Linear+41 Positive

Operator Insights

What are engineering leaders learning from deploying Linear?

7 accounts: 1 left it or turned it down and 6 adopted it

Left it or turned it down1

obiefernandezHacker News commenter describing ZAR's toolingZAR
Professional team. Product engineering at ZAR. Comment posted March 24, 2026.
Hacker News, published Mar 24, 2026Read the write-up (opens Hacker News in a new tab)
Observation
ZAR dropped Linear for a homegrown experiment-tracking system once it had an organization-wide knowledge graph.
Takeaway
If your team plans around experiments and has stopped working from a backlog, test whether a tracker still earns its place before you renew.
What happened
The commenter says that once ZAR had an organization-wide knowledge graph, its next step was to drop Linear and replace it with a homegrown experiment-tracking system that points product engineers at empirical data, arguing that working from a backlog is counterproductive.
- Public practitioner write-up (Hacker News comment)
- One commenter's account
Deployment details
Professional team. Product engineering at ZAR. Comment posted March 24, 2026.
- Public practitioner write-up (Hacker News comment)
- One commenter's account
Team
Product engineers
Workload
Prioritizing product work
Product version
Linear, replaced by an in-house system
In SignalsCounted once, as the commenter's Hacker News record. In the record

Adopted it6

Shashank SinglaFounderNiptao
Small team. Software startup. Write-up published June 10, 2026.
Niptao blog, published Jun 10, 2026Read the write-up (opens Niptao blog in a new tab)
Observation
Niptao runs a Claude Code loop that uses Linear as its work queue, state machine and memory.
Takeaway
Give an agent loop one place for state; Linear works as the queue when every request becomes a ticket.
What happened
Niptao turns bug reports from a group chat into Linear tickets, and a Claude Code loop reads them over MCP, using Linear as its work queue, state machine and memory. The post reports the first two fixes from the loop merged.
“Linear is the work queue, the state machine, and the memory.”
Shashank Singla
- Public engineering write-up
- Two merged fixes reported
Deployment details
Small team. Software startup. Write-up published June 10, 2026.
- Public engineering write-up
- Two merged fixes reported
Team
Small engineering team
Workload
Bug fixes from chat reports
Product version
Linear over MCP with Claude Code
Measurement
Merged fixes (2)
In SignalsCounted once, as Niptao's own engineering post. In the record

RampAug 31, 2026Linear (customer story)Linear's customer story credits Ramp's in-house agent with three of every four merged pull requests, working from six years of specs and customer requests kept in Linear.
Zach BruggemanCo-creator of Ramp's Inspect agentRamp
Fintech. 1,600 people. On Linear since 2019. In-house background agent since November 2025.
Linear (customer story), published Aug 31, 2026Read the write-up (opens Linear (customer story) in a new tab)
Observation
Linear's customer story credits Ramp's in-house agent with three of every four merged pull requests, working from six years of specs and customer requests kept in Linear.
Takeaway
GILD's read: the tracker's history is the agent's context, so specs and customer requests written for people pay off again when an agent traces a bug through them. The cost moves to review; size reviewer time before you widen delegation.
What happened
In November 2025 Ramp handed most of its code to Inspect, a background agent three engineers built in two weeks. Each session runs in a sandbox on Ramp's own infrastructure and starts from the web, a Chrome extension, or Slack. Through Linear's agent API, Inspect reads the specs, customer requests, and roadmaps Ramp has kept in Linear since 2019. This lets it trace a bug from a customer request to a spec to the code. Linear's story says a coding agent wrote about 90% of that integration in 30 minutes, and that Ramp's bottleneck moved from writing code to reviewing it.
Vendor customer story, published with Linear's film
Vendor-reported merge share
Deployment details
Fintech. 1,600 people. On Linear since 2019. In-house background agent since November 2025.
Vendor customer story, published with Linear's film
Vendor-reported merge share
Team
Ramp engineering; designers also start sessions
Duration
Since November 2025
Workload
Most code changes at Ramp, including bug fixes traced from customer requests
Product version
Ramp's Inspect agent connected through Linear's agent API
Measurement
Share of merged pull requests written by Inspect (75%), as reported in the story
In SignalsNot counted. Linear published this story, and Ramp's own account of Inspect never mentions Linear.
CoinbaseAug 11, 2026Linear (customer story)Coinbase made Linear the source of truth its agents read, then asked engineers to write no code by hand for two weeks.
Chintan TurakhiaLeads AI accelerationCoinbase
Crypto exchange. 3,800 people. On Linear since November 2025. In-house agent called Forge.
Linear (customer story), published Aug 11, 2026Read the write-up (opens Linear (customer story) in a new tab)
Observation
Coinbase made Linear the source of truth its agents read, then asked engineers to write no code by hand for two weeks.
Takeaway
GILD's read: an agent can only pick up work the tracker describes, so Coinbase moved requirements, bugs, and status into one record before it scaled Forge. The trade is discipline: every request has to land in Linear in a form an agent can act on, or the agent works from a partial picture.
What happened
Turakhia told the team from day one that agents would depend on Linear, so Coinbase uses Linear for requirements, designs, bug reports, shipping status, and team structure. In January 2026 Turakhia asked the engineering organization to delete their IDEs and write no code for two weeks. A bug flagged in Slack becomes a Linear issue that is labeled, sized, and assigned. Forge, Coinbase's own agent, drafts a fix and posts the pull request back to Slack for review. Linear's story reports that in near-weekly Speedrun sessions, 120 engineers produced 80 pull requests in one sitting.
Vendor customer story, published with Linear's film
Vendor-reported session counts
Deployment details
Crypto exchange. 3,800 people. On Linear since November 2025. In-house agent called Forge.
Vendor customer story, published with Linear's film
Vendor-reported session counts
Team
Coinbase engineering and product organization
Duration
From November 2025; two-week IDE-free experiment in January 2026
Workload
Bug fixes, tests, and pull requests drafted by agents from Linear issues
Product version
Linear's agents first, then Coinbase's Forge agent called from Slack, GitHub, or Linear
Measurement
Pull requests per Speedrun session (80 from 120 engineers), as reported in the story
In SignalsCounted once, via Chintan Turakhia's own Arize Observe 2026 talk. In the record
daily.devMar 23, 2026daily.dev blogdaily.dev logs all its bugs, feedback and tasks in Linear and built a coding agent that takes delegated Linear tickets.
Ido ShamunEngineering teamdaily.dev
Professional team. Developer news platform. Write-up published March 23, 2026.
daily.dev blog, published Mar 23, 2026Read the write-up (opens daily.dev blog in a new tab)
Observation
daily.dev logs all its bugs, feedback and tasks in Linear and built a coding agent that takes delegated Linear tickets.
Takeaway
Keep tickets in Linear rich enough to hand to an agent, and connect the agent through Linear's native agent API.
What happened
daily.dev logs bug reports, user feedback and feature tasks in Linear and built Huginn, a coding agent that takes tickets delegated in Linear, plans across repositories and opens pull requests through Linear's native agent API.
“At daily.dev we log everything in Linear: bug reports, user feedback, feature tasks.”
Ido Shamun
Public engineering write-up
Architecture described; no metrics
Deployment details
Professional team. Developer news platform. Write-up published March 23, 2026.
Public engineering write-up
Architecture described; no metrics
Team
Engineering
Workload
Bug reports, feedback and feature tasks delegated to an agent
Product version
Linear with its agent API
In SignalsCounted once, as daily.dev's own engineering post. In the record
CommureSep 10, 2025Linear blogCommure reads Linear dashboards daily for triage and SLA misses, and less often for where engineering time goes.
Myles BaumannSenior Product ManagerCommure
Healthcare technology. Software for over 400,000 clinicians. One product, many teams. Linear Dashboards on Enterprise.
Linear blog, published Sep 10, 2025Read the write-up (opens Linear blog in a new tab)
Observation
Commure reads Linear dashboards daily for triage and SLA misses, and less often for where engineering time goes.
Takeaway
GILD's read: a dashboard earns its place when it answers a decision made on a schedule, such as SLA misses or unplanned work. Dashboards are on Linear's Enterprise plan, and per-person cycle time is a reason to ask a question, never a score.
What happened
Commure uses Linear dashboards daily to stay on top of triage, SLA performance, and postmortem follow-through. One tracks high-priority issues over time by team. Another shows how often teams hit P0, P1, and P2 SLAs and the lead time at each severity. An investment chart splits effort between projects and unplanned work such as feature requests and operations. A cycle-time-by-assignee chart is a prompt to ask about poorly scoped tickets or hidden blockers, and Commure says it is not a performance metric on its own.
Vendor customer story
No results published
Deployment details
Healthcare technology. Software for over 400,000 clinicians. One product, many teams. Linear Dashboards on Enterprise.
Vendor customer story
No results published
Team
Commure product and engineering teams on one product
Workload
Triage, severity SLAs, postmortem follow-up, and planning across teams
Product version
Linear Dashboards and Insights on the Enterprise plan
In SignalsNot counted. Linear published it on its own blog, and Commure has published no account of its own, so GILD counts it nowhere in the Signals.
Damian GalarzaFeb 13, 2026damiangalarza.comA developer runs a Claude Code loop that takes the next Linear issue, builds it and opens a pull request.
Damian GalarzaDeveloperCreatorSignal
Individual. A project built on live streams. Write-up published February 13, 2026.
damiangalarza.com, published Feb 13, 2026Read the write-up (opens damiangalarza.com in a new tab)
Observation
A developer runs a Claude Code loop that takes the next Linear issue, builds it and opens a pull request.
Takeaway
Write Linear issues an agent can finish on its own, then let the loop pull the next one.
What happened
Galarza runs a Claude Code agent loop on his CreatorSignal project that picks the next Todo issue from Linear over MCP, builds it, reviews it and opens a pull request.
“I decided to give this a try on my recent project CreatorSignal that I’ve been building during my live streams.”
Damian Galarza
Public practitioner write-up
Personal project
Deployment details
Individual. A project built on live streams. Write-up published February 13, 2026.
Public practitioner write-up
Personal project
Team
One developer
Workload
Building features from Linear issues
Product version
Linear over MCP with Claude Code
In SignalsCounted once, as the developer's own post. In the record

Left it or turned it down1

ZAR dropped Linear for a homegrown experiment-tracking system once it had an organization-wide knowledge graph.

Adopted it6

Niptao runs a Claude Code loop that uses Linear as its work queue, state machine and memory.

Linear's customer story credits Ramp's in-house agent with three of every four merged pull requests, working from six years of specs and customer requests kept in Linear.

Coinbase made Linear the source of truth its agents read, then asked engineers to write no code by hand for two weeks.

daily.dev logs all its bugs, feedback and tasks in Linear and built a coding agent that takes delegated Linear tickets.

Commure reads Linear dashboards daily for triage and SLA misses, and less often for where engineering time goes.

A developer runs a Claude Code loop that takes the next Linear issue, builds it and opens a pull request.