Peak claim: "A complete alternative to Zapier." Later: "a platform to create AI agents that work for you." Final outcome: Product dies September 14, 2026. Founder returns to Google as VP of Product for Chrome. Not an acquisition. No official reason given. Lesson: You can be right about the product and still lose the interface.

Khosla partner Sandhya Venkatachalam had already backed Bank's previous company, Timeful. No Series A. No valuation disclosed. A few databases round this to $8.2M. The two announced checks add to $8.1M.

Do not confuse this with the UK parcel-delivery startup also named Relay ($10M, then a $35M Series A under a different legal entity). Scrapers keep merging them. This is Relay.app Inc., San Francisco, Jacob Bank.

Jacob Bank (LinkedIn, X: @jebank) — Founder & CEO. Cornell CS, Academic Excellence Award. Dropped a Stanford AI Lab Ph.D. to co-found Timeful with Yoav Shoham and Dan Ariely. Google acquired Timeful in May 2015 (amount undisclosed). Bank spent a little over six years inside Google as a product lead across Gmail, Google Calendar, Google Chat, and the Workspace developer platform.

He left Google in July 2021, when his first daughter was about six months old, and started a company called Collab AI. He did not know the product yet. Quote, February 2025:

Why they renamed it Relay: "so many things we do should be a relay race where the computer does some stuff and the user does some stuff."

He pulled former Gmail and Calendar people with him. Named: Thijs van As (founding Head of Product, Google Flights then Booking.com), Annika Crowley (founding designer, Timeful then Gmail/Chat), Jeroen Jillissen (Google Director of UX for Calendar/Gmail). At public launch in October 2023 the company was 11 people: 3 in the US including Bank, 1 in Canada, 7 across Germany, France, Spain, Switzerland, and the Netherlands. Google Calendar had been built out of Zurich. When Google tried to pull people back to the office in fall 2021, some of those engineers called Bank instead.

Khosla partner Sandhya Venkatachalam, who backed Timeful, backed him again:

The track record was real. The challenge was eating Zapier.

Wander (2021–mid 2023): Bank later called this "wandering in the desert." Eight or nine prototypes: automated to-do list, contextual knowledge base, standup tool, employee onboarding, performance reviews. None stuck. Design partners in the 2022 early access: Ramp and Lumos.

V1 (late 2022): Collaborative workflows at the "intersection of Zapier and Asana." All-hands prep, board meetings, onboarding. Target: companies of 30–500 people. Closed product, $5M Khosla seed. Bank later: by the end of beta they realized a category might exist, "but it wasn't us who was going to create it." Too much friction.

V2 (Oct 11, 2023, GA): "A complete alternative to Zapier." $3.1M from a16z the same day. Hundreds of triggers and actions. ChatGPT-powered AI Autofill and AI Classify. The stated differentiator was human-in-the-loop: any step could pause for a person to approve, fill missing data, or add a line before the workflow continued.

Named early customers in the launch press: Ramp, Skyflow, Warp, Motion, Lumos, Tavus.

V3 (2025–2026): Stop saying automation, start saying agents. Bank said changing the homepage from "the new standard for automation" to "a platform to create AI agents that work for you" lifted website conversion 25%. February 3, 2026: "biggest launch ever," Relay.app Agents. "Build your own AI team." April 10, 2026: he published "Relay.app vs. Claude Cowork." Three months later the company was winding down.

Integration gap, in his own words (Feb 2025): Relay had about 120 native integrations. Zapier had 7,000. He called Zapier's number a vanity metric and said the real target was 300–500. They never got there.

On July 16, 2026, Relay killed new signups, cancelled paid subscriptions through Paddle, and started refunding unused annual time. The homepage, the FAQ, and Bank's LinkedIn post do not give a reason. Commenters asked. No answer in the retrieved thread.

Free accounts were deleted August 15 at 23:59 PT. Paying customers keep access, extra steps, and extra AI credits through September 14 at 23:59 PT. Then the product, the data, and the connected-app tokens are gone. Export is JSON, Markdown prompts, and CSV. There is no importer into Zapier, Make, or n8n.

This is not a Google acquisition. No deal terms. The app does not live on inside Chrome. On August 17, 32 days after the shutdown notice, Bank announced he was VP of Product for Chrome, leading product and developer relations. "Several" Relay colleagues are joining. Thijs van As's LinkedIn now shows Google / Chrome. The others are not confirmed.

Bank's explanation, August 17:

Three reasons he gave: Chrome is "the primary workspace for productivity." It is an "open platform for the entire web" so you don't need to worry which APIs or MCP servers exist. And it is "the perfect place to collaborate with agents."

That last sentence is the obituary.

Relay spent five years teaching small teams to let software do work across Gmail, Calendar, Slack, and Notion, with a human still holding the last click. Chrome already sits on every one of those tabs. Gemini is already in the browser. Bank is walking the same product idea into a distribution channel Relay was never going to beat.

Timeful ended in an acquisition and put him inside Google. Relay ends in a shutdown and puts him back, with a better title and a harder lesson.

If your product only works as a tab, the browser can eat you. Relay's workflows lived in other people's apps via APIs. Chrome already had the session, the page, and the user. Bank's own Chrome post is the tell: you don't need APIs or MCP. The moment Google decided the browser should do work instead of display it, Relay's surface area shrank to a round-trip.

Human-in-the-loop is a feature, not a moat. Bank was right that companies will not hand the send button to a model. He was also early enough that every AI agent product copied "ask a human before the dangerous step." Being first to the cautious design did not lock anyone in.

A pedigree Zapier-alternative is still a feature until the incumbent copies it. Gmail/Calendar leaders, Khosla, a16z, HITL, native AI. By February 2025 Bank still described the field as everyone "circling around the same opportunity." By April 2026 he was explaining himself versus Claude Cowork.

You can have early PLG PMF and still not own the category. 440 paying customers, 1,200 weekly active teams, 25% of paid from YouTube, founder-led LinkedIn. That is real. Bank's Timeful scar was "300,000 downloads the first weekend and retained none of them," so Relay optimized for retention. Retention on a small base does not automatically produce a durable company when ChatGPT, Claude, Cursor, and Chrome move up the stack.

The integration gap is not a vanity metric if that's the job. Bank called Zapier's 7,000 connectors vanity. Users still hit missing apps. A nicer canvas that can't connect the long tail is not a replacement.

Big Tech will hire the people and let the product die. Timeful (2015) was a product acquisition into Calendar/Gmail. Relay (2026) is a shutdown plus a talent hire. July notice named no buyer. August, Bank is VP Product on Chrome. Customers get a JSON export and a deletion date. Google gets the intuition inside a product that already has distribution. That is a founder outcome, not a customer outcome.

A graceful shutdown is still a shutdown. 30/60 day windows, cancelled billing, prorated refunds, export tools. Customers will remember that. Investors still have a dead seed company and no Chrome deal to point at.

Do not confuse an $8M seed with a Zapier fight. Zapier had two decades and a brand that means "automation" the way Google means search. $8.1M and a team of nine is enough to build a beautiful product. It is not enough to displace the noun.

Founded: July 2021 (as Collab AI) Public launch: October 11, 2023 Paying customers: 440 (Bank, Feb 9, 2025) Weekly active teams: ~1,200 (same date) Headcount: 11 at launch (2023), 9 as of March 2025 Shutdown announced: July 16, 2026 (no reason given) Free users cut off: August 15, 2026 Paid users cut off: September 14, 2026 Money raised: $8.1M (Khosla, a16z, Neo, BoxGroup, SV Angel) Valuation: Never disclosed Series A: Never happened Acquisition: None. Product dies. People go to Google. Founder outcome: VP of Product, Google Chrome

The irony is almost too clean. Jacob Bank left Google to build AI that does real work across the apps you already live in, without taking away your judgment. Five years, eight or nine prototypes, two seeds, and one "biggest launch ever" later, Google hired him to put that product in the one app that already contains all the others. Relay was not a failed Zapier killer. It was a five-year prototype for Chrome. The prototype got killed so the platform could ship.

I hope this story serves as a reminder that even good ideas can fail, and staying adaptable matters just as much as building fast.

If you found it helpful, pass it on. Thank you.

Cheers,

Ram

StartupObituary.com 

👉 My simple ask: It took hours to put together this post for you. I hope you forward this email to at least one founder friend or share on your social channels 🙏.

Startup Obituary is for educational purpose only not a business advice.