Agentic Coding Productivity

Feb 1 – Apr 11, 2026 · AIA · S&P · Dev Platform · Lifecycle · SP Success
Detection via commit messages (Co-Authored-By tags, noreply@anthropic.com) + reliable PR body patterns (Claude Code/Cursor Agent signatures). False positives excluded (“agentic” product feature, CSS cursor, augment config). Full-agentic classification based on last 30 days (Mar 12 – Apr 11) to reflect current adoption, not historical averages.

Executive Summary

1. Full agentic adoption is accelerating — but concentrated in two teams

Across all 5 teams, full agentic PRs grew from 7% in early Feb to 42% in the week of Apr 6. But adoption is sharply uneven: Dev Platform and S&P both surged from ~15% to 53% in the second half of the period, while AIA stayed flat at ~19%, and Lifecycle / SP Success remain near zero (6% and 1%). The gap is widening, not closing.

2. Full agentic engineers ship 1.5–2.5× more PRs per person

In S&P’s latest week, 3 full-agentic engineers produced 5.7 PRs/person vs 2.3 for AI-assisted — a 2.5× gap. In Dev Platform over the last month, full-agentic engineers averaged 12.7 PRs/person vs 8.4 for AI-assisted (1.5×). PR sizes are similar across both groups — the multiplier is in shipping frequency, not scope. Full-agentic engineers don’t write bigger PRs; they ship more of them.

3. The biggest productivity factor is project momentum, not tooling alone

The strongest predictor of weekly output is what a team is building, not how. The clearest evidence: Lifecycle jumped from 1.3 to 5.0 PRs/active/week and SP Success held steady at 2.8–3.8 — both with near-zero agentic adoption. Their output growth came entirely from project phase, not tooling. Similarly, Dev Platform ramped from 19 to 31 total PRs/week as they built out the Workflows & Agents platform features, and S&P surged to 24 PRs/week while executing the PNP 4.2 pricing migration — in both cases, all engineers shipped more, not just the agentic ones. Meanwhile AIA declined from 23 to 16 PRs/week with no comparable initiative driving output. Agentic tooling acts as a multiplier on existing project momentum — it amplifies output when teams are actively shipping, but doesn’t create momentum on its own.

648
Total PRs Merged
5 teams, 10 weeks
10
Full-Agentic Engineers
>50% in last 30 days (≥4 PRs)
33%
Full Agentic Rate (Last 30d)
117 / 359 PRs (Mar 12 – Apr 11)
159
Combined Full Agentic PRs
152 commit + 7 body-only (reliable)
89%
Top Agentic (Last Month)
bertenator: 25/28 PRs
227k
Lines Added
Capped at 5k/PR

Key Insights

1. Combined detection captures 25% full agentic usage overall

Scanning commit messages (Co-Authored-By tags) + reliable PR body patterns found 159 full agentic PRs (25%) across all time. In the last 30 days alone, the rate is 33% (117/359), showing accelerating adoption. Unreliable keywords (“agentic”, CSS cursor) were excluded to avoid false positives.

2. 10 full-agentic engineers in the last month

Measured over the last 30 days (≥4 PRs, >50% full agentic): jc-montanari (100%), josh-hewlett (92%), bertenator (89%), KeithRyanWong (80%), zmcnellis (75%), andredtr (75%), 4vanger (75%), memo-pineda (73%), jdodsoncollins (64%), morganthrapp (62%).

3. Dev Platform leads with 6 of 10 full-agentic engineers

55% of Dev Platform PRs in the last 30 days are full agentic, and 6 of its 11 engineers are full-agentic. S&P has 3 full-agentic engineers (54% last-30d rate). AIA has 1 (andredtr at 75%).

4. Adoption is binary, not gradual

zmcnellis: 0% through 03/23, then 15/16 in the last 3 weeks (75%). morganthrapp: 0/6 before mid-March, then 5/8 (62%). When engineers adopt, they go all-in within one week.

5. Full-agentic engineers ship more PRs per person

On S&P, full-agentic engineers averaged 5.7 PRs/person in the latest week vs 2.3 for others. The productivity multiplier is in frequency, not scope — PR sizes are similar across both groups.

Avg PRs per Active Contributor

Avg Lines per Active Contributor

Full Agentic PRs per Week (Combined Detection)

Full Agentic Rate by Week (%)

Total PRs per Week (Stacked by Team)

Cross-Team Summary (Combined Detection · Full-Agentic = >50% in Last 30 Days)

TeamTotal PRsFull AgenticFull Agentic RateFull-Agentic Engineers (Last 30d)Team Size
Dev Platform19685
43%
bertenator zmcnellis memo-pineda KeithRyanWong morganthrapp 4vanger11
S&P12146
38%
jc-montanari josh-hewlett jdodsoncollins7
AIA12323
19%
andredtr10
SP Success1392
1%
7
Lifecycle693
4%
3