Cristina Cordova: Linear chose agents before a traditional enterprise roadmap

Linear’s COO explains why she delayed enterprise sales, kept support human and changed course as agents began creating work.

CCCristina CordovaGast · Linear
Kanal
Latent Space
Veröffentlicht
19. Nov. 2026
Video
1:23:32
Lesezeit
14 Min.
Original
Englisch
Auf YouTube öffnen
Beim Scrollen schrumpft das Video in die rechte untere Ecke und läuft weiter. Untertitel folgen Ihrer Lesesprache.
LesespracheIn 2 Sprachen verfügbar
Diesen Artikel gibt es noch nicht auf Deutsch, daher wird das Original (Englisch) angezeigt.
ZUSAMMENFASSUNG

In 30 Sekunden

Linear put enterprise expansion on hold when it had only one salesperson, first proving it could win growing companies already using another tool. COO Cristina Cordova says the company later set aside parts of its enterprise roadmap to build for agents: more than half of Linear issues are now created by agents, while humans still use its interface. She describes a deliberately small sales team, human-reviewed AI support drafts and a gradual approach to pricing new AI features.

  1. 1One salesperson was not an enterprise sales strategyThree weeks into the job, Cordova crossed enterprise off Linear’s goals and targeted growth-stage companies first. She wanted evidence the team could replace an existing tool before moving further upmarket.
  2. 2An agent-facing API is not enough to keep the product usefulCordova says Linear brings agent workflows into its own interface. More than half of issues are now agent-created, while the company also tracks work humans create in the app.
  3. 3Advanced customers help shape features for everyone elseLinear learned from builders using agents and incorporated their needs into features such as automatic issue triage.
  4. 4AI drafts support replies, but people send themCordova says the support team reviews and edits drafts instead of automatically sending an AI response to every customer.
  5. 5New AI features prompted usage charges alongside seat pricingLinear keeps seat pricing for its core service, charges for some new compute-intensive features and tells users when free betas may become paid.
  6. 6The enterprise roadmap gave way to agent integrationsCordova says Linear postponed some large-workspace features to build for agent workflows, believing the change could also attract enterprise customers.
  7. 7A small sales team had to earn engineers’ trustShe argues that measured hiring and product knowledge built collaboration with engineering more effectively than hiring a very large team at once.

Das Wichtigste

  1. Cordova says more than half of Linear issues are now created by agents; the company also watches work people create in its own interface.
  2. Win and support customers you can serve well before making enterprise sales a company goal.
  3. Train salespeople on the product and hire for experience selling to technical users.
  4. Use AI to draft support replies, with a person deciding what goes to the customer.
  5. Test new AI features before settling on pricing and tell customers when a free beta is expected to become paid.
  6. Cordova argues that an API alone is not enough: software needs agent workflows that make its own interface useful.
VOLLTEXT

Den bearbeiteten Text nach Kapiteln lesen

Nur das Transkript
01

The business behind a product-led tool

Cordova describes Linear’s shift from self-service startup customers to selling to larger organizations. She explains why its founders wanted business expertise.

Asked how Linear can sell its product after its CEO declared issue tracking dead, Cristina Cordova distinguishes an issue as a unit of work from manually moving that work through every step. She says customers agreed with the premise but wanted to know what they were buying instead.

Cordova is Linear’s COO. She previously led platform work, partnerships and self-service growth at Notion, after spending seven and a half years at Stripe across partnerships, sales and product.

She does not think every young company needs a COO. A founder who mainly needs someone to sell may be better served by hiring a salesperson or sales leader than an executive whose role is still unclear.

Linear’s founders were strongest in design, engineering and product. As the business expanded, they wanted help evaluating and building functions such as sales, where they did not always know what excellent performance looked like.

When Cordova joined, Linear had large customers but few above roughly 500 seats and no public-company customers, she recalls. It was better known among startups and had largely grown through self-service adoption.

Customers such as Vercel and Ramp could sign up and expand without speaking to Linear. Larger buyers often wanted a contract, pilot and evaluation before putting data into a new system, requiring a different sales motion.

02

A work trial becomes a go-to-market blueprint

Her trial used a real six-to-twelve-month business plan. Early priorities included storytelling, in-product communication and technical sales.

Linear uses work trials in hiring, and Cordova completed one herself. Her assignment was a go-to-market plan for roughly the next six to twelve months, closely resembling the actual job.

She spoke with people in sales, customer success, support and marketing while building the plan. The work tested the fit for both sides and gave her a starting blueprint once she joined.

One priority was making Linear’s brand known beyond the niches where it already had a following. The aim was to tell a more complete story and reach organizations that did not yet consider the product.

At the time, marketing mostly consisted of release emails and social posts. Fewer than 10% of users subscribed to the emails, so many active users never saw news about features; in-product notices offered another route.

Product-led growth could take the company only so far, Cordova says. Buyers choosing a system central to product development may want to build trust with people rather than make a self-service purchase.

Linear had one experienced salesperson, who later became its sales leader. The question was how to expand the team in a way that matched the product and the founders’ expectations.

Early sales hires had experience selling technical products to engineers or designers. Cordova says those customers often prefer a real demonstration over polished slides, so new hires also needed deep product training.

The founders did not want to outsource support. Since support was often a customer’s first human contact with Linear, the company built a global internal team and tried to keep its responses fast and high quality.

03

Delay enterprise until the team can serve it

Cordova replaced an enterprise goal with winning growth-stage companies. She describes how product differences and references made switches possible.

Asked whether moving upmarket must happen in stages, Cordova recalls a company offsite three weeks after she joined. She crossed enterprise off the company goals: with one salesperson, she did not believe the team was ready.

She set a nearer goal of winning growth-stage companies already using another product. Successfully replacing those tools would give her confidence that the company could pursue enterprise customers later.

Customers who are well served today can also provide positive references to larger prospective buyers. Cordova saw a similar progression at Stripe, which first won new businesses before regularly replacing incumbent providers.

Switching a core tool has costs, so Cordova does not think being modestly better is enough. Linear’s founders had used older tools themselves and understood the frustration many potential users associated with them.

The first sales pitch emphasized a substantially better experience. Demonstrations and conversations helped show why a team should replace something that was already working well enough.

04

From issue tracking to agents and human interfaces

Linear adapted its pitch as agents began doing work previously tracked manually. Cordova argues an agent API and a useful interface need to coexist.

The pitch has changed as product development workflows change. Cordova describes a traditional chain in which a person reports, labels and assigns a bug, another fixes it, and someone updates and closes it manually.

Linear has oriented more of the product around agents performing parts of that workflow. The unit of work still matters, but she expects less manual movement of tasks through the system.

Asked whether people will still need an interface, Cordova says a database exposed through an MCP server alone is unlikely to be a sufficient business. She wants useful agent workflows inside Linear as well as integrations with external tools.

Linear launched its MCP server in 2025, she says. More than half of issues are now created by agents; the company also measures work people create directly in its web, desktop and mobile applications.

Cordova says the human interface still matters because people need to understand and direct work even when agents create or carry out tasks. An agent-facing product and a product for people are not mutually exclusive.

Work can be assigned by a person or passed from one agent to another. A Linear feature called Loops can assign an agent to an issue and begin a coding task, she says.

The host asks whether the declaration that issue tracking is dead endangered Linear’s existing business. Cordova says the company is small enough to change its product rather than defend an older workflow indefinitely.

She contrasts that choice with a larger company facing a conflict between a new interface and an established revenue stream. In her view, manual issue handling is the behavior being displaced, not necessarily the issue itself.

Customers understood the claim, she says, but naturally asked what Linear would become. The company’s answer centers on coordinating work and agents across Linear and other tools.

05

Learn from advanced customers and train sellers

Frontier customers requested new workflows, and Linear turned some lessons into product features. Sales representatives must learn the product before customer calls.

The host asks whether customers requested the shift. Cordova says some frontier AI customers were already changing how they built products and wanted Linear to support their agent workflows.

Other customers were moving more slowly, but later began asking about the same changes. Linear built integrations and APIs for tools outside its own product as well as agents within it.

Cordova says listening to technically advanced builders is not simply a matter of teaching them the product. Their practices can show Linear what other customers may need next.

The company can then package those lessons for users who are not yet operating at the frontier. One request was for help triaging a backlog, a task customers did not want to maintain by hand.

Linear built triage intelligence to help route work automatically. Cordova notes that a useful capability need not always appear to the user as a separately built agent; it can be part of the product.

The host asks how salespeople learn an engineering tool. Cordova says Linear is designed for people outside engineering too, including sales and operations teams who need to understand a product roadmap.

Sales staff use Linear internally and complete product training. Representatives must pass a certification process before taking a customer call alone, she says.

Some demonstrations, such as code review and GitHub integration, are more involved because they require connected demo environments. Technical knowledge remains part of earning the customer’s trust.

06

Smaller teams, human support and technical sales

Linear avoids a separate sales-development team and uses people to review AI drafts. Solutions engineers help with integrations and migrations.

AI has also made some development tasks accessible outside engineering. Cordova says salespeople can ship website copy edits and marketers can propose changes to data handling, with an engineer reviewing the work.

Linear has tried to stay relatively flat and efficient. It has never had sales development representatives, a separate function often responsible for outbound prospecting or qualifying leads before handing them to a salesperson.

Cordova says a technical prospect may dislike spending half an hour answering qualification questions when what they want is a demo. Tooling and automation can help an account executive prepare useful outreach directly.

The customer experience team uses AI to draft responses, but a person checks what is sent. Cordova rejects a model in which every customer first receives an automatic answer and a human intervenes only after a complaint.

A draft may be longer than necessary or miss the most useful answer. Human review lets the team choose and edit the response; Cordova says fast, helpful support also shapes how customers perceive the brand.

She clarifies that Linear has not downsized the support team. AI assistance has helped the company avoid increasing headcount as quickly as it otherwise might as the business grows.

Marketing uses AI to transcribe customer interviews and find interesting passages, but people still write customer stories. Cordova says the team wants the result to feel specific and crafted rather than relying on a style of AI writing it dislikes.

Solutions engineers partner with enterprise and upper-growth account executives on deeper technical questions. Those might include custom API integrations or moving tens of thousands of users from another tool.

An engineer may build a working integration or importer as a demonstration. Showing a concrete path can make a migration less daunting than a sales promise alone.

Linear generally does not place engineers inside a customer’s systems to implement the product. Cordova says the team believes the tool should be straightforward enough for customers to integrate.

More customizable systems may require dedicated administrators and permit different teams to use the same labels differently. Cordova argues Linear’s more constrained structure can make its data easier for agents to understand.

A salesperson also helps an internal champion persuade people in IT, procurement or leadership who may not use the product every day. Engineers who like a tool do not necessarily have experience selling it internally.

07

Enterprise buying and experiments with AI pricing

Purchasing speed depends on the organization and its budget. Linear retains seat pricing while experimenting with usage charges for new AI features.

Buying timelines differ widely. Cordova describes a thousand-person company where a CEO made a top-down switch in about a month; elsewhere, approval may require repeated attempts to persuade a CTO.

Budget cycles can also delay a decision. Cordova says she does not see every established company moving faster, though an AI allocation can sometimes provide funds that were not available for an ordinary software purchase.

The host asks whether new AI budgets explain some unusually fast revenue growth. Cordova says many customers are experimenting, often through one-year deals or paid pilots rather than long commitments to a changing product.

Longer agreements can give a trusted vendor predictable revenue and a customer pricing certainty, but both sides accept tradeoffs. AI budgets and experimentation patterns may also change over time.

Cordova contrasts concentrated revenue with a business such as Linear, which she says has 40,000 paying customers. A handful of very large contracts can produce a high revenue figure much faster than a broad seat-based business.

Linear now charges for some features based on use, while its core seat-based pricing remains. Cordova says outcome-based pricing is attractive in theory but difficult when customers disagree about the value of a task such as triaging an issue.

For now, some charges reflect the costs Linear incurs. A coding session or a question that requires an agent to retrieve context may cost more than a simple action.

Agent chat is free at the time of the discussion, and Cordova says Linear uses frontier models for it. Users of coding sessions can choose between available OpenAI and Anthropic models and harnesses.

For coding sessions, she says, Linear shows the published API rate and charges a small amount on top. Loops uses AI credits, while many other AI features are included without an extra charge.

The distinction matters in customer conversations: the company is not removing seat pricing or charging usage fees for the same previously free core functionality. Some new features begin in beta while the team learns their costs and value.

Linear tells customers when a beta is free but expected to become paid, and gives notice before charging. Cordova says users may still be surprised months later, so repeated, clear communication matters.

She prefers to be conservative about initial prices and lower costs when possible, rather than underprice a costly service and later make a large increase. The goal is to build trust while learning what the model should be.

08

Change plans as the market changes

Cordova describes moving resources from a conventional enterprise roadmap toward agent workflows. She weighs speed against quality and staying close to customers.

Asked whether her original go-to-market plan worked, Cordova says many of its team-building and launch ideas eventually shipped, though not always on the original schedule. Product launches also require continued customer education.

The later agent shift changed priorities. Work planned for larger workspaces and organizational structures was delayed while Linear invested in integrations and workflows for agents.

Cordova believed those features could still win enterprise customers, just through a different route than the team initially expected. In her account, moving early also gave Linear an advantage over competitors that waited.

The company entered a difficult 2023 market in which startups were cutting costs and fundraising was harder. Because many customers were startups or growth-stage businesses, Linear had to find growth without relying on customers’ natural expansion.

When product direction changes, the business side must trust engineering, product and design to build for emerging needs, then work out how to support a business around those choices. Cordova says it is difficult to predict the exact future in advance.

Her advice is to stay close to customers while retaining the ability to change direction. The pace of model and product releases means a long, fixed roadmap can become outdated quickly.

She noticed the pace in summer 2025, when a season that had usually slowed sales and launches stayed busy. Older assumptions about avoiding launches on particular days mattered less as the market accelerated.

Fast does not mean shipping every feature immediately. Cordova says Linear sometimes delays a launch to reach its quality bar, choosing a tradeoff between speed and polish that makes sense for its position.

She believes some traditional business moats have weakened as models and competitors change rapidly. A team needs to recognize shifts and invest in people who can respond, rather than assume yesterday’s advantage will persist.

09

Hiring deliberately and growing existing leaders

She compares Linear’s small-team philosophy with lessons from Stripe. Work trials, technical headcount and support quality reflect the company’s priorities.

Why not hire a hundred salespeople and go directly after large enterprises? Cordova says that would conflict with Linear’s preference for doing more with fewer people and remaining nimble.

A sales team larger than the rest of the company would also change how the organization worked. Cordova says persuasion and human relationships are valuable, but the function had to fit the company’s product culture.

Growing steadily helped sales earn trust across the organization. Engineers and product managers know that salespeople understand customer needs and can help work through a difficult implementation.

Work trials convey the same deliberate approach to hiring: Linear wants to see whether experience gained elsewhere translates to the actual problems and working style of this company.

At Stripe, Cordova saw business hiring begin to outpace engineering and product hiring. The company adopted a rough floor of 40% for engineering, product, design and data roles to preserve its intended makeup.

The number was a management choice rather than a universal rule. If sales wanted to grow faster, it could help the company recruit engineers too; the appropriate mix depends on the kind of business being built.

Linear does not enforce the same hard threshold, but Cordova tracks the distribution of headcount. Engineering recruiting can be difficult, so she watches whether the company is investing enough in it.

One lesson she would handle differently is the temptation of a successful company to hire above a good leader simply because it can attract someone more senior. She wants to recognize existing people and give them a chance to grow.

A founder who succeeds at one scale has usually not done the next stage before, and early leaders can learn as well. Cordova says she would add outside leadership when the business truly needs capabilities it lacks, not merely because a stronger résumé is available.

She also wants to preserve Linear’s support standard as it grows. Cordova says the company is willing to pay more for that quality and that she has not received an escalation about a poor support experience in her time there.

10

What investing taught an operator

Cordova explains why she moved from Notion into investing and back to operating. She uses that investor perspective in board discussions and fundraising.

At Notion she found herself enjoying company-building work outside her original platform and partnerships remit, including helping hire people leaders, recruiters and product managers.

After filling those roles she returned to the work she had originally been hired to do and realized she wanted a different scope. Venture investing seemed to offer breadth across companies and sectors.

Investing proved different from executing inside a company. Cordova felt her operating experience was particularly useful after a startup had found product-market fit, when it needed to build out the business.

Before that stage, she says, an investor may do most good by backing strong founders, making introductions and giving them room. She personally enjoys being inside a company and getting work done.

The host asks what operating taught her about investing and vice versa. Cordova says selling a founder on an investor resembles sales: many investors offer money, so each must explain what else they bring.

Her operating background gave her a concrete way to offer help, while other investors may bring pattern recognition or a network. An introduction to someone with relevant experience can be valuable in its own right.

Returning to a company after investing also changed her understanding of board meetings and fundraising. She has a better sense of what investors want to see and what help, including customer introductions, she can ask from them.

Asked why Linear has few C-suite executives beyond its founders and Cordova, she says her role initially covered many functions in a small company. She expects leadership and titles to change as the team develops.

At roughly 170 people, Linear still tries to remain relatively flat. It gives people wider responsibility and ownership without assuming that each new scope immediately requires another organizational layer.

Mehr aus diesem Kanal

3 Folien1:25:52EN
Latent Space· 10. Nov. 2026Podcast · 2 StimmenZeno Rocha on building email tools for people and agentsGast: Zeno Rocha · Moderation: Shawn WangThe Resend co-founder explains deliverability, agent-facing interfaces and why he writes product information for machines to read.Product StrategyAI agentsProduct-Led Growth
4 Folien2:43:49EN
Latent Space· 9. Nov. 2026Podcast · 9 StimmenWhat AI Labs Buy: Expert Data, Realistic Workplaces and the Art of EvaluationGast: Garrett Lord, Sahil Bhaiwala, Nishant Ranka, Nick Magazine, Jonas Mueller, Collin Schweiker, Steve KellingHandshake AI opens its training environments and benchmarks for biology, investment banking and safety, showing where capable agents still fail and how experts find the gaps.AI agentsEvalsData quality & curation+2
3 Folien23:15EN
Latent Space· 8. Nov. 2026Podcast · 2 StimmenJay Parikh: GitHub must adapt as coding agents change developmentGast: Jay Parikh · Moderation: Shawn WangGitHub's scale, model choices and security are changing as agents join developers, says Microsoft CoreAI's Jay Parikh.Coding-AgentenEntwicklungs-Workflows & TestsInfrastruktur & Deployment+2
3 Folien1:42:28EN
Latent Space· 7. Nov. 2026Podcast · 2 StimmenJamin Ball on Anthropic's S-1, Compute Economics and the Frontier Model RaceGast: Jamin Ball · Moderation: swyxAn investor examines Anthropic's filing, from cash flow and convertible notes to the financing and competitive pressures facing frontier AI labs.Foundation ModelsAI cost & performanceInvesting+2
50:37EN
Latent Space· 6. Nov. 2026Podcast · 2 StimmenAdit Abraham on Why Document Parsing Still Breaks AIGast: Adit Abraham · Moderation: Allen ParkThe Reducto co-founder discusses missing table rows, production evaluation, document editing and growing a company while making a steak sandwich.AI agentsEvalsCompany Building
50:50EN
Latent Space· 6. Nov. 2026Podcast · 3 StimmenFabian Theis on Virtual Cells, Organoids and Closing the Lab LoopGast: Fabian Theis · Moderation: RJ Honicky, Brandon AndersonThe Helmholtz Munich researcher explains perturbation models, biological benchmarks and the work required to connect prediction with experiments.AI for scienceComputational BiologyDeep Learning+2