Town raises $108M to bring AI assistants to whole teams
Town's $108M Series B, Kapital's $30M Woz deal, monday.com's Harmony voice agent, and more.
In this issue
- AI App BuildersTown raises $108M to bring AI assistants to whole teams
- DevelopersKapital buys coding-agent startup Woz for $30M
- CRM / Voicemonday.com launches Harmony voice agents for sales
- OperationsVessla exits stealth with $3.3M for property AI
- AccountingPlinto raises €2.9M to catch invoice errors with AI
Town raises $108M to bring AI assistants to whole teams
Source: Town / GlobeNewswire · Oct 19, 2026
Town, the AI work assistant from a team led by former Plaid CTO Jean-Denis Greze, has raised a $108 million Series B at a $1 billion valuation, co-led by Forerunner Ventures and Menlo Ventures with a16z and First Round also in the round. Each person gets an assistant Town calls a “Townie” that connects to their email, docs, calendar and core systems. The new bit is that Townies can now work directly with each other, asking questions and handing off work, and can also reach customers, partners and candidates over email and text.
The speed here is the headline: Town launched publicly only in June with a $55 million Series A and says it hit seven figures of ARR in weeks, with paying customers up 4x and teams up 9x since July. If you run a small team, the part I’d actually think about is Town’s stance that your company owns the context its assistants build and that nothing you connect gets used to train a model by default. I’d still pilot it on a bounded workflow before wiring it across the whole business, but model-agnostic, ownership-first language is a better starting point than the lock-in you usually get.
Kapital buys coding-agent startup Woz for $30M
Source: Kapital / PR Newswire · Oct 19, 2026
Mexico City-based Kapital, one of the larger B2B financial institutions in Latin America, has agreed to acquire Woz, a Y Combinator-backed startup building coding agents for complex business use cases, in a $30 million cash-and-stock deal. Woz co-founders Ben Collins and Brad Eckert join as Kapital’s Head of AI and CTO, and Woz’s whole team moves over. Woz’s flagship product WOZCODE uses patented token-reduction technology to make frontier coding models from Anthropic and OpenAI faster and cheaper, which the company says has saved customers more than $2.5 million.
This is a bank buying an AI developer-tools company to feed its own risk and underwriting stack, Kapital’s OctavIA, rather than a deal you’ll use directly. My read for a small business: it’s another sign that the “make coding models cheaper and more reliable” layer is becoming valuable infrastructure, not a toy. If you’re already leaning on AI coding tools, the token-reduction idea is worth watching, because cost per build is the number that quietly decides whether scaling AI is affordable.
monday.com launches Harmony voice agents for sales
Source: monday.com / Business Wire · Oct 19, 2026
monday.com has launched Harmony, a voice AI agent platform built for enterprise sales teams, using voice technology it acquired from OneAI and running on Twilio. Harmony relies on small, task-specific models for qualifying leads, capturing data and handling objections, which it says deliver roughly 200 millisecond latency and a contact rate of up to 70%. It runs the full sales cycle, including unlimited phone numbers with branded caller ID, predictive and power dialing, compliance controls and native CRM write-back, and is available in the US, UK, Australia and Singapore.
The customer story is worth treating as a single data point, not a guarantee: FCA-regulated UK broker Bright Car Finance says Harmony’s agent now follows up every approved applicant and drove a 9% increase in deals closed in under a year. My honest judgment for a small sales team: the real problem this solves is leads going cold before a human reaches them, but it’s a mid-market and enterprise product, so it only pays off if you have genuine call volume. I’d want to hear a live demo and check the consent and compliance controls before letting an agent talk on my brand.
Vessla exits stealth with $3.3M for property AI
Source: Seedcamp · Oct 19, 2026
Stockholm-based Vessla has emerged from stealth with a $3.3 million pre-seed round, more than twice oversubscribed and led by Seedcamp, with Inception Fund, Tiny VC and operators from Lovable and OpenAI joining. It’s building a shared AI platform that automates workflows across the systems property owners and managers already use, pulling information, applying each organisation’s rules and carrying out work through connected software while keeping people in charge of judgment calls. It says it’s already working with some of the Nordics’ largest commercial and residential property owners.
This is real estate operations rather than a tool for every small business, so I won’t oversell it. The pattern is the same one I flagged for fitness studios and gyms: a single maintenance request or booking still runs through a chain of separate tools and manual copying, and agents are being pointed at that coordination tax. If you run a facility or a property-heavy business, the practical test is whether an agent can actually complete the handoff across your existing stack rather than just surface a summary.
Plinto raises €2.9M to catch invoice errors with AI
Source: Tech.eu · Oct 20, 2026
Copenhagen-based Plinto has raised €2.9 million from People Ventures and Emblem for its AI-powered financial transaction monitoring. Its agents check transactions against contracts, budgets and internal policies, investigate discrepancies, collect evidence and then follow up with suppliers, request credit notes or prepare corrections for approval. It integrates with ERP, banking, documents and Slack or Teams, and the company says it currently has paying customers in logistics, real estate, facility services and food production.
The numbers it cites are company claims, so treat them as illustrations: around 430 discrepancies a month in one area of a customer’s business, and a duplicate invoice submitted under a different supplier name caught before a second payment. My judgment for a small business on thin margins is that this is a real pain point, because duplicate and over-charged invoices are usually found only after the money is gone. For most very small teams the volume won’t justify an enterprise tool yet, but if you’re processing a high count of supplier invoices a month, catching errors before payment is exactly where the time and money go.
News roundups are editorial and not sponsored. Facts are sourced from the links above; pricing and terms may change — verify before purchasing.