in Kylon
People and Agents can join several Rooms. Skills travel with the Agent. Apps, Automations and Webhooks belong to one Room.
Workspace = companyThe circled Agent is the same one, in two Rooms
People
SkillSkillSkillAgent Its Skills go with it
# sales
Quote follow-upApp
Follow up daily at 9:00Automation
Lead from the website formWebhook
# ops
ScheduleApp
Plan every MondayAutomation
Notice from the order systemWebhook
# finance
ReconciliationApp
Month-end reconciliationAutomation
Payment received noticeWebhook
WorkspaceCompanyOne per company
RoomDepartment or projectSales, ops, finance
AgentAI colleague with a roleJoins several Rooms
SkillCompany methodsGoes with the Agent
AppBusiness ledgerLives in one Room
AutomationScheduled workLives in one Room
WebhookOutside noticesLives in one Room
01Meet AppsLesson 01 · Make your company AI-native
App: Turn spreadsheets into a small system the team shares
Sheets scattered everywhere
Quote follow-upApp · sales Room
CustomerStatusNext follow-upOwner
Aster DesignTo follow upThu
Birchwood HomeQuotedNext Mon
Northbank HotelWon
Built from one sentence
Tell an Agent what to track and it builds it.
Every record has its own discussion
All the back and forth with a customer sits under that record.
Agents work from the fields
Status and next follow-up tell it what to do.
01Meet SkillsLesson 01 · Make your company AI-native
Skill: How-to instructions written for Agents
Your veteran's method
SKILL · Quote follow-up
When to use
Step
Common mistakes
Agents that have it follow it
Correct it once in chatSkill updatedEvery time after, the new way
Like a job manual
Write down how we do this here.
Write once, give it to the Agents that need it
Every Agent that has it works the same way.
Gets sharper with use
Correct it once and it's recorded. No need to say it twice.
01Meet AutomationsLesson 01 · Make your company AI-native
Automation: Give an Agent a goal, not a script for every step
Fixed-step workflow
TriggerStep 1Step 2Step 3?A case nobody wrote in Stops here
Kylon Automation
On schedule, or a Webhook
Give it a goal
Check recordsDraft follow-upAsk a personPicks the next step for the situation
ReportsThe Agent posts them to the Room
SendExternal email and messagesA person clicks Send
The Agent bringsRoom rulesCompany knowledgeMemorySkill
Goals, not scripts
It picks the next step from what it finds.
Uses your know-how
Room rules, knowledge, Memory, Skills.
Two kinds of output
Reports post themselves. External messages wait for Send.
Kylon can run fixed-step workflows too, as code Automations. Automations don't promise to fix every error on their own.
01Live demoLesson 01 · Make your company AI-native
Have an Agent analyze your company
kylon.io
1
Tell the Agent what the company does
Give it your website and a short intro: products, customers, departments
2
It maps the business workflows
What each department repeats every week, and which systems it uses
3
It flags the best first step to hand to AI
The owner reviews it and decides where to start
01Demo resultLesson 01 · Make your company AI-native
The Agent's analysis:where to start, at a glance
Ada · Business analysisAcross the three departments, sales quote follow-up repeats every week and is the easiest to miss. Start here.
01SalesRepeats weeklyFollowing up after a quote goes outHand to AI first
02OpsRepeats weeklyScheduling and delivery remindersSecond wave
03FinanceRepeats weeklyInvoicing and reconciliationAccounting softwareWatch for now
Example. In class, run it live with each attendee's own website and intro.
01How to chooseLesson 01 · Make your company AI-native
The first task for AIshould meet three conditions
01
It repeats every week
Frequent enough that the time saved shows.
02
It has a set method
A veteran can explain every step.
03
Mistakes are easy to spot and fix
Start with internal work, or work that produces a draft first.
01Hands-on: four diagnostic questionsLesson 01 · Make your company AI-native
Map your companywith four questions
Question
What it finds
Your answer
Which processes make money?
The steps a sale really depends on
What does the team repeat every week?
Work that's settled enough to hand over
Which step goes wrong or stalls most?
The first gap to close
How much staff time does it take now?
A baseline to compare against later
Let an Agent fill in a first pass, then the owner and team leads edit it.
01Common questionsLesson 01 · Make your company AI-native
In the first few days,the three questions we hear most
Q1
What do I do after signing up?
Have an Agent map the company and pick the first task.
Q2
What's the difference between a Room and a thread?
A Room is a department or project. A thread is one piece of work inside it.
Q3
Why does the Agent keep asking questions?
The task isn't clear yet: what to do, how far to go, who approves.