State of AI in Portfolio Management 2026 by Tempo

Tempo Software’s 2026 State of AI in Portfolio Management report looks at how big companies plan and run projects when AI is part of the mix. The survey polled 300 senior leaders including executives, PMO leaders, portfolio managers, and project managers to see how AI is actually being used and where older planning software breaks down. The main message is simple: just having or dabbling in AI isn’t a real advantage anymore. Real gains happen when teams let AI agents do real execution work, connect big-picture strategy to daily tasks and manage people and AI together without letting projects drift off course.
Key Insights
1. Almost Everyone Is Using AI Now
- 91% are piloting or actively using AI: 47% actively use AI in project delivery and 44% are piloting it. In plain terms, AI is now standard rather than special.
- Almost nobody is sitting it out: 7% plan to roll out AI in the next 6 to 12 months, and only 2% have no plans.
- Basic AI use isn’t a winning edge: Teams that simply say they “use AI” don’t automatically beat teams that are still piloting or haven’t started yet.
2. Real AI Agents Are Still Less Common Than Basic AI Tools
- Only 33% have AI agents doing real delivery work: This means AI is actually executing tasks like coding, QA, or documentation not just acting as a chat helper. Another 56% are piloting agents, and 12% aren’t (or haven’t yet).
- How much delivery work touches AI varies a lot:
- 11% say less than 10% of their project management and delivery work involves AI.
- 11% say 10%–25% involves AI.
- 40% (the biggest group) say 26%–50% involves AI.
- 17% say 51%–75% involves AI.
- 7% say more than 75% involves AI.
3. Teams Hit Plenty of Friction With AI
- Top management headaches: Leaders most often struggle with AI output quality and oversight (43%), fitting AI into project management workflows (38%), governance/compliance/risk (36%), and tracking AI cost versus value (36%).
- Other common AI challenges: Avoiding duplicate AI work (34%), figuring out what AI did versus what humans did (30%), linking planning to execution (27%), planning team capacity with AI (24%), and replanning quickly (22%).
- Day-to-day friction points:
- 46% say teams end up duplicating work with AI.
- 42% struggle to tie AI spending to return on investment (ROI).
- 41% say their planning tools and delivery tools don’t connect properly.
- 39% can’t see what AI actually costs them.
- 39% can’t clearly tell AI work apart from human work.
- 34% can’t replan fast enough when things change.
4. What Typical Projects and Tool Stacks Look Like
- Most projects take several months: 47% usually run 3–6 months and 19% run 6–12 months. Another 12% take 1–2 months, 7% take 2–3 weeks, 6% take over a year, 9% vary, and 0% take under two weeks.
- Heavy, multi-tool setups: 100% use project management/planning tools, 74% use ERP, 66% use ITSM, 66% use Excel or custom in-house tools, and 49% use CRM (with SCM at 35%, and other chart categories like HR/engineering/general platforms also showing up in the stack breakdown).
- What teams use planning software for most: Resource and capacity planning (78%), project comms and docs (77%), cost and budget tracking (74%), reporting and dashboards (68%), time tracking (67%), risk management (61%), scenario planning (57%), roadmapping (50%), and support ticket triage (46%).
5. Specific AI Portfolio Features Are Still Early-Stage
- Reporting is #1 in live use: 47% actively use AI to prepare reports or dashboards (31% piloting, 22% not adopted).
- Risk and planning features lag further behind in active use:
- Identifying risks and flagging issues: 33% in use, 40% piloting, 27% not adopted.
- Capacity and resource planning: 29% in use, 39% piloting, 32% not adopted.
- Prioritizing or reprioritizing work: 26% in use, 35% piloting, 39% not adopted.
- Timeline forecasting: 25% in use, 45% piloting, 30% not adopted.
- Agents executing PM tasks on their own: 21% in use, 38% piloting, 40% not adopted.
- Scenario planning and modeling: 20% in use, 42% piloting, 38% not adopted.
6. Putting AI Agents Into Production Cuts Friction
- Attribution gets much clearer: Among teams actively using AI, trouble telling AI work apart from human work drops from 55% (no agents deployed) to 16% (agents deployed in production).
- Cost and ROI visibility improves: Inability to see AI costs drops from 55% to 22%, and trouble tying AI spend to ROI drops from 43% to 19%.
- Tool connection and duplication ease up: Disconnected planning/delivery tools drop from 47% to 30%, and duplicate AI work drops from 53% to 40%.
7. Adoption Varies Sharply by Industry
- Software leads by industry: 47% of software companies have AI agents in production, followed by retail/e-commerce (38%), transportation/logistics (29%), consulting (29%), manufacturing (27%), and construction/engineering (5%).
8. What Helps Teams Deliver High Project Volume Quickly
- Most run dozens of projects at once: 43% run 11–25 concurrent projects and 30% run 26–50 (with 13% running >50, 12% running 5–10, and 1% running <5).
- AI agents give a clear speed boost: Deploying AI agents adds +9 percentage points to hitting high volume with projects finishing under six months, and running 3+ AI use cases adds +3 points.
- Size and multi-team complexity slow things down: Having 1,000–5,000 employees shifts that metric by -5 points, and complex projects involving 4+ teams shift it by -12 points.
9. Frontline PMs Feel the Mess More Than Senior Leaders
- Project managers see more hidden costs: 54% of individual PMs say they can’t see AI costs, compared with 36% of portfolio managers and 28% of PMO leaders.
- PMs also see more rework and ROI blind spots: 58% of individual PMs report duplicate AI work (vs. 41% of portfolio managers and 40% of PMO leaders), and 50% say they can’t tie AI spend to ROI (vs. 42% and 33%).
10. Cross-Team Dependencies Are Still Tough Except for Top Performers
- Almost every project spans multiple teams: 48% of typical projects involve 4–6 teams, 34% involve 2–3 teams, 17% involve 7+ teams, and only 1% involve one team.
- More teams means harder coordination: Overall, 40% find managing dependencies very (36%) or extremely (4%) challenging (52% somewhat, 8% not challenging). High difficulty climbs from 27% (2–3 teams) to 44% (4–6 teams) to 53% (7+ teams).
- Strong cross-team managers use more AI in real work: Leaders who manage cross-team projects well are more likely to have AI agents deployed (38% vs. 25% of struggling peers), run AI on 26%+ of work (68% vs. 29%), and have 3+ AI use cases live (57% vs. 26%, alongside a 70% chart figure).
11. A Tiny Elite Group Has All 7 Capabilities Live
- Only 18 out of 300 teams (6%) have all seven AI capabilities operational and AI agents deployed in production.
- Their pain points nearly vanish: Compared to the overall baseline, this group has far less trouble with AI-vs.-human attribution (6% vs. 39%), planning-to-delivery connection (6% vs. 41%), slow replanning (6% vs. 34%), unseen AI costs (6% vs. 39%), unclear ROI (17% vs. 42%), and duplicate AI work (28% vs. 46%).
- They also see better project pacing and fewer severe bottlenecks: Fewer of their projects run 6+ months (17% vs. 25%) and fewer rate dependencies as very or extremely hard (28% vs. 40%).
12. Big Wishlist, Small Current Tooling
- What leaders want most (top-three rankings): Connecting planning to execution (46%), continuous real-time replanning (43%), managing AI and humans in one system (43%), real-time AI cost tracking (42%), cross-team visibility into AI work (42%), tying AI spend to ROI (41%), and modeling portfolio changes (41%).
- Most don’t have tools that do it well yet: Only 22%–30% say they have a tool that handles any of those capabilities well including just 30% for connecting planning to execution (meaning 70% lack a strong tool for their #1 pick).
- High expectations vs. current reality: 89% expect an AI-ready platform to give early drift warnings, 89% expect continuous course correction, 86% expect planning-to-execution alignment, and 82% expect unified human-AI management yet only 25%–30% currently have a tool that delivers those well.
Conclusion
In short, the report shows that companies are surrounded by AI, but most haven’t wired it into real execution or connected it to how portfolios are actually planned and tracked. Teams that get real speed, clarity, and fewer headaches aren’t just sprinkling AI onto old spreadsheets or dashboards, they’re putting AI agents into production work, stacking multiple practical capabilities together, and tracking human and AI contributors in one place. For portfolio leaders, the big shift ahead is moving from passive reporting to real-time orchestration where strategy and day-to-day work stay in sync.
You can check out the full report here.