What we shipped in Q1 2026, and what's coming in Q2

Every quarter I write this post, I find myself surprised by how much ground we covered. Q1 2026 was no different. When I look at the list of things we shipped between January 1st and March 31st, it genuinely doesn't feel like three months of work. But it was — and I want to give proper credit to the team, provide context for why we prioritized what we prioritized, and be honest about what didn't ship and why.

The headline launches

Multi-branch logic shipped on February 14th after eight months of design, build and rebuild. I've written about this feature in detail elsewhere, but the numbers are worth repeating here: in the six weeks since launch, 4,200 workflows now use branching logic, average agent complexity is up 34%, and our NPS from Growth and Enterprise customers increased by 11 points. This was our most requested feature in the history of the product, and I'm proud of how well it shipped.

AI suggestions v2 shipped on March 1st. The v1 system had a 31% acceptance rate — meaning customers were accepting the suggested next step roughly one in three times. After our three-layer rebuild (semantic preprocessing, integration-aware context, streaming architecture), v2 launched with a 58% acceptance rate in controlled testing. In production over the first two weeks, it settled at 62%. We'll keep iterating toward 75% — at that point, suggestions will be correct more often than not on every single use case.

The Agent Templates library launched on March 3rd with 50 pre-built workflows. This was a bet on a hypothesis we'd been developing for a year: that most customers don't want to build automation workflows from scratch, they want to start from something that already works and customize it. The usage data confirmed the hypothesis immediately. Within two weeks of launch, 34% of new agent creations started from a template rather than a blank canvas. Time-to-first-active-agent dropped from 23 minutes to 8 minutes for template users.

The supporting features

Seven features shipped in Q1 — three headliners and four supporting features that deserve more credit than they typically get.

CSV export for run logs was the highest-voted item on our public roadmap for four consecutive months. It sounds simple, but the implementation required redesigning our log storage schema to support efficient export of large run histories without degrading query performance for active users. We shipped it on January 28th and the immediate adoption rate — 23% of Growth and Enterprise customers used it in the first week — validated just how much customers had been waiting for it.

Date-based triggers let you run an agent on a schedule — at a specific time, on a specific day, or on a recurring cadence down to the minute. This sounds basic, but it unlocks an entirely new category of automation: morning briefings, end-of-week reports, monthly invoice processing, quarterly data audits. 1,200 date-based agents were created in the first three weeks after launch.

Webhook authentication improvements added support for HMAC signature verification, API key headers and OAuth 2.0 bearer tokens on incoming webhook triggers. This was primarily an enterprise ask — before this feature, any request to a Flowpath webhook URL would be accepted without verification, which was a security concern for customers exposing their agents to external triggers.

The Greenhouse integration for HR teams rounded out our HR automation coverage. Combined with our existing BambooHR and Workday integrations, Flowpath now covers the three most common HRIS platforms used by our Growth and Enterprise customers.

What didn't ship

Scheduled agents — the ability to run an agent on a cron schedule rather than just a date-based trigger — was planned for Q1 and slipped to Q2. The feature is built and tested, but we found a performance regression in our execution engine under the load pattern that scheduled agents produce (many agents triggering simultaneously at the top of the hour) and chose to fix it properly rather than ship a degraded version.

The mobile app was scoped for Q1 and pulled from the roadmap entirely in February. We started building it, realized we were building something mediocre, and made the call to stop. Mobile is on our roadmap for H2 2026 as a properly scoped project rather than a rushed Q1 feature.

Q2 roadmap

Scheduled agents will ship in the first two weeks of April — the performance regression is fixed and we're in final testing. The Slack bot for agent monitoring (trigger agents, view run status, receive error alerts without opening the Flowpath app) is targeting a mid-April launch. An agent marketplace where customers can publish and monetize their own templates is targeting end of May. And a major run logs update with saved filter views, anomaly detection alerts and a new execution timeline visualization is targeting end of June.

Q2 will be another big quarter. The theme, if there is one, is making agents easier to monitor and operate at scale — the features that matter most to the customers who are running 50+ agents and need better tooling to stay on top of them.