Company profile · 4 min read
Cofia: automations discovered from a team’s existing work
Cofia detects repeatable work patterns from team activity, proposes custom AI agents and lets users review and launch the resulting automations.
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
Published · Updated
Cofia is trying to make automation emerge from the work a team already does. Instead of asking a business user to identify a workflow, write a prompt and maintain a builder, it watches for repeatable patterns, proposes a custom agent and waits for the user to review and launch it.
What it does
The YC profile describes Cofia as AI automations that implement themselves. Its launch post says the system learns repetitive tasks such as prospect-list building, outreach, scheduling, recruiting-pipeline maintenance and data reports, then offers an automation when it recognizes a pattern.
The interesting implementation detail is that Cofia says it works from system events and anonymized network traffic rather than requiring screen recordings or a user to describe every step. It also says people can see what it is processing, review the proposed agent and launch it. The public homepage was not readable in the checked source packet, so current pricing, integrations and availability are not established beyond the official YC materials.
Why I’d look closer
The buyer is a growing team with repetitive cross-tool work and no dedicated automation engineer. The founders’ operating context fits: Moses Wayne previously led monetization engineering at Duolingo, and Paola Martinez led product and retention at Brilliant.org.
The advantage is opportunity discovery. Most teams know automation exists in theory but cannot spare the time to map every repeated task. The tradeoff is observation risk. System events and network traffic can reveal sensitive customer data, credentials, hiring decisions and business strategy. Anonymization is a claim and a design requirement, not permission to collect everything.
What I’d ask
What events and traffic are collected, and how can an administrator exclude systems or fields? Can a user inspect the proposed steps, source data and permissions before launch? How are generated agents versioned, monitored and stopped? Does Cofia ever infer a workflow from another team’s data, and what is the deletion path?
My editorial take
Shortlist Cofia if your team has visible repetitive work but nobody owns workflow discovery. Start with low-risk internal reporting or draft-only coordination, require explicit approval before writes or sends, and compare the automation’s maintenance cost with the manual task it replaces. The product’s promise is useful; the data boundary decides whether it is deployable.
Quick facts
| Field | Sourced detail |
|---|---|
| Product | Pattern detection that proposes custom AI automations from existing work |
| Buyer | Operations, sales, recruiting and other repetitive knowledge-work teams |
| Examples named | Prospect lists, outreach, scheduling, recruiting pipelines and reports |
| Pricing | Not published in the checked pages |
| Main question | Can Cofia discover useful workflows without collecting more work data than the team intended? |
Sources checked
| Source | Checked |
|---|---|
| YC company profile | 2026-09-19 |
| Cofia YC launch | 2026-09-19 |
| Cofia homepage | 2026-09-19 |
Cohort context
Cofia is listed in Winter 2026. In our 2026-09-18 directory snapshot, 126 of 199 listed companies in that cohort have YC’s primary industry label B2B (63.3%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.
Public website snapshot
Observed 2026-09-19T16:20:07.561Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.
| Signal | Homepage observation |
|---|---|
| Product description metadata | Observed |
| Canonical link | Not observed in this response |
| H1 or H2 heading | Not observed in this response |
| Typed structured data | Not observed in this response |
| Docs/developer link | Not observed in this response |
| Pricing link | Not observed in this response |
| llms.txt link | Not observed in this response |
| Markdown alternate | Not observed in this response |
Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.
About the author
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.
First1000 ↗ · X ↗