An illustrated example
KerjaBareng for early-stage startups
Seven people doing product, sales and support at once. What is needed is not more tools but a way to make seven feel like twelve without adding payroll.
This page is an illustration, not a case study. The company described below is one we made up as a plausible example for this kind of business — not a customer, not a testimonial. The cost figures are computed from our published plan prices.
- Team size
- 7 people: 3 engineering, 2 product, 2 commercial
- Active divisions
- Product, Data & Analytics, Sales & CS
- Tools they use today
- Linear or Notion, Slack, a metrics spreadsheet
Where it usually jams
Three things that hold early-stage startups back
Non-product work eats the founders’ hours
Writing specs, tidying meeting notes, answering the same question again — all necessary, none of it moves the product forward.
Context lives in heads, not in a system
The reason a decision was made six weeks ago is written down nowhere, so the same argument comes back.
AI tools sit apart from where work happens
Copying context from the tracker into a chatbot and pasting the answer back is its own chore, and nobody wants it.
One working day
What a day looks like once it is running
Four moments in sequence over a single day — no grand promises, just the order in which things actually happen once agents share the work.
Morning: the daily question to the main assistant
“What three things need my decision today?” is answered from the workspace itself — stalled issues, waiting approvals.
Midday: the agent drafts the feature spec
A “spec feature X” issue is assigned to the Product agent; it reads the related issues and drafts user stories for a human to check.
Afternoon: a data question gets answered without a queue
The commercial side mentions the Data agent in a channel and gets a number plus how it was computed, without waiting for the one person who writes SQL.
End of week: decisions are stored as memory
Important decisions are recorded once in Company memory, so next month nobody asks “why did we do it that way?”
The agent line-up
The agents that make sense for a business like this
Each division gets one agent. Names, instructions and permissions are yours to change — what follows is only a sensible starting point.
Bystep Agent — the main assistant
Answers cross-division questions from the workspace itself: what is stuck, what waits on a decision.
The Product agent
Drafts specs and user stories from existing issues, acceptance criteria included.
Bimo — Data & Analytics
Answers questions about numbers and explains how they were computed, rather than tossing out a figure.
What changes
- Founders get their hours back for product and customers
- The reasoning behind decisions is stored, not re-argued
- The monthly bill stays predictable while the runway is short
What it would cost
Seven seats on the Team plan: Rp 693,000 a month. Plenty of teams this size start on the Free plan first, just to prove the way of working.
The AI figure above is estimated from real run prices (Rp 130–500 per task) and is covered by the plan quota. Your exact number depends on how often your agents work.
See full pricing detailsStart working alongside your team and AI agents today.
14 days free, no credit card.