Longitudinal persistence rates after gap closure for low-code applications,…

Longitudinal persistence rates after gap closure for low-code applications, bots, and agents

2026-05-17 · agentic-ai mlops-deployment workforce-skills cost-performance low-code-development organisational-design tools-infrastructure · low · source → · wiki →

key claims

No accessible public empirical source consulted in this item publishes a reusable persistence or survival rate after the original gap is closed for enterprise low-code applications, bots, or agents, although relevant evidence could still exist in proprietary analyst reports or enterprise-internal datasetsBinzer et al. (2024)Viljoen et al. (2024)Ajimati et al. (2025)Forrester (2023)Mitchell (2026)
The accessible low-code research base is longitudinally weak on retirement rates but consistently strong on the recurring governance, unofficial asset, and maintenance burden conditions that make persistence likelyViljoen et al. (2024)Ajimati et al. (2025)Mitchell (2026)
Microsoft Power Platform documents inventory, timestamp, owner, and usage telemetry that can support internal persistence-window analysis for applications, flows, and agents if historical records are retainedLearn (2026)Learn (2026)Learn (2026)
Microsoft explicitly documents a six-month inactivity workflow for apps and flows, which provides one concrete persistence checkpoint but should be treated as an operational review trigger rather than as a public benchmark normLearn (2026)Learn (2026)
UiPath's accessible official documentation shows centralized governance, intake shutdown, and dependency-aware deletion controls, while the consulted UiPath material does not itself provide public retirement or abandonment rates for automationsUiPath (2026)UiPath (2026)UiPath (2026)
The strongest accessible bot-retirement guidance supports end-of-life planning and overlap removal as good practice, but it still stops short of reporting measured post-replacement survival percentagesPega (2021)Mitchell (2026)
Governance maturity changes the observability of persistence more clearly than it changes any publicly provable persistence rate, because mature estates capture owners, dependencies, usage, and exit events while weakly governed estates do notLearn (2026)Learn (2026)UiPath (2026)Viljoen et al. (2024)
A practical internal measurement design is to use three, six, twelve, and twenty-four month windows as early decay, first inactivity, medium-term stabilisation, and long-tail persistence checkpoints, while treating those windows as operational conventions rather than public standardsLearn (2026)Learn (2026)Mitchell (2026)

Research Question

What longitudinal evidence exists on persistence rates after the original gap is closed for low-code applications, bots, and agents in live enterprise estates?

Findings

(Populated from §6 Synthesis above.)

Executive Summary

Accessible public research does not currently publish a reusable longitudinal persistence rate for enterprise low-code applications, bots, or agents after the original gap is closed. Relevant longitudinal evidence could still exist inside proprietary analyst reports or enterprise-internal datasets, so this conclusion is bounded to accessible public evidence rather than to all possible evidence. The strongest accessible evidence instead shows repeated governance and maintenance burden conditions that make persistence plausible, together with platform telemetry and lifecycle controls that make internal measurement feasible. The best-supported current conclusion is that persistence after gap closure is a measurable internal lifecycle problem, not a solved public benchmarking problem. Comparisons across governance maturity and ownership models are still decision-useful, but they mostly compare observability and decommission readiness rather than externally published survival curves.

Key Findings

No accessible public empirical source consulted in this item publishes a reusable persistence or survival rate after the original gap is closed for enterprise low-code applications, bots, or agents, although relevant evidence could still exist in proprietary analyst reports or enterprise-internal datasets.
The accessible low-code research base is longitudinally weak on retirement rates but consistently strong on the recurring governance, unofficial asset, and maintenance burden conditions that make persistence likely.
Microsoft Power Platform documents inventory, timestamp, owner, and usage telemetry that can support internal persistence-window analysis for applications, flows, and agents if historical records are retained.
Microsoft explicitly documents a six-month inactivity workflow for apps and flows, which provides one concrete persistence checkpoint but should be treated as an operational review trigger rather than as a public benchmark norm.
UiPath's accessible official documentation shows centralized governance, intake shutdown, and dependency-aware deletion controls, while the consulted UiPath material does not itself provide public retirement or abandonment rates for automations.
The strongest accessible bot-retirement guidance supports end-of-life planning and overlap removal as good practice, but it still stops short of reporting measured post-replacement survival percentages.
Governance maturity changes the observability of persistence more clearly than it changes any publicly provable persistence rate, because mature estates capture owners, dependencies, usage, and exit events while weakly governed estates do not.
A practical internal measurement design is to use three, six, twelve, and twenty-four month windows as early decay, first inactivity, medium-term stabilisation, and long-tail persistence checkpoints, while treating those windows as operational conventions rather than public standards.

Assumptions

This item treats gap closure as the point when an equivalent governed capability exists, even if organisations may disagree on when that capability is fully adopted.
This item assumes that persistence measurement uses retained historical records or repeated extracts rather than a single current-state snapshot.
The proposed 3, 6, 12, and 24 month windows are treated as operational checkpoints because the consulted sources do not provide a stronger public benchmark schedule.

Analysis

The evidence weighs most heavily toward a bounded negative answer: the accessible public literature does not disclose a reusable persistence rate after gap closure. That conclusion is stronger than a simple session-local search failure because the consulted empirical studies, official platform documentation, and adjacent completed repository items all converge on governance mechanisms and lifecycle controls rather than on published survival curves. Platform documentation gives strong evidence for internal measurability, but that strength should not be overstated into a claim about actual cross-firm persistence outcomes. The most decision-useful comparison across governance maturity and ownership models is therefore about whether an organisation can observe and enforce retirement at all, not about whether public benchmarks prove a universal rate difference.

Risks, Gaps, and Uncertainties

Public cross-organisational benchmark data on persistence after the original gap is closed remains absent from the consulted accessible evidence base.
The seeded Forrester and IEEE pages did not produce usable evidence, which leaves this item more dependent on official platform documentation and accessible empirical studies.
The consulted vendor documentation is authoritative for platform capabilities but not independent evidence of actual estate-level retirement outcomes.
Agent-specific public lifecycle evidence is thinner than low-code application evidence, even though Microsoft now includes agents in its inventory surface.

Open Questions

Which enterprises are willing to publish anonymised cohort-level persistence data for low-code applications, bots, and agents across 3, 6, 12, and 24 month windows?
What is the cleanest denominator for measuring persistence after gap closure: all created assets, all approved assets, or only assets whose original bridging need has been formally closed?
Does named central ownership materially reduce long-tail persistence, or does it mainly improve observability and clean-up execution?

Output

Type: knowledge
Description: This item establishes that the strongest current evidence is a gap claim plus an internal-measurement design claim: public benchmarks for persistence after the original gap is closed are not yet accessible, but mature platform telemetry now makes internal lifecycle measurement and retirement analysis feasible.
Links: