Monday Agents Month
Definition
A company-wide initiative at monday in which employees built and submitted AI agents over the course of a month. Per nizan-shifman in 2026-05-27-directions-vertical-pivot-and-prediction-markets, the program produced ~3,000+ submitted agents and ran an internal contest with ~10 sub-tracks. Treated by the team as a use-case mining ground for the vertical-use-case-led-brain direction: the catalog is mostly slop, but a small subset are genuine workflow agents whose value compounds with deeper context.
Key points
- Volume: “more than 3,000” submissions across the org.
- Mostly slop. saar-arbel’s framing: most submissions are thin Cursor / Claude wrappers that solve a narrow workflow without needing real cross-team context. They do not benefit from a Brain.
- A small set look genuinely Brain-shaped:
- inbar-markovich — Engy: always-on employee-sentiment agent that replaced a $60K/yr survey contract (real-time leader-accessible insight vs annual pulse).
- yuval-ovadia — Simona: AI candidate-matching agent saving ~600 hr/yr by surfacing strong candidates from one role into adjacent open roles.
- amit-frishberg — Web→mobile feature-port agent: reads web repo via internal MCPs, generates iOS/Android specs, kicks off coding agents. First feature shipped 3 hours after release.
- Roni — legal-request triage / handling agent. (Surname not captured.)
- shiran-navi — published an HR-team “agentic week” recap covering Engy, Simona and others.
- Filter for Brain-fit: the use case has to require either (a) durable, accumulated context that no single human currently holds, or (b) cross-team signal aggregation that is impossible to produce ad-hoc. Candidate-matching, sentiment-aggregation, and code-port-from-other-platform pass; “draft my email” wrappers do not.
- Action: nizan-shifman to share the full submission list with the team for systematic mining.
Evidence
- 2026-05-27-directions-vertical-pivot-and-prediction-markets — Nizan’s live walk-through of Shiran Navi’s recap and the named examples; Saar’s “mostly slop” filter.
Open questions
- Of the ~3,000 submissions, what fraction actually need a Brain to be 10× better — vs work today as standalone wrappers? Hypothesis: low single-digit percent.
- Are the strongest examples (Engy, Simona, Amit’s mobile-port) reusable patterns the Brain product could productize across customers, or are they Monday-specific?
- Is the program a one-off contest or recurring? Recurring would matter as a continuous use-case feed.
- Internal MCPs that Amit’s agent uses — are those Monday-internal-only, or is some equivalent already published in the Brain prototype’s stack?