Town is one of the fastest unicorns of this cycle, and the founding team is a big part of why investors moved so quickly. CEO Jean-Denis Greze was CTO at Plaid, where TechCrunch reported he grew the engineering team about 17.5x in four years; co-founder Tony Vincent is reported to have come out of Google’s AI org, though that background is thinly sourced. They launched Town publicly in June 2026, hit seven-figure annual recurring revenue within weeks, and according to The Next Web’s coverage of the raise grew paying customers fourfold and paying teams ninefold since July. The $1B valuation came on a $108M round co-led by Forerunner and Menlo, with existing backers a16z and First Round continuing, after a $55M Series A around the June launch.
The product is the part to internalize before any conversation, because the engineering questions almost certainly grow out of it. Each person gets a Townie: an assistant wired into their email, documents, calendar, and core work systems that learns their projects, relationships, and voice, then takes work off their plate. It’s built around multiple agents cooperating. Townies work across a company, ask each other questions, and hand off tasks, sharing only what their person would share and routing real decisions back to the human. They also reach outside the company, emailing and texting customers, partners, and candidates. Teams publish shared workflows so one process change propagates to every member’s Townie. Customers own the context their assistant builds and can take it with them, and by default their data isn’t used to train models.
Before you prep algorithms, picture the system underneath: a permissions model, a context pipeline, and a way to tell whether an agent did the right thing. That’s where I’d expect the interview to live.
Who Town hires, and the five-day rule
The roles listed on Town’s investor job boards this fall track the product directly: backend, iOS, desktop, and security engineers, growth and product engineers, plus three AI-leaning tracks worth naming, an AI engineer focused on evals and agent quality, an AI product engineer, and an AI context and data-infrastructure engineer. Agent quality and evals are a standalone function here, as is the pipeline that pulls a person’s scattered work context into something a Townie can reason over.
The firm rule is location. The engineering postings say the role is onsite “five days a week in person” in the Financial District “right near Montgomery BART,” and some, including the backend role, list it as “either SF or NYC.” No remote engineering option shows up, so if you’re not in or moving to one of those two cities, settle that with the recruiter before investing prep time. Visa sponsorship isn’t stated on the postings I read, so if you need it, make that your first recruiter question. The listings also repeat a specific profile: someone who already runs an AI-native coding workflow to ship fast and is comfortable “defining technical direction in ambiguous, fast-moving environments.” Town doesn’t publish levels, but the $250K–$300K band and the “senior or staff engineer” language on the eval role read senior, so a mid-level candidate should expect to clear a senior bar or ask the recruiter where they’d land. That profile isn’t filler at an agent company, and I’d expect at least one interviewer to watch how you actually work alongside AI tooling, not just whether you use it (my read, not something Town has published).
For difficulty calibration, the AI-native company interview guides hub and the AI-startup interview difficulty index are the right places to compare loop shapes, and the agent-heavy ones like Decagon and Harvey AI are closer analogues than a big-lab research interview.
No one has posted a Town loop yet
A November 2026 search of Glassdoor, Blind, and Reddit turns up no round-by-round Town engineering account, which is normal for a company hiring at scale for only a few months. The one Town-specific process signal comes from its own founding-technical-recruiter listing, which says “the interview loops and evaluation criteria are largely in place,” with tenured hiring managers who “know what good looks like,” and that “the bar is deliberately, uncomfortably high.” Read that as a real, intentionally selective process, not an improvised one. What follows is reasoning from that plus the founders’ background, so get the written agenda from your recruiter and trust it over anything here.
A company this new usually opens with a recruiter screen on motivation, level, location, and what you’d own, and because the team is still small I’d expect a founder or hiring manager in the loop early rather than only at the end (again, my read). Greze scaled a large engineering org at Plaid, so I’d bet he runs a structured, high-signal loop here too: a genuine coding exercise plus a design or architecture discussion over a loose conversation, weighted toward practical, product-shaped problems more than abstract puzzles. Don’t copy Plaid’s actual loop onto Town, though. It’s a different company at a far earlier stage, and nobody outside Town knows the format, the number of rounds, or whether the coding happens live or as a take-home. Prep the substance and leave the schedule open.
What the work implies each track tests
Build prep from the role, because the tracks pull in different directions. None of the prompts below are reported Town questions. They’re drawn from what the product has to solve, and they make decent practice:
- AI engineer (evals and agent quality): how do you measure whether a Townie’s action was right when there’s no single ground-truth label, and keep an eval suite trustworthy as both the agent and the models under it change? The posting wants someone who has “built or owned LLM eval systems” and is comfortable with model-routing tradeoffs.
- AI context and data infrastructure: design the pipeline that ingests one person’s email, docs, and calendar into a queryable context store, with per-item permissions so a Townie only surfaces what its person is allowed to share.
- Backend: model the cross-Townie handoff, where one assistant asks another for information and has to enforce permission and consent boundaries between two different people’s data.
- iOS or desktop (client) engineer: design the on-device consent flow for when a Townie wants to act for its user, keep a local context cache in sync with the server without leaking across accounts, and handle the OS-level permissions a desktop assistant needs to read mail and calendar.
- Product or growth engineer: instrument and improve the invite and activation funnel for a tool whose value only shows up after it has ingested enough context to be useful.
- Security: an assistant can email and text external parties for a user, so enumerate the abuse and data-exfiltration paths and what you’d gate or detect.
The agent-permission and context-boundary theme runs through all of them, so fluency there reads as real engagement with Town’s core bet. Server-side roles lean on the multi-tenant, high-ingest problems in the system design interview guides, the security track overlaps with the security engineer interview guides, and a behavioral screen, if the loop has one, would likely probe your appetite for a five-day-in-office pace and for ambiguity. The AI-era interview notes are worth reading given how explicitly Town hires for an AI-native workflow.
What the posted $250K–$300K band covers, and what it leaves out
Two of Town’s engineering postings, the backend engineer and the AI engineer for evals, list a base range of “$250K – $300K.” Neither states an equity grant, and the iOS posting I read carried no number, so treat $250K–$300K as the base band for senior engineering rather than a guaranteed total. Town’s founding technical recruiter role is posted at $150K–$200K, so the engineering band sits well above the go-to-market side. California and New York require pay ranges on postings, so read the figure on the live listing for your exact role rather than trusting an aggregator, then cross-check levels.fyi and Blind for comparable early-stage SF AI companies. Put base, equity, and any sign-on into one annualized figure with the total comp calculator before comparing offers.
At a company that went from launch to $1B in four months, equity is the number that moves, and the posted base says almost nothing about it. Ask directly about the grant size, the preferred-share price, the strike price, and the option pool, because an early senior hire’s equity can swing total value far past a $50K base spread. That’s where your negotiation should focus, and any range someone quotes from outside is a market estimate, not Town’s offer, until a Town letter says so.
| Town interview and company facts (as of November 2026) | Detail | How well-sourced |
|---|---|---|
| What it does | Workplace AI assistant. Each employee gets a “Townie” wired into their email, docs, calendar, and work systems; Townies cooperate across teams, hand off tasks with permission, and contact external parties. SOC 2 Type 2; customer context is portable and not used for training by default | Reported (The Next Web coverage and company statements) |
| Funding and valuation | $108M Series B at a $1B valuation, announced October 19, 2026, co-led by Forerunner Ventures and Menlo Ventures, with a16z and First Round continuing. Prior round was a $55M Series A around the June 2026 public launch | Reported (The Next Web) |
| Founders | Jean-Denis Greze (co-founder and CEO, formerly CTO of Plaid) and Tony Vincent (co-founder, reported to be ex-Google AI) | Greze/Plaid-CTO widely reported (SiliconANGLE, AIWeekly; Plaid scaling per TechCrunch). Vincent/Google appears only in job-board briefs, unverified |
| Traction | Public launch June 2026; seven-figure ARR within weeks; paying customers up 4x and paying teams up 9x since July; thousands of businesses using Townies daily | Company-claimed figures relayed by The Next Web, not independently verified |
| Engineering roles and location | Backend, iOS, desktop, security, growth/product, plus AI engineer (evals/agent quality), AI product engineer, and AI context/data-infrastructure engineer. Onsite five days a week in SF Financial District near Montgomery BART; some roles (backend) also list New York | Reported (Town postings on investor job boards, read November 2026) |
| Interview loop | No public loop and no candidate reports on Glassdoor, Blind, or Reddit. Town’s own recruiter listing: loops and evaluation criteria “largely in place,” bar “deliberately, uncomfortably high.” Plausible shape: recruiter screen, founder/hiring-manager round, a coding exercise, a design discussion. Format and round count unknown | One quoted process signal from Town; the rest inferred. Confirm with the recruiter |
| Compensation | Base band of $250K–$300K on the backend and AI-eval engineering postings; no equity figure stated. Founding technical recruiter role posted at $150K–$200K. Check the live CA/NY listing for the exact range; equity is the swing factor at a new unicorn | Base figures from Town’s postings; equity not disclosed. Verify with the recruiter |
If the loop looks anything like the above, the way I’d try to stand out is to engage Town as the permissions-and-context problem it really is. The founders are betting that an assistant worth paying for has to understand a person’s actual work and act on it safely across a whole company. Come ready to argue the hard cases, consent between two people’s Townies, evals with no clean ground truth, a funnel that only pays off after ingestion, and just as ready to say where the bet breaks.
Useful next steps:
