9/1359 → 34/990; +2.77 percentage points of formation share.
What used to mean bolting scripts onto legacy screens has quietly split into two very different businesses inside this batch of companies. One track, from loopfour to Velos to TableFlow, is still classic robotic process automation rebuilt around LLMs and computer vision, replacing overseas back-office labor with AI workers that read PDFs and click through the same software a human would. The other track has narrowed hard into vertical depth: Parallel targets hospital medical coding, Cardinal Gray targets DMV lien filings, Sameday targets trade dispatch, and Finic targets fraud detection, each one betting that owning a single messy workflow beats offering a general automation platform. That narrowing shows up in the exit already on the board, StackAI's 2026 acquisition by Asana, which suggests buyers now want a wedge into a specific back-office process rather than another do-anything tool, and it lines up with this formation's 5.19x growth outpacing the broader Workflow Automation category's 2.24x. The tension underneath is that these agents are being asked to act like software workers inside systems built for humans, which is exactly why the exception-handling problem shows up twice: one live investigation flags that supply and field operations still resolve disruptions through phone calls and spreadsheets rather than any system of record, and loopfour's whole pitch is literally "you approve only the exceptions." None of this was viable before 2023, when GPT-4-class models first made reliable computer use and document extraction cheap enough that hiring an "AI worker" could plausibly undercut an outsourced back office instead of just automating rule-based clicks.
Robot fleets turn physical failures into distributed-software incidents, but debugging also requires synchronized sensor, action, environment, and operator context.
Map deployment-stage robotics companies against tools for fleet telemetry, incident reconstruction, regression diagnosis, and reliability operations.
Supply and field operations coordinate normal flows in systems of record, while disruptions are resolved through fragmented calls, messages, and spreadsheets.
Catalog high-cost exceptions across logistics and field operations, then map how context, decisions, and resolution state move between organizations and software.
With builder platforms like StackAI, Magic Loops, and Autotab letting non-engineers assemble automations quickly, the lack of engineering rigor around these workflows implies a coming need for lightweight testing/eval tools to catch regressions before automations run against production data (echoing TableFlow and Tagsafe's data-quality/analytics focus).
Check back around June 2027 for new YC batches, to see whether independent evidence for "No-code/low-code automation testing and eval suites" has appeared.
Tools like Sola, Autotab, and Magic Loops build agents to execute multi-step workflows autonomously; as these scale past simple tasks, edge cases and failures will require a standardized escalation/handoff mechanism to humans rather than silent failure, similar to how RPA historically needed exception queues.
Check back around May 2027 for new YC batches, to see whether independent evidence for "Exception-handling / human-in-the-loop escalation infrastructure" has appeared.
Companies like Finic (compliance, fraud detection) and Automat (RPA on documents) are building automations that act on sensitive enterprise systems, which creates demand for a dedicated layer that manages credentials, scoped permissions, and audit trails for non-human agents rather than each startup building its own.
Check back around May 2027 for new YC batches, to see whether independent evidence for "Agent identity & permissioning layer" has appeared.
Included in YC's Automation membership during the measured recent window.
Included in YC's Automation membership during the measured recent window.
Included in YC's Automation membership during the measured recent window.
Included in YC's Automation membership during the measured recent window.
Included in YC's Automation membership during the measured recent window.
Included in YC's Automation membership during the measured recent window.
Included in YC's Automation membership during the measured recent window.
Included in YC's Automation membership during the measured recent window.
Included in YC's Automation membership during the measured recent window.
Included in YC's Automation membership during the measured recent window.
Measured-window members of Automation that could not be assigned to a narrower shared YC tag or industry without forcing a classification.
A primary internal company segment within Automation, derived deterministically from the shared yc tag “Operations”. This is a classification of current window members, not a causal layer or a separate time trend.
A primary internal company segment within Automation, derived deterministically from the shared yc tag “AI Assistant”. This is a classification of current window members, not a causal layer or a separate time trend.
A primary internal company segment within Automation, derived deterministically from the shared yc industry “Engineering, Product and Design”. This is a classification of current window members, not a causal layer or a separate time trend.
A primary internal company segment within Automation, derived deterministically from the shared yc tag “Sales Enablement”. This is a classification of current window members, not a causal layer or a separate time trend.
A primary internal company segment within Automation, derived deterministically from the shared yc tag “Robotic Process Automation”. This is a classification of current window members, not a causal layer or a separate time trend.
These share a word with “Automation” but are tracked as separate categories. The real overlap below is the actual share of Automation’s companies also found in that category — not a guess.
Cambrian Signal, “Automation,” publication yc-2026-10-02-semantic-v8, snapshot October 2, 2026.
cambrian.jafar.com/formation/automation