Puneet Kohli Put a Sales Pipeline Inside the Job Search

The job seeker and the recruiter stare at the same labor market through radically different windows. One has search syntax, databases, filters, a pipeline and reminders. The other has a résumé called FINAL-v7, seventeen browser tabs and the suspicion that somebody, somewhere, has invented a better way to live. Puneet Kohli noticed the imbalance while doing two jobs at once: building computer-vision systems at Apple and informally helping people with résumés and interviews. His eventual answer was Careerflow.ai, but the company began with something smaller and more revealing - a borrowed trick.

Recruiters use Boolean searches to locate candidates and hiring managers. Kohli and Nikita Gupta, a recruiter who would become his co-founder and Careerflow's chief operating officer, wondered what would happen if they put that power on the candidate's side of the table. They built a quick proof of concept in a couple of weeks and released it. About 30,000 people tried it. Nobody needed a grand theory after that. The experiment had found a queue.

Kohli left Apple. The little search utility expanded into LinkedIn optimization, job tracking, résumé building, matching, cover letters and interview practice. In April 2023, he told Founder Club that Careerflow had reached 42,000 users and was adding roughly 2,000 a day. Three years later, the company said it had helped more than one million job seekers and crossed 10 billion tokens processed through OpenAI's API. These numbers belong to different chapters, but their line of descent is unusually clean: first make one hidden professional advantage available to everyone, then follow the work that appears around it.

The funnel hiding in plain sight

One of Kohli's useful observations is almost comically unromantic: the job search resembles sales. A role is a lead. A recruiter is a contact. An application sits in a stage. A follow-up has a date. An interview advances the opportunity. Rejection closes it. This is not the language people use when speaking about their ambitions, which is precisely why the mechanics are easy to neglect. Dreams are lyrical. Pipelines require housekeeping.

Careerflow turned that housekeeping into product. The company's job tracker gives applications the sort of stage-based view salespeople have used for years. Its other tools work around the same funnel: improve the candidate's presentation, match skills to a role, generate tailored materials, prepare for the conversation and show coaches what is happening across a group. A résumé matters, but it is one object inside a moving system. Kohli built for the motion.

“We essentially figured out the key pieces of the job search process that are a pain point for job seekers, and we tried to optimise it using tools and technology.”Puneet Kohli, 2023

The idea had been accumulating before the product existed. Kohli said he and Gupta began mentoring during the pandemic. By the time technology layoffs accelerated in 2022, they knew the familiar points of friction: candidates struggled to locate the right people, tailor their story and maintain momentum through a process designed by someone else. Kohli had the technical view. Gupta had the recruiter's view. The product grew in the parallax between them.

First, teach machines to see

There is a neat continuity in Kohli's career. He studied information technology at COEP Technological University and computer science at Texas A&M. He worked as a software development engineer at Amazon Web Services, managed computer-vision engineering at Leia, and joined Apple in 2020. By 2021 he was a senior computer-vision engineer. Then he moved from helping machines see to helping candidates be seen. It is a tidy line only in retrospect. At the time, it meant leaving an established engineering role because strangers had crowded into an experiment.

The earlier technical career also explains Careerflow's instinct to convert ambiguity into signals. The platform scores LinkedIn profiles, compares skills with job descriptions and organizes applications into visible states. None of these can make a hiring decision fair, nor can software manufacture experience. They can, however, expose missing information and reduce the clerical drag around a stressful process. Careerflow's practical wager is that candidates make better decisions when the machinery stops hiding.

Careerflow joined the Techstars Oakland accelerator in 2023. The company had been bootstrapped until shortly before Kohli's April interview, when he said it had raised a little more than $100,000 from angels in its network. A 2025 ASU+GSV profile later listed $800,000 raised, a $12 million post-money valuation and investors including Techstars, Ganas Ventures, Oasis Capital, the Greylock Ventures Scout Program and Rise Ventures. By then the company reported more than 700,000 individuals served and more than 20 organizational partners.

The customer who should not carry the whole bill

Growth created a less charming problem: who pays? Kohli has been candid that a product for job seekers serves an audience living through a financially awkward moment. Charging that person may work as a transaction, but it limits the mission and the market. He has also said that revenue should have been designed into the business earlier. The internet's familiar advice - grow now, monetize in some enchanted future - looked less applicable once he began speaking with investors.

Careerflow's answer was an institutional layer. Career coaches, workforce programs and university career centers could offer the tools to their populations while seeing aggregate progress and intervening where needed. The company now presents products for universities, bootcamps, outplacement providers and workforce organizations alongside its consumer service. Its 2026 materials name Stanford, UC Berkeley, SUNY, Claremont Lincoln University, FourBlock, NuPaths and Metana among institutions using the platform.

That model also completes the sales-pipeline analogy. An individual needs order. A coach needs visibility across many individuals. An institution needs evidence that a program works. Each level asks a different question of the same underlying activity. Careerflow packages the view accordingly.

A suspiciously normal operating system

For somebody building AI software, Kohli has amusingly little interest in performing technological sophistication. He calls himself old school about internal tools. Google Workspace is a default. Apple Notes holds the notes. When people advertise their fluency with Notion, his response is that he would rather spend that time building the business. It is the sort of opinion that creates either instant relief or a twelve-tab rebuttal.

The preference suggests a useful split in his thinking. Put complexity where it changes the user's outcome. Keep the founder's own machinery plain enough to disappear. Kohli applies a similar filter to learning. He told Founder Club he is not much of a book person because he struggles to finish them; videos, podcasts and Y Combinator material suit him better. He named The Good Time Show and its co-host Aarthi Ramamurthy, who advised Careerflow, among the content and connections he valued. Silicon Valley, where he had lived for more than five years by 2023, provided a network he could call when stuck.

“All I need is some notes that I can look at and so I just use Apple notes on my computer.”Puneet Kohli on productivity tools

His public recommendations add another dimension. A career coach described him joining her Slack group to answer users' Careerflow questions. A participant in a FourBlock military-transition cohort recalled Kohli mentoring the group on LinkedIn and walking them through the product. These are small scenes, not scale metrics. They show a founder remaining close enough to watch confusion happen in real time, which is often where the next feature is hiding.

The platform keeps widening

In January 2026, Careerflow said it had passed 10 billion OpenAI API tokens, a marker reflecting résumé generation, profile analysis, matching and mock-interview activity. Kohli described the milestone as evidence of user and partner trust. Later that year, the company announced a human-evaluation partnership with Online Mind2Web, a benchmark for browser-using agents, and described a vetted expert network spanning more than 100 domains. Career technology had begun to overlap with the less visible work of evaluating AI itself.

Kohli's own newsletter writing has become more direct about that labor market. He discusses fraudulent recruiters, AI-mediated hiring and what laid-off workers should do with the finite runway of severance. The tone is tactical because the market is not theoretical to his readers. In one July dispatch, he noted that Careerflow's own hiring team encounters applicants using stolen headshots and face-swaps, while users struggle to decide whether recruiters are real. Founder and customer now stand on opposite sides of the same fog.

The aspiration Kohli named early remains intact: a holistic solution for the career cycle. The phrase risks sounding like every platform diagram ever drawn, yet Careerflow's origin keeps it grounded. It did not begin by declaring ownership of a category. It began by moving one tool across a market boundary and watching 30,000 people reach for it. The job search will always contain uncertainty. Kohli's project is to remove the uncertainty caused merely by poor organization, asymmetric tools and invisible process. Ambition is difficult enough without FINAL-v8.