← Analyst Firms 10 min read

Forrester on Developer Experience and AI Tools: The Honeymoon Is Over

Executive Summary

  • Only 13% of enterprises report positive EBITDA impact from AI investments, and fewer than one-third can connect AI value to P&L — Forrester predicts enterprises will defer 25% of planned AI spend to 2027 (Forrester State of AI Survey 2025, n=1,400+ AI decision-makers).
  • 48% of firms have already cut headcount due to AI, yet change management and employee experience rank among the least prioritized areas — a dangerous inversion that explains why adoption stalls after initial excitement.
  • Forrester’s Developer Survey 2025 finds coding (48%) and testing (47%) are the top SDLC phases for AI adoption, but trailing phases like development insights (33%) reveal that AI remains a narrow coding accelerator, not the full-lifecycle tool vendors promise.
  • Forrester predicts CS enrollment drops 20% and time-to-hire for developers doubles — the paradox is that AI makes routine coding cheaper while making the architects who govern AI output more scarce and expensive.
  • Forrester coined “Agentic Software Development” (ASD) in March 2026 as the successor to copilot-style assistance — a Forrester Wave evaluation of ASD tools is planned for H2 2026, which will become the definitive enterprise buyer’s guide for this category.

The Forrester Framework: From TuringBots to AppGen

Forrester’s approach to AI in software development differs from Gartner’s in one important way: where Gartner evaluates vendors (Magic Quadrant for AI Code Assistants), Forrester builds conceptual frameworks for how the entire discipline changes. Three concepts define their thinking:

TuringBots — Forrester’s umbrella term for AI agents that operate across the software development lifecycle. The firm coined this term before “AI code assistants” or “copilots” became standard vocabulary. TuringBots are not limited to code generation; they encompass analysis, design, testing, security review, and delivery. VP Principal Analyst Diego Lo Giudice predicts that by 2028, TuringBots will collapse SDLC silos entirely, making the traditional lifecycle “less visible” as development becomes real-time (Forrester, “The Future Is Now: TuringBots Will Collapse The SDLC Silos,” May 2024).

Agentic Software Development (ASD) — Defined in March 2026 as “the use of AI agents that can plan, generate, modify, test, and explain software artifacts across multiple stages of the SDLC, working alongside human developers with a degree of autonomy” (Lo Giudice, March 2, 2026). Forrester draws three explicit boundaries that distinguish ASD from copilot-style tools:

  1. Agency over assistance — systems execute tasks independently, not reactively
  2. SDLC breadth — spans design, documentation, refactoring, testing, review, and delivery
  3. Professional focus — built for software engineers in real codebases, excluding low-code/citizen developer scenarios

AppGen Platforms — Forrester’s most ambitious prediction: Application Generation platforms will replace both traditional development and low-code within 4-6 years. AppGen platforms are “AI-native ecosystems that use agentic TuringBots to automate and orchestrate across the SDLC, employ intermediate modeling languages to bridge business intent and executable code, and enable multimodal development in a single iterative flow” (Lo Giudice and Bratincevic, 2024-2025). Microsoft, Oracle, OutSystems, and Airtable are building toward this model.

The Honeymoon Problem

Forrester Principal Analyst Devin Dickerson wrote the most honest first-person account of AI coding tool limitations published by any analyst firm (“The AI Coding Honeymoon — And What Comes After,” January 2025). His findings:

  • Initial productivity felt dramatic — “hundreds of lines of perfectly formatted code” generated rapidly
  • Context switching between web interfaces and IDEs created significant cognitive overhead
  • Architectural drift accumulated through small pragmatic decisions into systemic design problems
  • Scaling to multi-file refactoring caused cascading errors, with “test coverage plummeted to barely 50%”
  • His conclusion: “Software fundamentals are now more important than ever: clean architecture, careful design, disciplined development practices, thorough testing”

This matches a pattern across multiple data sources. Developers perceive large productivity gains while measured outcomes tell a different story. The METR RCT (n=16 experienced developers, 246 tasks, July 2025) found developers were 19% slower with AI tools despite believing they were 20% faster. Forrester’s own position acknowledges this gap without citing the METR study directly.

Developer Experience Data: What the Surveys Show

Forrester’s Developer Survey, 2025

Forrester’s annual developer survey provides the firm’s proprietary adoption data:

SDLC Phase AI Adoption Rate
Coding 48%
Testing 47%
Finding development insights 33%

The coding and testing numbers are notable for two reasons. First, they are significantly lower than Gartner’s projection of 90% adoption by 2028, suggesting a slower reality on the ground. Second, the gap between coding/testing and other SDLC phases confirms that AI adoption remains concentrated in the easiest-to-automate tasks.

Forrester’s State of AI Survey, 2025 (n=1,400+ global AI decision-makers)

The broader enterprise AI picture is sobering:

Metric Finding
Positive EBITDA impact from AI 13% of firms
Can connect AI value to P&L Fewer than one-third
Expect AI payback within 1 year Nearly 50%
Commit to 3-year AI horizons Only 14%
Have cut headcount due to AI 48%
Cite security/risk as top concern 40%
Have documented AI policies ~75%
Mandate responsible AI training Very few

The most alarming finding: nearly half of enterprises have already cut headcount in response to AI, yet change management and employee experience are among the least prioritized investments. This is the organizational equivalent of buying a Formula 1 car and firing the pit crew.

2026 Predictions: The Reckoning Year

Forrester’s 2026 predictions, published October-November 2025, constitute the most bearish major-analyst AI forecast:

Financial correction. Enterprises will defer 25% of planned AI spend to 2027. CFOs will increasingly demand measurable ROI before approving AI budgets. “In 2026, the AI hype period ends as the pressure to deliver real, measurable results from secure AI initiatives intensifies” — Sharyn Leaver, Forrester Chief Research Officer (October 2025).

Shadow AI costs. Ungoverned generative AI in commercial applications will cost B2B companies more than $10 billion in combined damages.

Governance overload. One-quarter of CIOs will be asked to resolve business-led AI failures — projects that business units launched without adequate IT involvement.

Training mandates. 30% of large enterprises will mandate AI training programs, up from ad hoc approaches today.

Developer talent paradox. Time to fill developer positions will double. CS program enrollment drops 20%. The demand shifts from coders to architects — professionals who can design systems, govern AI output, and think across the full lifecycle. VP Research Director Chris Gardner states it bluntly: “Coders — who take requirements, write code, and pass work forward — will die. Developers will thrive” (September 2025).

Vibe coding becomes vibe engineering. The prompt-to-code pattern that defined 2025 (25% of Y Combinator startups in the current cohort have codebases that are 85%+ AI-generated) will expand to encompass the full SDLC — analysis, planning, testing, and optimization — not just code generation.

The Developer Role Bifurcation

Forrester identifies two distinct developer archetypes emerging (Lo Giudice, March-April 2025):

Product Engineers focus on outcomes rather than code mechanics. They accept AI-generated code, request regeneration on failures, and rely on AI for debugging. The bar for entry drops; the bar for judgment rises.

High-Coding Architects maintain deep understanding of coding principles, ensure security and performance standards, review and edit AI-generated code, and provide the context that makes AI output production-grade. These roles become scarcer and more valuable.

Chris Gardner adds a workforce dimension: adoption in regulated sectors (financial services, government) now approaches that of early AI adopters, suggesting IP and competitor code concerns are “largely resolved” for most enterprises. The remaining barriers are organizational, not technical.

How Forrester Differs from Gartner

Dimension Forrester Gartner
Primary lens Developer experience and workflow transformation Vendor evaluation and market positioning
Key framework TuringBots → ASD → AppGen Magic Quadrant for AI Code Assistants
Adoption forecast 48% of developers using AI in coding (2025) 90% by 2028
Tone on AI Bearish on near-term ROI, bullish on long-term transformation Cautious on quality (2,500% defect risk) but bullish on adoption
Prediction style Conceptual frameworks with 4-6 year horizons Vendor rankings with specific market share data
Unique contribution Coined “Agentic Software Development” with formal definition First to evaluate AI code assistant vendors in Magic Quadrant

Both firms agree on the governance gap. Both flag quality risks. But Forrester is distinctly more pessimistic about near-term enterprise ROI and more focused on how development workflows — not just tools — need to change.

Key Data Points

  • 13% of enterprises report positive EBITDA impact from AI (Forrester State of AI Survey 2025, n=1,400+)
  • 25% of planned AI spend will be deferred to 2027 (Forrester prediction, October 2025)
  • 48% of firms have already cut headcount due to AI (Forrester State of AI Survey 2025, n=1,400+)
  • 48% of developers use AI in the coding phase; 47% in testing (Forrester Developer Survey 2025)
  • 33% of developers use AI for development insights — the adoption ceiling for non-coding tasks (Forrester Developer Survey 2025)
  • 30% of large enterprises will mandate AI training in 2026 (Forrester prediction)
  • 20% predicted drop in CS program enrollment (Forrester prediction, November 2025)
  • 2x increase in time-to-hire for developer positions (Forrester prediction, November 2025)
  • $10B+ in projected B2B damages from ungoverned shadow AI (Forrester prediction, October 2025)
  • 4-6 years until AppGen platforms become the industry norm (Forrester, 2024-2025)
  • 376% ROI claimed for GitHub Enterprise Cloud in Forrester TEI study (vendor-commissioned, July 2025 — flag: paid study)
  • 40% efficiency gain in test design, 30% in development, 15% in requirements gathering at Intesa Sanpaolo (single case study via Forrester, September 2025)

What This Means for Your Organization

Forrester’s research delivers a message that most AI vendors do not want executives to hear: the hard part is not selecting a tool. The hard part is the organizational rewiring that makes AI tools productive.

The 13% EBITDA figure should be in every CFO briefing. When nearly half of enterprises expect AI payback within a year but only 13% report actual financial lift, the problem is not the technology — it is the gap between buying licenses and changing how work gets done. Forrester’s finding that 48% of firms have already cut headcount while deprioritizing change management explains the disconnect. You cannot subtract people and add AI without redesigning the workflows those people operated within.

For organizations making AI tool decisions in 2026, the Forrester framework provides a useful maturity ladder. Most enterprises are still in copilot territory — code completion and testing assistance. The firms that pull ahead will be those that move to Agentic Software Development: delegating meaningful multi-step tasks to AI systems while maintaining governance and review. The coming Forrester Wave for ASD tools (H2 2026) will be the first authoritative third-party evaluation of this emerging category. Wait for it before committing to multi-year enterprise agreements.

The developer talent bifurcation is the most consequential workforce finding. If your organization is still hiring primarily for coding ability, you are optimizing for a skill that AI commoditizes daily. The scarce resource is the architect who can design systems that AI assists well, review AI output critically, and govern the full lifecycle from intent to production. Restructuring hiring profiles and career paths now — before the 2x time-to-hire prediction hits — is the highest-leverage move available.

Sources

  1. Forrester State of AI Survey, 2025 (n=1,400+ global AI decision-makers) — Independent survey. High credibility. Cited in multiple Forrester publications including “Three Questions That Will Define AI In 2026” (January 2026). https://www.forrester.com/blogs/three-questions-that-will-define-ai-in-2026/

  2. Forrester Developer Survey, 2025 — Independent survey. Credibility: strong for directional data, though sample size not publicly disclosed. Cited in Forrester’s 2026 software development predictions. https://www.forrester.com/blogs/predictions-2026-software-development-goes-from-jamming-to-full-orchestra/

  3. Forrester 2026 Technology & Security Predictions (October 28, 2025) — Sharyn Leaver, Chief Research Officer. Press release. https://investor.forrester.com/news-releases/news-release-details/forresters-2026-technology-security-predictions-ais-hype-fades-0

  4. “Predictions 2026: Software Development Goes From Jamming To A Full Orchestra” — Diego Lo Giudice, VP Principal Analyst (November 4, 2025). https://www.forrester.com/blogs/predictions-2026-software-development-goes-from-jamming-to-full-orchestra/

  5. “The AI Coding Honeymoon — And What Comes After” — Devin Dickerson, Principal Analyst (January 21, 2025). https://www.forrester.com/blogs/the-ai-coding-honeymoon-and-what-comes-after/

  6. “Agentic Software Development: Defining The Next Phase Of AI-Driven Engineering Tools” — Diego Lo Giudice (March 2, 2026). https://www.forrester.com/blogs/agentic-software-development-defining-the-next-phase-of-ai-driven-engineering-tools/

  7. “Don’t Fire Your Developers! What AI-Enhanced Software Development Means For Technology Executives” — Chris Gardner, VP & Research Director (September 2025). https://www.forrester.com/blogs/dont-fire-your-developers-what-ai-enhanced-software-development-means-for-technology-executives/

  8. “Vibe Coding: Innovation Or Chaos?” — Diego Lo Giudice (March 11, 2025, updated April 15, 2025). https://www.forrester.com/blogs/the-question-is-no-longer-if-but-how-ai-is-transforming-software-development/

  9. “The Future Is Now: TuringBots Will Collapse The SDLC Silos” — Diego Lo Giudice (May 28, 2024). https://www.forrester.com/blogs/the-future-is-now-turingbots-will-collapse-the-software-development-life-cycle-siloes/

  10. “Predictions 2025: GenAI Reality Bites Back For Software Developers” — Christopher Condo, Principal Analyst (October 23, 2024). https://www.forrester.com/blogs/predictions-2025-software-development/

  11. Forrester Total Economic Impact of GitHub Enterprise Cloud (July 2025) — Vendor-commissioned study. Flag: GitHub paid for this research. 376% ROI claim should be treated as upper-bound estimate. https://tei.forrester.com/go/github/enterprisecloud/

  12. “AppGen Is Here: Say Goodbye To Software Development As You Know It” — Diego Lo Giudice and John Bratincevic (2025). https://www.forrester.com/blogs/appgen-is-here-say-goodbye-to-software-development-as-you-know-it/

Created by Brandon Sneider | brandon@brandonsneider.com March 2026