How Work Gets Done
Circleback's eight-person team built its automations feature after Haghani noticed generic notes failed for different customer types; he built the extraction logic, showed it to Jacyna, and they shipped it — no PRD, no product manager, no review cycle. This profile maps how Circleback actually operates day to day, what principles govern it, and what kind of candidate the culture rewards or rejects, so you can judge fit before applying.
The rhythm is set by the two founders — Haghani, a former Stripe engineer, and Jacyna, from Tableau — who still write code, triage support, and run the business on their own product. Haghani uses Circleback as his personal ATS: he holds the action button on his phone to start recordings, reviews meeting prep generated from past emails, and uses the watch-app complication he built to capture in-person conversations without friction. When they say "we" in interviews — "we rolled out automations," "we realized general notes won't serve all customers", they mean the two of them and the engineers who pick up tickets. No product manager translates specs; the spec is the conversation they had with a customer that morning.
Decisions move at the speed of a Slack huddle. The automations feature, prompt-driven extraction of entities, action items, and problem areas from every meeting, shipped after he noticed they failed across use cases. He didn't write a PRD; he repeated the process, and they pushed it. The same pattern holds for privacy controls: Haghani has said they think "a lot about" ensuring one-on-ones don't leak to the whole company, and that concern shapes the sharing model. Authority is earned by shipping. The engineer who designs the watch complication owns it. The founder who feels the pain of context-switching owns the capture flow.
The team is distributed: job postings list San Francisco for every open role, while BuiltIn lists Vienna as headquarters, but the operating cadence is synchronous by default. Y Combinator's Winter 2024 batch reinforced weekly user calls. Angel investor Taggar told TechCrunch the output "felt like there was a human editor at play who curated key points." Glassdoor reviewers rate work-life balance 3.3 out of 5 and culture 3.4. The trade-off is explicit: zero bureaucracy, but zero cover. If the iOS app slips, the person who owns mobile feels it.
Compensation mirrors that transparency. Zero G Talent's board shows six active roles: Design Engineer, Senior Product Engineer, Senior Software Engineer, Product Engineer, Software Engineer, Founding Marketer, all San Francisco-based with bands from $140,000 to $280,000 and a median around $240,000. No equity ranges are published, but the $2.5 million seed round (Transpose Platform, Rebel Fund, Pioneer Fund, and angels including Taggar, Oliver Jung, and JJ Fliegelman) suggests standard early-stage grants. The salary spread is narrow because the team is narrow; a founding marketer sits near the senior product engineer, and both know what the other makes.
The operating principle is simple: the person closest to the problem decides. Haghani put it plainly in a 2026 founder video: "what matters more is not how fast you ship, but what you decide to build and how you decide to build it in order to be able to continue composing the systems that you build together." At eight people, that composition happens in real time, in public channels, with the founders in the mix.
The Operating Code
Circleback's principles read like a manifesto for a team that treats AI not as a feature but as the substrate of daily work. Haghani describes a culture built on three interlocking commitments: use the best tools without asking permission, make every task easier the next time someone does it, and hold the product to an opinionated standard that admits no "good enough."
The unlimited-AI-budget rule is the most visible expression of the first commitment. "Right now, we have an unlimited budget for using AI. And we trust the judgment of everyone on the team," Haghani said in a 2026 walkthrough of his workflow. The caveat, "Maybe that doesn't scale when we're 10x the size", is stated openly, not buried in a policy doc. The principle is judgment over guardrails: if an engineer believes a model call improves output, they make it. That trust extends to ops automation. Haghani automated payroll logins and other administrative friction so he could stay in "high-leverage" work; the same permission applies to every hire.
The second commitment, incremental improvement without whack-a-mole, shapes how the team evaluates product changes. "Because if you don't have good systems in place to make incremental improvements, you're constantly playing this game of whack-a-mole: maybe we made the overview better, but now the topics are too long." The solution is evals for anything quantifiable: action-item detection must distinguish "I'll send this now" (already done) from genuine follow-ups. The eval suite is the guardrail that lets the team ship fast without regressing on the thousand small quality dimensions that define the note-writing experience.
Which brings the third commitment: opinionated quality. Circleback's notes are engineered to read like a person wrote them. The team maintains a living style guide: never use "discussed" ("Obviously, things were discussed." It's a meeting. It doesn't add any value"). "Everything — not just code quality, but architecture, product, even the copy we put in — goes through that AI review to ensure consistency with how we usually do things."
The architect-versus-builder distinction governs how agents are deployed. "We don't want the agent architecting the thing. The engineer should architect it and then have the AI agent build it out." Email drafts are fine; autonomous sends are not. Data-access and security reviews get a dedicated sub-agent pass, but the human remains the decision point. This pattern — human intent, machine execution — repeats across support, sales, and recruiting. A weekly engineer-on-support rotation forces every developer to hear raw customer feedback; inbound sales and support threads spawn in shared Slack channels so the whole company sees what prospects ask and why deals stall. Personal agents (finance, people-ops, customer-intel) are being migrated into those same channels, turning private tooling into collective memory.
Privacy is the counterweight to capture-everything. "Giving our customers confidence that the things captured will be shared with the right people, and nothing shared inadvertently" is framed as a product requirement, not a compliance checkbox. The sharing layer, who sees a one-on-one transcript, is treated as a trust problem as much as a technical one.
Haghani's calendar reflects a final principle: protect maker time. Thursdays are meeting-free by design, reserved for end-to-end shipping of whatever has been queued all week. The founder has consciously weaned himself off daily coding because "our bottleneck is bringing on more great people to increase our throughput of product decisions." The organization's operating principle, in miniature, is that the highest-leverage work is deciding what to build and how to compose systems so they interoperate — not the typing itself.
The Hiring Filter
Circleback's hiring bar is built for a team that has never had a recruiter. Founders run every search themselves, which means the signals they trust are the ones they can verify in minutes. TheAntiJobBoard notes explicitly that Circleback has no recruiting function; founders handle hiring alongside everything else. That constraint shapes every filter.
The first filter is specificity. Applications that mirror the language of the job description clear automated filters at a materially higher rate, per TheAntiJobBoard's analysis of Circleback's Ashby pipeline. A recycled CV gets rejected fast; they notice. The company's own job posts:
| Role | Salary Range (San Francisco) | Source |
|---|---|---|
| Design Engineer | $170,000 – $280,000 | Zero G Talent |
| Senior Product Engineer | $170,000 – $250,000 | Zero G Talent |
| Senior Software Engineer | $170,000 – $250,000 | Zero G Talent |
| Product Engineer | $145,000 – $240,000 | Zero G Talent |
| Software Engineer | $145,000 – $280,000 | Zero G Talent |
| Founding Marketer | $140,000 – $200,000 | Zero G Talent |
each describes a distinct problem set. Candidates who lead with Circleback's problem, not their own ambition, advance. TheAntiJobBoard frames it bluntly: Circleback is focused on growing; show you understand that.
Autonomy is the second filter. TheAntiJobBoard lists "relevant experience and ability to operate autonomously" as the core criteria. In a sub-50-person B2B productivity company, there is no onboarding ramp. The interview emphasizes relevant experience and culture fit over abstract problems; a startup this size wants proof you can ship, not that you can whiteboard. Circleback does not include a take-home stage, a deliberate choice that removes a common proxy for "can do the work" and replaces it with a paid two-day work trial in San Francisco. That trial is the highest-fidelity signal in the loop: you either produce usable output alongside the team or you don't.
Persistence compounds. The median B2B productivity application gets no response ever. One follow-up at day five roughly doubles reply rates, per TheAntiJobBoard's benchmarking. At Circleback, where the median first-contact-to-offer window is seven days (three days faster than the category median), a candidate who follows up once signals the same urgency the founders operate with. Silence is not a strategy here.
The compensation band reflects the bar. Zero G Talent's board data shows eight salaried roles with a typical range of $95k–$259k (median $240k). The spread is wide because the roles are not leveled against a public framework; they are priced to the specific leverage each hire unlocks. A Senior Software Engineer is expected to own entire product surfaces. A Founding Marketer owns the growth narrative. The numbers are not bands — they are bets.
What gets rejected is equally legible: generic applications, candidates who need structure to produce, anyone treating the process as a numbers game. The hiring bar selects for people who have already done the work of understanding Circleback's exact problem, can demonstrate they've solved comparable ones without supervision, and will follow up once because they know the cost of delay. The paid work trial then confirms it in real time.
Two Companies, One Name
The employee feedback available for Circleback splits cleanly across two different businesses that share the name, a distinction that matters for any candidate evaluating fit. The bulk of public reviews on Indeed describe Circleback Lending, a consumer-lending business that ceased operations between 2017 and 2022. A separate, smaller body of first-hand accounts, primarily a 2024 YouTube walkthrough featuring Ali Haghani, describes the frontier-tech AI company (Y Combinator Winter 2024, roughly eight people) that is the subject of this profile. Glassdoor and SimplyHired listings do not clearly separate the two entities. Below, each account is attributed to its source and date so you can weigh relevance yourself.
Circleback Lending (2015–2022): the defunct lending business
Praise. Multiple reviewers called the workplace "wonderful," "comfortable," and "family-like" in the earlier years. A May 2015 review noted "great people at the top that are thrilled about hearing your thoughts and suggestion about change and culture." A November 2016 reviewer wrote, "I loved getting up and going to work… Great management and c-level executives took care of their employees." An April 2020 review highlighted schedule flexibility and culture: "My favorite is probably the flexibility I have with my hours. I also love the company culture and enjoy working with everyone." A May 2022 review added that HR "was quick to shut it down" when drama appeared. A September 2017 review on the Circleback Inc. page (distinct from the Lending pages) described "monthly meetings… to make sure everyone is up to date" and "relaxed work schedule. Flexible management."
Criticism. The tone shifts sharply in 2016. An April 2016 review stated: "The Management and Human Resources Department is horrible… Human Resources should advocate for the employees, rather than shut them out." A March 2016 review alleged "No respect for employees… Years of service means nothing… certain employees are given incentives when the rest are not." An August 2016 review said: "Everything at the company works, except for the leaders… potential to be very successful if leadership at the top… can be improved." By December 2016, a reviewer noted the company "needed better management structure in order to survive." A January 2017 review reported: "Management was changed and an outside team was brought in to eventually sell the business." Reviews from 2019, 2021, and 2022 uniformly cite the closure as the reason they would not recommend the company; one 2019 review said "the company could not find additional funding to stay afloat."
Aggregate scores (Indeed, 14 reviews). Work-life balance 3.9/5; Pay and benefits 3.8/5; Culture 3.8/5; Management 3.3/5; Job security and advancement 2.4/5, the last figure reflecting the shutdown.
Circleback (AI meeting-intelligence product): current team, 2024 evidence
The only direct, attributed employee accounts for the current company come from a 2024 YouTube video featuring Ali Haghani discussing daily operations. Each point below is attributed to "the 2024 team video."
The 2024 team video confirms the operating rhythm detailed above: no-meeting Thursdays, weekly support rotations, internal show-and-tells, an unlimited AI-tool budget, eval discipline for quantifiable outputs, and a deliberate migration of processes from private tooling into shared Slack channels. The founders also name hiring as the current bottleneck and describe a cultural maxim of making every task easier for the next person.
Third-party product validation (not employee sentiment, but relevant signal)
Time.com's June 2024 "Best AI Note-Taking Tools" review named Circleback its "top pick for an in-meeting note-taking bot for its excellent summarization and speaker recognition features, as well as streamlined interface and premium features," noting it is "among the most expensive of the tools we tested" but "worth it."
Glassdoor and SimplyHired: ambiguous attribution
Glassdoor lists 16 reviews for "Circleback" (no "Lending" suffix) with no dated excerpts provided in the research. SimplyHired shows a "Circleback" entity headquartered in Vienna with 11–50 employees and salary bands (Software Engineer $180k, Senior Software Engineer $210k, Design Engineer $225k) that align with the current AI company's San Francisco postings on the Zero G Talent board (board salary band typically $95k–$259k, median $240k). Because neither source separates the two companies cleanly, treat these as directional only.
What this means for you
If you are evaluating the current eight-person AI startup, the only primary employee voices on record are the 2024 team video, which describes a high-autonomy, engineer-driven rhythm with explicit rituals (no-meeting days, support rotations, show-and-tells) and an unlimited AI-tool budget. The lending-era reviews reflect a different business, a different lifecycle stage, and a closure that the current company has not experienced. Weight them accordingly.
Who Lasts
The profile that succeeds at Circleback today reads less like a traditional employee and more like a founding engineer who happens to draw a salary. The team runs lean, eight people as of the YC Winter 2024 batch, and the operating rhythm the founders describe on the record leaves little room for passengers. "We have a big culture of not just doing the thing like the task at hand, but also doing it in a way that makes it so the next time you or someone else goes to do that same thing, it's easier or the output is better," Haghani said in the 2024 demo walkthrough. That mindset — treat every task as infrastructure for the next person — is the clearest signal of who lasts.
Autonomy is real and so is the expectation that you use it. "Right now, we have that budget for like using AI. Um and we trust the team's judgment," Haghani noted. Candidates who need a spec handed to them, or who wait for permission to try a new model or tool, will stall. The team ships evals for anything quantifiable, "we have evals in place for things where it's quantifiable or there's a right or wrong answer", and shares discoveries in weekly show-and-tells: "Hey, this is a cool tool that I started using this week and here's how I'm using it." Engineers who treat AI as a curiosity rather than a default lever get left behind; Haghani admitted he barely opens a code editor anymore, doing most work "directly through agents."
Communication defaults to public Slack threads, not private DMs or email. "Anytime we get a customer email, there's like a new thread on Slack with the investigation and you can kind of do the whole back and forth there." Processes that once lived in a founder's "open claw" are being moved into shared channels. People who hoard context or prefer synchronous meetings over searchable threads create friction. The product itself reinforces this: the team builds an AI note-taker that records, transcribes, summarizes, and assigns action items across Slack, Notion, HubSpot, Salesforce, and Zapier, and they dogfood it for their own meeting hygiene, "we really try to never have our notes say the word discussed cuz it's fully pointless."
The flip side is visible in the lending-era reviews (2015–2022) from a different business that shared the name. Those reviews flagged "management faced that outcome," "no respect for employees," and "HR should do the same." Job security scored 2.4 out of 5 on Indeed. That company closed. The current startup shows no evidence of those patterns, Glassdoor and SimplyHired list the AI company separately, and the 2017 Circleback Inc. review (5.0 from a DevOps intern) describes "employees all know each other and work well together… relaxed work schedule, flexible management." But the lending-era feedback is a reminder: when a small team loses alignment, the damage concentrates fast.
Who burns out? People who need structure handed to them, who equate presence with output, who treat "non-promotable tasks" like meeting notes as someone else's problem, or who can't operate when the product direction shifts every few weeks, "the bounds of what's possible is now more so your imagination, not so much the skills that your team or you have, or the bandwidth that you have," Haghani said. The salary bands on Zero G Talent's board ($145k–$280k for engineering roles in San Francisco) reflect senior-level ownership expectations, not junior ticket-taking.
Who thrives? Engineers who write their own specs, default to public channels, treat AI as a baseline capability, and get energy from making the system better for the next person. The culture selects for builders who don't ask permission to improve the toolchain, they just ship the improvement and drop the link in Slack. The next automations feature will ship the same way the first one did: a Slack huddle between two founders, an engineer who picks up the ticket, and a link dropped in a public channel, no org chart required.
Working in frontier tech? Zero G Talent tracks the openings: see every open Circleback role, browse frontier tech jobs, the companies hiring, and the people building the field.