Progress Software 2026 Predictions

Ed Keisling, Chief AI Officer at Progress Software

AI Plumbing: As the models, frameworks, and standards change, many solutions that have already been built will quickly become outdated, unsupported, or insecure. There will be a significant effort to refactor these pilots which will impede new AI development. Organizations will need to pause to work on the plumbing to define and build the new frameworks and standards needed for their organizations that are scalable, flexible, and secure.

Smaller is better: Models will continue to get larger and more capable, but organizations will find the best value with smaller, tailored, more cost-effective models that they can deploy internally and at the edge.

The hype is real and there is reality in the hype: Generative AI disillusionment accelerates as there is more awareness of AI’s influence on the news cycle through LLM generated audio and video content. The use of AI companions, the saturation of AI content in social media, and adult use of LLM’s all result in more calls for regulation.

Integrate or get left behind: Market disruption will begin with companies that offer non-integrated products. Offerings that lack integration and data/sematic "secret sauce" will be most at risk. Integrated platforms and solutions will gain share by unlocking the value of existing customer data across products.

Mark Troester, Vice President of Strategy at Progress Software

1. Data Security Crisis: AI Infrastructure Becomes Prime Attack Target

We should expect at least one major data breach involving AI infrastructure data through public LLMs. It could be a security breach at an LLM provider exposing conversation data, a sophisticated attack that extracts training data from models, or an insider threat at a major AI company. What makes this particularly challenging is that the value of AI for infrastructure management is real and significant—we can’t just abandon these tools. Instead, organizations should consider adopting a tiered approach to AI usage with public LLMs approved for learning and generic questions, enterprise LLMs with Data Processing Agreements for internal code, and private deployments required for production infrastructure and credentials.

2. Invisible Prompt Injection: The Stealth Threat to Infrastructure AI

Invisible prompt injection, uses techniques literally invisible to humans—zero-width Unicode characters, right-to-left override characters that make text appear one way to humans but differently to AI, homoglyphs that look like normal characters but have different Unicode values, and white-on-white text in rich text formats. Organizations need detection through reasoning chain monitoring, containment through proper sandboxing, rapid recovery capabilities, and continuous learning from observed attack patterns to improve defenses over time.

3. Micro Agents for Repeatable Safe Jobs: The Pragmatic Path

Micro agents make sense for several compelling reasons—lower risk because limited scope means limited blast radius, easier validation since narrow tasks are simpler to test comprehensively, regulatory compliance because focused agents are straightforward to explain and audit, and cost effectiveness since small, specialized models are significantly cheaper than large general-purpose models. Focus on supervised micro-automation rather than full autonomy. Implement tiered automation—fully automated for proven low-risk tasks, approval-required for medium-risk tasks, human-led with AI assistance for high-risk activities.

4. Regulatory Pressure: EU Sets the Pace for AI Governance

AI regulation is coming, and the EU will lead the way with the AI Act, GDPR extensions, and new algorithmic transparency requirements. Organizations using AI for infrastructure management will face significant compliance obligations including data residency requirements. Training data for custom models must comply with GDPR data minimization. Transparency requirements will force operational changes.

5. Economic Reality Check: Preparing for the AI Bubble

OpenAI, Anthropic, and other LLM providers are pricing below cost to gain market share, burning venture capital while many organizations rely on free or heavily discounted AI services. When subsidies end, LLM API prices could increase three to ten times overnight, free tiers will disappear or become severely limited, and organizations will face massive, unexpected AI infrastructure costs. Supply chain disruptions will follow—smaller LLM providers will consolidate or exit, GPU manufacturers may reduce production creating future shortages, and organizations built on providers that exit the market will be stranded. The impact on infrastructure automation could be dramatic.

Philip Miller, AI Strategist at Progress Software

1) Agentic AI Orchestration: people become "managers of agents," not just prompts

The centre of gravity is shifting from single-shot prompts to multi-agent workflows that plan, call tools, verify, and hand off to humans where it counts. McKinsey frames this as "seizing the agentic AI advantage," arguing that redesigned processes, not model choice, create enterprise impact. The countersignal is also useful: Gartner warns of "agent-washing" and predicts >40% of agentic projects will be scrapped by 2027 for lack of clear value.

2) AI Coding moves from assistance to software supply-chain leverage

Coding copilots are quickly becoming table stakes, but the next horizon is end-to-end dev workflows, requirements synthesis, test generation, secure refactoring, and compliance evidence. GitHub’s enterprise research with Accenture found significant time savings and satisfaction gains, and broader industry reporting points to sustained productivity lifts as adoption deepens. Expect SRE and data-engineering agents to join the party, tightening feedback loops between app code, data products, and infra.

3) Scaling valuable AI: from scattered wins to rewired operations

The enterprises that break out next year will operationalize three disciplines: (a) productizing AI use cases with clear owners and SLAs; (b) connecting unstructured and structured knowledge so agents operate in context; and (c) making governance a feature that accelerates—not slows—delivery. McKinsey’s 2025 State of AI shows adoption is high but scaling practices are still rare; organizations that do adopt them capture more value.

4) Agentic Browsing enters the enterprise—starting with low-risk, high-value pilots

The browser itself is becoming an agentic workspace. Perplexity launched Comet, a question-centric browser; OpenAI introduced ChatGPT Atlas, a browser with an "agent mode" that can navigate, summarize, compare, and act on the web. For business, this opens up curated pilots: market scans with source trails, vendor due-diligence drafts with citations, compliance watch-lists, and structured research notebooks, but bound by policy, sandboxed data, and red-teaming to prevent leakage.

5) MCPs & A2A: from plugins to an interoperable agent fabric.

Model Context Protocols (MCPs) turn brittle, one-off integrations into standard, policy-aware connections—so assistants can securely reach apps, data, and tools with least-privilege access and full auditability. Agent-to-Agent (A2A) handshakes then let multiple assistants delegate, verify, and coordinate tasks across vendors and teams.