The technology sector leads on AI headlines but trails on deployment numbers. Just 32% of firms have AI running in production for technical execution tasks — the single highest figure in this dataset — and every other measured task sits well below that ceiling.
What's moving
All figures below carry Early Signal confidence grades. Treat them as directional indicators, not settled benchmarks.
- Technical execution leads at 32% of firms, growing at +9 points per quarter [Early Signal]. This is the furthest-along task in the dataset.
- Competitive intelligence sits at 30% of firms, also at +9 pts/quarter [Early Signal]. Nearly a third of technology firms have AI running here in production — a notably high floor for a non-core-engineering task.
- Marketing technology management is deployed at 15% of firms [Early Signal].
- Workflow automation design shows a split in the data: one reading puts it at 4% of firms, another at 12% [both Early Signal]. The range likely reflects definitional differences between firms — scripted rule-based automation versus agent-driven design. Neither figure is high.
- Onboarding and knowledge transfer sits at 8% of firms [Early Signal].
- Content repurposing and reporting to leadership each sit at just 3% of firms [Early Signal] — the lowest recorded deployment rates in this industry.
The +9 pts/quarter velocity is consistent across every task. Whether that rate holds into Q3 2026 is unconfirmed.
Capability vs deployment
Keep these lanes separate. A capability score is not a deployment figure.
For workflow automation design, three separate capability assessments exist:
- SkillOpt-based automated skill optimisation for AI agents scored 77% quality [Plausible], per VentureBeat AI (2026-06-11).
- Custom agents enabling sustained, context-aware automation across developer tools scored 76% quality [Plausible], per GitHub Blog (2026-06-11).
- A pre-mediation automation system demonstrated 54% quality on structuring negotiation preparation [Early Signal], per ArXiv cs.AI (2026-06-11).
For CI/CD pipeline management, a tool targeting pipeline abstraction — not core code generation — scored 78% quality [Plausible], per Hugging Face Blog (2026-06-11). Note the scope constraint: this score applies to the abstraction layer, not to writing or debugging core code.
The gap between the 77–78% capability scores and the 4–12% production deployment rate on workflow automation is real. Capability scores at this level do not automatically translate to production rollout. Integration cost, organisational readiness, and risk tolerance all create drag.
Roles most exposed in Technology
Based on the tasks with the highest current deployment rates:
- Competitive intelligence analysts — 30% of firms already have AI in production here. The task is automatable at scale and does not require physical presence or regulatory oversight.
- Developer operations and release engineers — CI/CD pipeline abstraction at 78% quality [Plausible] targets the coordination layer of their work, not the creative engineering layer. The narrow scope matters, but pipeline management is a significant portion of some roles.
- Technical writers and onboarding specialists — 8% deployment today, but knowledge transfer is a well-defined, document-heavy task. The ceiling is higher than current numbers suggest.
- Marketing operations roles within technology firms — 15% deployment in marketing technology management [Early Signal] suggests this function is further along than the headline automation conversation implies.
What to watch
The data is thin in several places:
- The 4% vs 12% discrepancy in workflow automation deployment is unresolved. Until firms report against a consistent definition, the real figure is unknown.
- The +9 pts/quarter velocity appears across every task simultaneously. This uniformity is statistically unusual and may reflect a measurement artefact rather than a genuine cross-task acceleration.
- Reporting to leadership at 3% deployment is either an opportunity or a wall. The task involves judgement, audience calibration, and political context — factors that consistently slow AI deployment regardless of capability scores.
Career Runway scorecard note: As of 2026-Q2, Career Runway has 15 published calls with 0 resolved. No accuracy record exists yet. Weight these signals accordingly.