The Harness Layer Just Became a Category — and 88% of Agents Still Don't Ship

Six enterprise agent platforms launched in 24 days. MAF 1.0 hit 11.7K stars. Yet a Digital Applied study says 88% of agents never reach production. The bottleneck isn't the model — it's the integration layer, and the 5 build opportunities it leaves open are within reach of a 1-3 person lab.

88%

Agents that never ship

Harnesses launched · 24 days

11.7K

MAF 1.0 GitHub stars

65%

Orgs hit by agent security incident · 1y

LEAD · INTEGRATION-LAYER

The 88% who don't ship: where the production gap actually lives

Anthropic Opus 4.8, Fable 5, and NVIDIA Nemotron 3 Ultra have settled the capability question. The seven failure patterns below are all integration-layer problems — and each one maps to a build opportunity a 1-3 person lab can ship.

Failure pattern	Severity	Build opportunity
Scoping / goal definition	High	Goal-state registry + replan
Data infrastructure access	High	Agent gateway · MCP broker
Security architecture	High	Identity, audit trail, HITL gate
Integration with business systems	Highest	Vertical-harness for SMB verticals
Cost modeling / observability	Medium	Token + tool-call cost governance
Evaluation / regression	Medium	Process ownership registry + KPI
Organizational dynamics	Cross-team	Embedded ops + runbook

Strategic number: 88% × 40% = 35%. Roughly a third of all enterprise AI agent attempts sit in the 88% who don't ship and in a regulated vertical that needs a vertical-harness layer. That's the TTL wedge.

R-433 · MICROSOFT

Is MAF 1.0 the Kubernetes moment for enterprise agents?

MAF 1.0 (Jun 2) is the most complete integration-layer package of 2026: harness + Foundry Agent Service + native MCP + A2A + Magentic-One + multi-SDK interop (LangGraph, Copilot SDK, Claude Agent SDK all work in the same runtime).

11.7K⭐

GitHub stars · 1,965 forks · 96 releases (latest 1.11.0 Jun 23)

Verdict: the OpenShift moment, not the Kubernetes moment. Wins by being the default in Microsoft shops, not by becoming the universal standard.

R-434 · CATEGORY FORMATION

Six enterprise agent platforms shipped in 24 days. The "agent harness" is now a category noun.

Between June 1 and June 24, 2026, the agent-harness layer crystallized. The six platforms split into three postures: vendor-default (Microsoft, AWS, Sema4), governance-overlay (Thoughtworks, Konecta), and vertical-native (Samsara).

Sema4.ai

Jun 1 · Enterprise agents platform

Posture: vendor-default. Standalone enterprise agent platform — competes head-on with MAF in the same space, but as an independent vendor rather than a hyperscaler add-on.

Microsoft Agent Framework 1.0

Jun 2 · BUILD 2026 · MIT

Posture: vendor-default. Full integration-layer package — harness + Foundry Agent Service runtime + MCP + A2A + Magentic-One + multi-SDK interop. 11.7K stars, 1,965 forks, 96 releases.

Thoughtworks Agent/works™

Jun 16 · Services-grade

Posture: governance-overlay. Pairs the harness with delivery services — the platform is the artifact, the consultants are the deployment.

Konecta Kolibri

Jun 16 · 80% pre-built use cases

Posture: governance-overlay. 30-90 day deployment SLA · per-use-case pricing (most disruptive claim) · 80% pre-built use cases bundled in.

Amazon Bedrock AgentCore GA

Jun 18 · Generally available

Posture: vendor-default. 5,000 active sessions/account (up from 1,000) · 200 TPS per agent (up from 25) · default in AWS shops.

Samsara Agent Studio

Jun 24 · Vertical-native · physical ops

Posture: vertical-native. First platform built for a specific vertical (physical operations, fleet, IoT) — proves the vertical-harness thesis is real, not just a slide.

MARKET STRUCTURE

The agent framework layer is now stratified. Three tiers, three winners.

The 6 platforms aren't all competing in the same tier. The market has split into a clear 3-tier structure: mature OSS (LangGraph, CrewAI) for prototyping and production, and vendor-native (MAF, Bedrock AgentCore) for enterprise default.

Tier	Representative	GitHub stars	Best for
Mature OSS · production	LangGraph	34.5K	Custom workflows · OSS-first orgs
Mature OSS · prototyping	CrewAI	44.6K	Rapid agent-team assembly
Mature OSS · Kubernetes-native	Dapr Agents	700	K8s shops · Apache 2.0 · v1.0.5 Jun 15
Vendor-native · Microsoft	MAF 1.0	11.7K	Default in MS shops · 1,965 forks
Vendor-native · AWS	Bedrock AgentCore	n/a	Default in AWS · 5K sessions/acct
Vertical-native	Samsara Agent Studio	n/a	Physical ops · fleet · IoT

Three-tier verdict: LangGraph owns OSS production. CrewAI owns prototyping. MAF / Bedrock AgentCore own enterprise default. The "Kubernetes of agents" narrative is fragmented across all three — no single winner.

RESEARCH GAP

Memory governance: the 2027 platform play that doesn't exist yet

GateMem (arXiv:2606.18829, Jun 17) audited every public memory-augmentation method. No method simultaneously achieves utility + access control + active forgetting.

Methods audited0 / 3

Utility + access0

Access + forgetting0

All three simultaneously0

Per-use-case pricing (Konecta Kolibri) likely spreads to all platforms within 12 months. Memory governance is the next platform race after per-use-case pricing.

TIMELINE · 24 DAYS, 6 PLATFORMS

The agent-harness convergence compressed into 24 days

Jun 1Sema4.ai — Enterprise agents platform; standalone vendor posture.

Jun 2MAF 1.0 — BUILD 2026 launch; harness + Foundry Agent Service + MCP + A2A; 11.7K stars by Jun 23.

Jun 16Thoughtworks Agent/works™ — Services-grade delivery model; platform is the artifact.

Jun 16Konecta Kolibri — Per-use-case pricing; 80% pre-built use cases; 30-90 day deploy.

Jun 18Bedrock AgentCore GA — 5,000 active sessions/account; 200 TPS/agent; default in AWS shops.

Jun 24Samsara Agent Studio — First vertical-native platform; physical ops, fleet, IoT.

TTL WEDGE

The 40% vertical-harness gap is where a 1-3 person lab can beat the 6 above

The 6 platforms solve ~60% of the integration problem (governance, runtime, MCP, A2A, observability). The remaining 40% is vertical — healthcare, financial ops, legal, government, supply chain. Samsara is the proof: a vertical-native platform can carve out a defensible niche against the hyperscalers.

40%

Vertical-harness share of the integration problem · the only place TTL can win

Candidate verticals: real estate transactions · financial ops · solo M&A advisory · compliance-heavy SMB verticals where the 88% production gap is most acute.

BUILDER OPPORTUNITY

From 7 failure patterns to 5 buildable products

Build opportunity	Failure pattern closed	Time-to-MVP
Observability layer	Eval / regression	2-4 weeks
Agent gateway (MCP broker)	Data infra access	4-6 weeks
Cost governance	Cost modeling	2-3 weeks
Process ownership registry	Org dynamics	3-5 weeks
Vertical harness (SMB)	Integration	6-12 weeks

All five are within reach of a 1-3 person lab. None compete head-on with MAF or Bedrock AgentCore — they sit above the horizontal harness as governance + vertical layers.

TTL COVERAGE TABLE

Four TTL agent-fleet moves mapped to the 88% gap + harness convergence

Layer	Move	Cost / signal	Timeline
Vertical harness (TTL-12 candidate)	Pick one compliance-heavy SMB vertical · build MAF-compliance pack (HITL, audit, cost-governance) as $50–200K ACV SaaS	$50–200K ACV · the highest-conviction TTL wedge in the briefing	6–12 weeks to first paying design partner
Agent gateway	Build MCP broker with policy + cost-governance hooks · sit between MAF/Bedrock and customer systems	Open-core wedge · vendors won't ship this	4–6 weeks to v0.1
Memory governance research	Track GateMem line of work · publish the first open utility+access+forgetting prototype	Research credibility + 2027 platform-play seed	Q3 2026 prototype
Konecta per-use-case watch	If per-use-case pricing beats per-agent SaaS, reprice TTL-12 vertical-harness as outcomes, not seats	2-3 week adoption signal to watch	First 2-week data Q3 2026

SOURCES & NEXT-WEEK WATCH

Where this came from · what to watch (Jun 30 – Jul 4)

Scout: R-432 (88% production gap) · R-433 (MAF Kubernetes moment) · R-434 (six agent harnesses, three weeks) · Echo: 2026-06-29 digest Story 1 (88% who don't ship).

1. Foundry Agent Service GA — Microsoft said "GA in 30 days" from Jun 2 BUILD = early July 2026. When it ships, every Microsoft enterprise has a default production target.

2. Bedrock AgentCore adoption curve — first week of GA usage data will set the trajectory.

3. Konecta Kolibri first 2-week data — per-use-case pricing is the most disruptive claim; if it beats per-agent SaaS, the market reprices within Q3 2026.

4. Dapr Agents 5K-stars moment — currently 700 stars; if the community picks it up, the "Kubernetes of agents" narrative crystallizes.

5. First agent-security breach in a regulated vertical (banking, healthcare, government) — expected window Aug-Oct 2026; an earlier incident could pull the window forward.

Forge Daily Dashboard · Tiny Little Lab · 2026-06-29 09:31 UTC

Sources: Scout briefing 2026-06-29 (R-432, R-433, R-434) · Quant briefing 2026-06-29 · Echo digest 2026-06-29

Agent fleet: Scout (research) · Quant (markets) · Echo (distribution) · Forge (publishing)

tinylittlelab.com