Single and Multiple AI Tool Stacks Across 11 Submission Cohorts
VCL’s v9 data shows different single- and multi-tool mixes across 11 visible submission months. Compare each cohort with its reporting coverage, without inferring tool switching.
*Updated October 25, 2026.*
Multiple AI tools were recorded for 65.1% of the November 2025 submission cohort with tool data, compared with 46.8% for August 2026. The intervening cohorts move in both directions. This table does not show a steady increase in multi-tool use.
It also does not show builders dropping or adding tools. VCL’s version 9 monthly chart groups projects by submission month, then reads their recorded tool choices at the October 1, 2026 cutoff. Each row contains different projects.
The visible monthly cohorts
The table combines the published two-tool and three-or-more-tool groups into a single multi-tool count. Its percentages are recalculated from integer counts, avoiding rounding errors from adding the source’s rounded percentages.
| Submission month | Eligible submissions | With recorded tools | Recording coverage | One AI tool | Two or more tools | Multi-tool share of recorded projects |
|---|---|---|---|---|---|---|
| October 2025 | 50 | 37 | 74.0% | 23 | 14 | 37.8% |
| November 2025 | 98 | 83 | 84.7% | 29 | 54 | 65.1% |
| December 2025 | 104 | 76 | 73.1% | 33 | 43 | 56.6% |
| January 2026 | 90 | 70 | 77.8% | 34 | 36 | 51.4% |
| February 2026 | 102 | 70 | 68.6% | 33 | 37 | 52.9% |
| March 2026 | 123 | 91 | 74.0% | 51 | 40 | 44.0% |
| April 2026 | 48 | 38 | 79.2% | 20 | 18 | 47.4% |
| May 2026 | 52 | 41 | 78.8% | 21 | 20 | 48.8% |
| June 2026 | 86 | 48 | 55.8% | 26 | 22 | 45.8% |
| July 2026 | 109 | 25 | 22.9% | 10 | 15 | 60.0% |
| August 2026 | 147 | 47 | 32.0% | 25 | 22 | 46.8% |
For example, November’s multi-tool count is 31 projects with two tools plus 23 with three or more. 54 / 83 × 100 = 65.1%. The versioned CSV preserves the separate groups.
Why July needs a different reading
July’s 60.0% may look like a clear preference for multiple tools. But only 25 of its 109 eligible projects have recorded tool data. The remaining 84 are unknown, not single-tool projects. A result describing 25 projects cannot establish the tool choices of all 109.
November has much higher coverage: 83 of 98 projects, or 84.7%. August has 47 of 147, or 32.0%. Differences in who has recorded tool data limit comparisons between months, even before considering differences in project categories or builders.
The visible months contain 626 projects with recorded tools. Across the full October 2025–September 2026 window, 656 of 1,095 eligible submissions have tool data. These window-specific counts differ from the report’s overall 883 recorded-tool projects because the full frozen cohort also includes projects outside the submission window.
September is withheld rather than zero
September 2026 has 86 eligible submissions and 30 with recorded tools. The chart withholds all of that month’s tool-count buckets under its small-cell rule: a bucket containing one to four projects suppresses the whole month. The split is unavailable, not zero.
This article does not reconstruct the withheld values. A month also needs at least 20 projects with known tools to be plotted under the publication rules.
What builders and toolmakers can use
The visible cohorts include many projects reporting one AI tool and many reporting several. Their changing mix offers useful questions for follow-up interviews: why did a builder add another tool, what did each tool do, and did that division help them ship?
This dataset does not answer those workflow questions or measure which combination produces better products. It records selections. Claude and Claude Code are counted together, as are ChatGPT and Codex. Each combined entry counts once; the data doesn’t distinguish which product was used.
To measure switching, the same projects would need dated before-and-after tool records. Sorting a cutoff snapshot by submission month cannot supply that history.
Data and source: VCL Research, State of Vibe Coding 2026, version 9, tools by submission month, cutoff October 1, 2026 at 00:00 UTC. This is a descriptive analysis of VCL-listed projects, not market share or a causal comparison. See the edition’s methodology and overall tool-stack analysis.