Executive Summary
Engineering activity across the startup ecosystem is diverging sharply in Q3 2026. While nearly half of tracked startups (49.1%) show negative commit velocity change — suggesting maturation or belt-tightening — over a quarter (26.2%) are experiencing extreme acceleration above +100%, with AgTech and HR Tech leading at +445% and +413% average velocity growth respectively.
This report draws on the VC Deal Flow Signal dataset — a weekly-updated corpus of engineering metrics derived from public GitHub activity across startup organizations. With five quarters of historical data (Q3 2025 through Q3 2026), we can now observe meaningful trends in where engineering teams are investing their energy and which sectors are attracting the most intense development activity.
AgTech (+445.5% avg velocity change) and HR Tech (+412.9%) are the fastest-accelerating sectors this quarter, driven by a wave of AI-powered agricultural platforms and next-generation workforce management tools. Meanwhile Enterprise SaaS (+61.7%) and Data Infrastructure (+54.6%) show the most moderate acceleration — mature sectors where engineering teams may be optimizing rather than expanding.
▲ Sector Velocity Ranking
Average commit velocity percentage change by sector, ranked from fastest to slowest. Each bar shows the mean of all startups' commitVelocityChange values within that sector.
Average Commit Velocity Change by Sector
The sector velocity ranking reveals a clear pattern: smaller, emerging verticals like AgTech (11 startups), HR Tech (11), and PropTech (16) are showing the most dramatic engineering acceleration. These sectors are still early in their software transformation cycles. In contrast, mature sectors with larger startup counts — Web3 (48), EdTech (34), E-commerce Infrastructure (27) — show more moderate average growth, with Data & Analytics bringing up the rear at +54.6%.
★ Top 10 Fastest-Accelerating Startups
These startups showed the highest commit velocity change this quarter. PhotoFlare leads the cohort with an extraordinary +1,600% velocity surge, followed closely by healthcare AI startup aipoch at +1,362%.
Commit Velocity Change — Top 10
PhotoFlare's +1,600% velocity change is the highest in the dataset — a Series A/B startup that has likely entered an intense feature-development cycle. The next cluster of 8 startups all registered exactly +999%, a ceiling effect suggesting the metric captures strong acceleration but may compress at extreme values. aipoch, a Seed-stage healthcare AI startup based in APAC, stands out for its combination of high velocity (307 commits/14d) and massive acceleration (+1,362%).
● Stage Distribution
The dataset is evenly distributed across funding stages, with no single stage dominating. This balance enables meaningful cross-stage comparisons of engineering behavior.
Startups by Stage
The distribution is remarkably even — Seed (26.2%) and Pre-seed (25.6%) together account for just over half of tracked startups (51.8%). Growth-stage (24.4%) and Series A/B (23.8%) companies make up the remainder. This balance is intentional: the dataset includes startups from earliest engineering formation through scaling, making it a representative cross-section of the venture-backed engineering ecosystem.
Notably, the extreme acceleration signals (velocity change above +999%) are concentrated in Pre-seed and Seed-stage companies, where small teams can produce dramatic percentage swings as they establish initial engineering velocity.
■ Signal Type Breakdown
Each startup is classified by its dominant engineering signal type. Framework migration dominates nearly half of all startups, reflecting the ongoing industry shift to modern architectures.
Engineering Signal Types Across 324 Startups
Framework migrations account for nearly half of all signals (49.7%) — an indication that the startup engineering community is actively modernizing architectures, likely adopting AI-native development patterns and cloud-native infrastructure. Engineering hiring bursts (26.9%) are the second-largest category, indicating active team scaling. Deploy frequency spikes (17.9%) suggest startups shipping more rapidly, while infrastructure buildouts (5.6%) form a smaller but significant category.
The dominance of framework migration signals suggests we're in a platform transition cycle — similar to the early-2020s React/Next.js migration wave, but now encompassing AI integration layers, edge computing frameworks, and real-time data architectures. Startups that aren't actively migrating their frameworks risk accumulating technical debt that will slow future development.
▼ Sector Deep Dive
A sector-by-sector analysis of engineering activity across all 15 sectors in our dataset. Each entry includes the startup count, average commit velocity change, and the dominant signal type.
| Sector | Startups | Avg Velocity Change | Dominant Signal |
|---|---|---|---|
| AgTech | 11 | +445.5% | Framework migration |
| HR Tech | 11 | +412.9% | Framework migration |
| PropTech | 16 | +364.0% | Framework migration |
| E-commerce Infra | 27 | +304.4% | Deploy frequency spike |
| Supply Chain | 23 | +230.1% | Framework migration |
| Legal Tech | 13 | +210.5% | Framework migration |
| EdTech | 34 | +174.1% | Engineering hiring burst |
| Robotics | 17 | +159.1% | Framework migration |
| Social & Community | 19 | +138.4% | Framework migration |
| Gaming | 17 | +136.4% | Framework migration |
| Web3 | 48 | +89.6% | Framework migration |
| Space Tech | 18 | +81.3% | Framework migration |
| Healthcare | 24 | +78.9% | Framework migration |
| Enterprise SaaS | 21 | +61.7% | Framework migration |
| Data & Analytics | 25 | +54.6% | Framework migration |
AgTech leads with the highest average velocity change (+445.5%), driven largely by AI-powered precision agriculture platforms. The sector's small startup count (11) means individual high-velocity outliers exert more influence on the average. HR Tech (+412.9%) is undergoing a similar transformation, with startups like iblai (sovereign AI infrastructure) and otter-sec representing a new wave of AI-native HR platforms.
At the other end of the spectrum, Data & Analytics (+54.6%) and Enterprise SaaS (+61.7%) show the most moderate growth. These are mature categories where rapid percentage changes are harder to achieve at scale — the startups here are more likely optimizing than rebuilding from scratch.
🌎 Geography Breakdown
Engineering activity is global, though the dataset skews toward startups with public GitHub presence. Geographic attribution is based on each startup's disclosed headquarters or primary operating region.
Of the 324 startups tracked, 193 (59.6%) have undisclosed geography in their GitHub organization profiles. Among those with known locations, the United States leads with 52 startups (16.0%), followed by the European Union (40, 12.3%), APAC (22, 6.8%), and the United Kingdom (7, 2.2%). LATAM (6) and Canada (4) round out the known-geography cohort.
Notable regional findings: aipoch (APAC, Healthcare) is the second-fastest accelerating startup in the entire dataset at +1,362%. saleor, an e-commerce API platform, shows strong engineering hiring (+110% contributor growth) from an undisclosed location. LATAM's ecomplus and eftechcombr both register +999% velocity changes from Pre-seed stage, suggesting emerging engineering hubs in the region.
⇄ Velocity Change Distribution
The overall velocity change across all 324 startups reveals a bimodal distribution: nearly equal shares are accelerating and decelerating, with a small stable middle.
The near-perfect split between startups accelerating (46.9%) and decelerating (49.1%) suggests a dynamic, churning ecosystem rather than a uniform boom or bust. The 26.2% of startups showing extreme acceleration above +100% are concentrated in early-stage companies — Pre-seed and Seed startups that are laying down their initial engineering velocity from a low baseline, where percentage swings are naturally larger.
However, some Growth-stage and Series A/B startups also appear in the extreme acceleration cohort, including saleor (+159%, Growth, E-commerce) and OpenCircuits (+386%, Growth, EdTech), indicating genuine scaling events rather than just baseline effects.
ⓘ Methodology
Data Source
All data is derived from public GitHub activity across 324 startup organizations. The VC Deal Flow Signal dataset tracks commit velocity, contributor counts, repository growth, and signal types on a weekly cadence. Organizations are classified by sector, stage, and geography based on their public profiles and disclosed information.
Commit Velocity
Commit velocity measures the number of commits pushed to a startup's GitHub repositories over a rolling 14-day window. The commitVelocityChange metric compares this period's velocity against the prior baseline, expressed as a percentage. A value of +100% means commit activity has doubled; +999% indicates an extreme acceleration from a very low baseline.
Signal Types
Startups are assigned one of four signal types based on their dominant engineering pattern: Deploy frequency spike (rapid increase in commit cadence), Engineering hiring burst (significant contributor growth rate increases), Infrastructure buildout (new repository creation at scale), and Framework migration (evidence of architectural changes in commit patterns). Startups with no dominant signal are tagged Traditional.
Stage Classification
Funding stages are self-reported or inferred from public information: Pre-seed (no institutional round announced), Seed (seed round completed), Series A/B (institutional venture rounds), and Growth (later-stage, often Series C+ or profitable).
Limitations
This dataset captures public GitHub activity only. Startups using private repositories for core development, or those on non-GitHub platforms, are underrepresented. The velocity change metric can produce extreme values (>+999%) for startups with very low baseline activity. Geography data is incomplete — 59.6% of startups have undisclosed locations. Sector classification is based on public descriptions and may not capture all subsector nuances.
Reproducibility
The full dataset is available for free at signals.gitdealflow.com/api/signals.json (JSON) and signals.gitdealflow.com/api/signals.csv (CSV). A formal methodology preprint is published on SSRN at ssrn.com/abstract=6606558. Historical data spans five complete quarters: Q3 2025 through Q3 2026.
License
Free for personal and editorial use. Attribution required: cite as "VC Deal Flow Signal (signals.gitdealflow.com), Q3 2026 data." Commercial redistribution requires permission. The underlying dataset is licensed under CC BY 4.0.