14 Best Agentic AI Companies to Watch in 2026
Not all agentic AI companies deliver real delegation. Founders learn which ones actually get work off their plate in 2026\.
Skydive 25 min read 
Table of contents
The word “agentic” now covers everything from a chatbot plugin to a system that owns outcomes. Picking from a vendor list without knowing the difference is why most trials fail.
The common assumption among founders and CEOs evaluating AI vendors in 2025 is: “If I pick the right agentic AI company from a reputable list and run a proper trial, I’ll finally get real work off my plate.” That assumption is what makes the word “agentic” so dangerous right now, because it is doing a lot of heavy lifting in 2025, and most of it is dishonest. Founders scanning vendor lists are making shortlisting decisions based on a label that now covers everything from a search-enabled assistant to a system that can receive a goal, break it into sub-tasks, call the right tools, handle failures, and return finished work without a single human prompt along the way.
That gap is the reason trials fail. See our AI teammates for how this works in practice.
“Practitioners struggle to identify the right platform or vendor for building agentic AI systems, reflecting the core problem: most definitions of agentic AI lead people to pick the wrong vendor.”
A genuinely agentic system passes three tests: it perceives a goal without being told each step, it selects and sequences tools on its own, and it handles failure without asking you what to do next. A chatbot with a web-search plugin passes none of them. According to the MarketsandMarkets Agentic AI Market Report (2026), the defining characteristic of agentic AI is the ability to perceive goals, plan multi-step actions, use tools, and execute tasks autonomously without step-by-step human instruction. That definition rules out most of what vendors are currently selling under the label.
These are not points on a spectrum. Robotic process automation handles rigid, pre-scripted workflows where every step is defined in advance. Copilots accelerate a human who is still doing the work.
Agentic AI is the category where the system owns the outcome, not just a step inside it. A founder who deploys an RPA tool expecting autonomous judgment will spend months scripting edge cases. A founder who deploys a copilot expecting delegation will still be the operator.
The MarketsandMarkets report draws this boundary explicitly: RPA automates scripted workflows, while agentic AI handles unstructured, multi-step execution where the path is not known in advance.
Key takeaways
- “Agentic AI” is doing dishonest work in 2025, most vendors using the label are selling workflow automation with a chatbot layer on top, not systems that complete real tasks across tool boundaries without an engineer in the room.
- The vendor with the longest feature checklist is almost never the one that gets work off your plate by Friday, category fit matters more than capability count.
- Agentic AI platforms split into three distinct categories, and enterprise comparison lists only surface one of them, the one built for six-month procurement cycles, not 15-person teams.
- Klarna’s AI assistant handled 2.3 million customer service conversations in its first month, that gap between demo and that kind of output is where most agentic AI trials quietly die.
- A 102-day average B2B evaluation cycle means the cost of picking the wrong vendor is often smaller than the cost of the evaluation itself for a small, fast-moving team.
- SkyDive closes the loop by letting a founder or ops lead deploy AI teammates that handle real work inside Slack, email, iMessage, and web chat, no workflow wiring, no engineer required, operational the same week.
Key Features of Agentic AI Platforms and the One Capability That Separates Real Delegation from Expensive Demos
Scan any vendor’s feature page long enough and a pattern emerges: the list keeps growing, the capabilities sound impressive, and none of it tells you whether a non-technical founder can actually get work off their plate by Friday. The common assumption is: “If I pick the right agentic AI company from a reputable list and run a proper trial, I’ll finally get real work off my plate.” It is one of the most widespread beliefs driving AI purchasing decisions right now, and it is what most feature comparison pages are designed to reinforce without ever actually delivering on.
The Five Features Every Legitimate Agentic Platform Must Have
Every credible agentic AI platform shares the same foundational five. A platform missing any one of them is a smarter chatbot wearing a bigger label:
- Multi-step task execution
- Tool and API integrations
- Persistent memory across sessions
- Multi-agent coordination
- The ability to act autonomously without step-by-step human instruction
These are table stakes. The problem is that checking all five boxes says nothing about who can actually deploy the thing. Feature parity is real. Deployment parity is not.
Three Platform Categories That Share the Same Feature List but Serve Completely Different Operators
The same five features appear across three very different categories. All three can claim multi-agent coordination and agentic AI capabilities in their marketing. Only one of them is built for a 25-person startup with no engineering bench.
| Category | Built For | Deployment Model |
|---|---|---|
| Enterprise orchestration suites | IT-led teams | Six-month implementation windows |
| Developer frameworks | Engineering teams | Custom agent logic written in-house |
| Multi-teammate platforms | Non-technical operators | Work running this week |
According to BCG, 74% of companies have yet to show tangible value from AI investments. The failure is rarely a capability gap. It is almost always a category mismatch.
74% of companies have yet to show tangible value from AI investments
The One Capability That Separates Agentic AI Platforms Worth Buying From Ones That Stall
Plain-language deployment without an engineer is the filter that separates real delegation from expensive demos. Most feature comparison tables bury it entirely, because it is hard to score on a matrix.
How to Choose an Agentic AI Platform Without Wasting a Month on the Wrong Category
Picking the wrong agentic AI platform is a category problem. Most founders discover this in week two, when the polished demo task from day one refuses to extend to a second tool without an engineer in the room.
The Three Categories That Actually Exist (and Why Enterprise Lists Only Show You One)
Agentic AI platforms fall into three distinct categories. Most vendor lists conflate all three under the same “agentic AI” label, and that conflation is the trap. A common pattern among founders evaluating this market is the frustration of comparing platforms that are fundamentally unlike each other, as if LangGraph and a multi-teammate deployment tool belong in the same shortlist.
| Category | Target Organization | Operator Profile | Deployment Speed |
|---|---|---|---|
| Enterprise orchestration suites | 200-plus person organizations | IT-led rollouts | Months |
| Developer frameworks | Teams with available engineering staff | Technical, custom builds | Varies |
| Multi-teammate platforms | 15–50 person funded startups | Non-technical operators | Same-day |
How to Pressure-Test Any Agentic AI Platform Before You Commit to a Trial
Week one of any trial is a controlled environment. One task, one tool, one polished outcome. Week two is where the category mismatch surfaces. Real work crosses functions. It touches the CRM, then email, then Slack, then a calendar. Enterprise orchestration suites handle that, but industry analysts note implementation typically runs three to six months with significant IT overhead, a structural mismatch for a 22-person B2B SaaS team that needs relief this week. Every week spent in the wrong trial is a week the operator is still the operator.
Why Your Team Size Is the Most Important Filter You Are Not Using. Team size determines which category is structurally available to you. Enterprise orchestration suites are genuinely powerful for the organizations they were built for.
What the Multi-Teammate Category Actually Requires From a Platform
The third category is the one most founders in the 15-50 person range should be evaluating, and it is the one most vendor lists bury or mislabel. Multi-teammate platforms are a distinct architectural choice: deployment without engineering involvement, task execution that crosses tools in a single workflow, and an operator model where the person configuring the agent is the same person who needs the outcome by end of day.
The pressure-test question for this category is whether a non-technical operator can extend the task to a second tool, a third integration, and a new workflow type without filing a ticket or waiting for a sprint. If the answer requires an engineer at any point in that chain, the platform has misrepresented its category.
Same-day deployment is the functional threshold that separates this category from the other two. A 22-person B2B SaaS team does not have three to six months and an IT lead.
They have a founder or an ops hire who needs the agent running before the next pipeline review. The multi-teammate category exists precisely because that constraint is real, and the platforms built for it treat deployment speed and operator autonomy as first-order design requirements rather than afterthoughts bolted onto an enterprise architecture.
The 14 Best Agentic AI Companies to Watch in 2026 - Ranked by Who Actually Gets Work Done
Procurement checklists reward vendors who can survive a six-month RFP process, not vendors who can survive your second week. That gap is where most agentic AI trials die, and it is why the list below is organized by a single criterion: who actually gets real work done for a non-technical operator, fastest.
Before reading a single entry, hold this thought. Klarna’s AI assistant handled 2.3 million conversations in its first month after launching globally in February 2024, doing work equivalent to 700 human agents. That number gets cited constantly as proof that agentic AI delivers.
It is precisely the wrong benchmark for a 15-50 person startup. Klarna’s deployment was a single-function vertical build, deep system integration, one well-scoped task domain. The conditions that made it work are the opposite of what a small team’s cross-tool operations actually look like.
Founders who use enterprise vertical wins to validate platform choices are confirming the wrong architecture for their actual problem.
There is a second distortion worth naming before you read a single entry. Agentic AI has become cover for workforce reduction. In documented cases, engineers at companies deploying these systems have been cut in large numbers, not because the technology failed, but because leadership used AI adoption to justify headcount decisions that were already in motion.
That means the benchmark “who gets work done” covers more than speed of deployment. It is about whether the platform removes operational drag from the people who remain, or whether it ships a surface-level wrapper that looks like productivity in a demo and stalls the moment it meets a real workflow. The entries below are evaluated on that harder standard.
The entries below are split honestly: some are the right tool for the right team, some are powerful in the wrong hands, and a few are benchmarks rather than purchases. Read each one with one question in mind: can someone who is not an engineer have real work running by Friday?
1. SkyDive - Best Agentic AI Platform for Operators Who Need Real Work Done
Most founders on this list will feel the pull of the enterprise platforms because the brand recognition is real and the feature lists are long. The hidden cost is that every one of those platforms assumes you have an engineer, an admin, or a six-week onboarding window. SkyDive flips that assumption: an operator describes what they need in plain language and has an AI teammate running inside Slack, email, or iMessage the same afternoon.
That speed matters most to a specific kind of operator, a founder or ops lead who is still doing the work and needs to stop being the operator this quarter, not next fiscal year. The operators we work with consistently face the same two pressure points:
- They cannot protect the time they need for strategic marketing and positioning work.
- They cannot get a day back each week because the coordination overhead, the status updates, the handoff confirmations, the follow-up loops, consumes it before noon.
SkyDive is built to remove exactly that overhead.
The architecture behind that is not accidental. SkyDive is multi-teammate by default, meaning you are not configuring a single bot for a single task. You are standing up a team of AI teammates that coordinate through complex, multi-step workflows without losing track of who owns what at each stage.
Each teammate can run code and commands, read and write files, browse the web, search the web, host web apps, process media, and give itself new tools as needed. Each teammate shows up in Slack, email, iMessage, and web chat, conversation surfaces the team already lives in, and iMessage support is novel across the competitive set, and carries memory and persistence across sessions, so context does not reset every time you return to a thread. Anyone on the team creates teammates. Permissions control view, use, or edit on a per-teammate basis.
Useful teammates spread the way useful things inside companies always spread. Built-in tools and capabilities, scheduled runs, and MCP (Model Context Protocol) support extend what teammates can reach without requiring custom engineering work. Pre-built teammate templates let a non-technical operator start from a working pattern rather than a blank canvas.
The question that separates SkyDive from the field is whether the operator, not the engineer, is the one who deploys it. Platforms that require developers to model offline and on-premise system architecture before anything runs are built for the engineer to build something for the operator, which is a different product solving a different problem. SkyDive is built from a description: you say what needs to happen, and a teammate starts doing it. That distinction is why deployment happens in an afternoon rather than across a six-week onboarding engagement.
Most beneficial when you are the person writing the Slack messages, chasing the status updates, and still blocking on your own strategic work. The goal is to get back a day this week and preserve the space to do the thinking that only you can do.
2. Automation Anywhere - Best for Enterprise RPA Teams Graduating to Agentic AI
Automation Anywhere holds a significant share of the enterprise RPA market and has been extending its platform toward agentic orchestration, making it the natural upgrade path for large operations teams already running bot-based automation. For a 15-50 person startup, it is overkill in almost every scenario: the implementation overhead, licensing structure, and IT governance requirements are calibrated for organizations with dedicated automation centers of excellence. Worth watching as a benchmark for where enterprise RPA is heading, not as a shortlist candidate for lean teams who need relief this quarter.
3. Klarna AI - Best Proof Point That Agentic AI Can Replace Customer Service Headcount
Klarna’s assistant handled two-thirds of all customer service chats in its first month, according to Klarna’s own February 2024 announcement, and the equivalent-to-700-agents figure, corroborated by Perspective AI’s case study, has become the canonical ROI benchmark for agentic AI. What the headline obscures is the architecture: a single-function vertical deployment requiring deep integration with existing ticketing and order systems. It succeeded because the task domain was tightly scoped. Founders should treat this as a proof point for what focused, well-resourced vertical AI can do, not as a model for deploying a general-purpose agentic platform across a startup’s messy, cross-tool operations.
4. GitHub Copilot Workspace - Best Agentic AI for Software Development Teams
Copilot Workspace moves meaningfully beyond autocomplete by letting engineers describe a goal and have the agent plan, scaffold, and execute across a codebase. For software teams, the productivity gain is real and well-documented in engineering communities. The limitation is narrow scope: this is an agent built for developers, operating inside developer tools, solving developer problems. A founder or ops lead trying to delegate sales follow-up, recruiting coordination, or financial reporting will find nothing here that helps. Right pick when your bottleneck is engineering velocity; wrong pick when your bottleneck is everything else.
5. Salesforce Agentforce - Best Agentic AI for CRM-Native Revenue Teams
Agentforce is the most credible enterprise agentic layer for teams whose entire revenue motion already lives inside Salesforce. The agents can act on CRM data, trigger workflows, and coordinate across Sales Cloud and Service Cloud without leaving the ecosystem. The deployment reality for teams without a dedicated Salesforce admin is friction-heavy: meaningful agent configuration typically requires an implementation partner and a multi-week setup window. Most beneficial when you have a Salesforce admin on staff, an existing data model worth building on, and a revenue team large enough to justify the platform cost.
6. n8n - Best Agentic Workflow Builder for Technical Founders Who Want Full Control
N8n has built a substantial developer following, with a large and active GitHub community, and an open-source ecosystem that consistently surfaces it in technical founder discussions about self-hosted workflow automation. The platform gives engineers precise control over multi-step agentic logic, tool connections, and data routing, which is exactly what a technical co-founder building a custom internal tool wants. The tradeoff is direct: n8n makes the operator the engineer. If you are a non-technical founder or ops lead, the control that makes n8n powerful for developers is the same control that will stall your deployment before the first workflow ships.
7. Blueshift - Best Agentic AI for Cross-Channel Marketing Personalization at Scale
Blueshift combines a customer data platform with AI-driven agents that autonomously orchestrate messaging across email, push, SMS, and web based on real-time behavioral signals. Best for growth and retention teams at mid-market and enterprise companies running high-volume lifecycle programs. Tradeoff: it’s purpose-built for marketing, teams looking for a general-purpose agentic platform will find its scope too narrow and its pricing structured for marketing budgets.
8. Arahi AI - Best Agentic AI Platform Tested Across Real Marketing Workflows
Arahi AI has been independently tested across ten marketing workflows including lead scoring, email sequencing, content generation, social scheduling, and reporting, making it one of the more transparently evaluated platforms in the agentic marketing space. Best for marketing ops teams that want evidence-based platform selection rather than vendor demos. Tradeoff: its brand recognition lags behind larger competitors, which can create internal buy-in challenges when presenting to leadership.
9. Fantech Labs - Best Agentic AI Deployment Partner for SMB Operators
Fantech Labs focuses on deploying task-specific AI agents for small and mid-sized businesses in sales, support, and operations, replacing full workflows rather than just automating individual tasks. Best for SMB operators who want a guided deployment partner rather than a DIY platform. Tradeoff: as a services-oriented firm rather than a pure SaaS product, scalability depends on their team capacity and may not suit companies that need instant self-serve onboarding.
10. Siemens Industrial Copilot - Best Agentic AI for Manufacturing and Industrial Operations
Siemens Industrial Copilot brings agentic AI into factory floors and industrial operations, enabling autonomous monitoring, diagnostics, and maintenance scheduling across complex equipment environments. Best for manufacturing enterprises and industrial operators where downtime costs are measured in six figures per hour. Tradeoff: implementation requires deep integration with Siemens’ own hardware and software ecosystem, limiting applicability for facilities running mixed or non-Siemens infrastructure.
11. JPMorgan Chase AI Agents - Best Benchmark for Agentic AI in Regulated Financial Services
JPMorgan Chase has deployed agentic AI across trading, compliance monitoring, and document analysis at a scale that sets the benchmark for what’s achievable in heavily regulated financial environments. Best for financial services leaders evaluating what enterprise-grade agentic AI looks like under strict compliance constraints. Tradeoff: like Klarna, this is an internal build, it functions as a proof point and competitive pressure, not a purchasable solution.
12. Kanerika - Best Agentic AI Consulting Firm for Enterprise Automation Roadmaps
Kanerika specializes in helping enterprise clients identify, design, and deploy AI agents across customer service, sales, and operations, with a strong emphasis on comparing agent types before committing to a build. Best for enterprise transformation teams that need a structured evaluation framework before selecting an agentic platform. Tradeoff: as a consulting-led firm, timelines and costs scale with project complexity, making it less suitable for teams that need rapid deployment.
13. DHL Agentic Logistics AI - Best Agentic AI Case Study for Supply Chain Operators
DHL has deployed agentic AI across its logistics network to autonomously reroute shipments, predict delays, and optimize last-mile delivery decisions in real time. Best for supply chain and logistics leaders benchmarking what autonomous decision-making looks like at global distribution scale. Tradeoff: DHL’s implementation is deeply proprietary and infrastructure-specific, it illustrates what’s possible but offers no transferable tooling for operators outside the logistics sector.
14. Pickaxe.ai - Best No-Code Agentic AI Builder for Non-Technical Product Creators
Pickaxe.ai lets non-technical founders build, brand, and monetize AI agent tools through a studio interface that supports client-facing deployment and usage-based billing, without writing a line of code. Best for product managers and solopreneurs who want to ship an agentic product to paying customers quickly. Tradeoff: the platform’s simplicity caps what’s architecturally possible; teams that outgrow its constraints will face a painful migration to a code-first stack.
Related Reading
- How to Measure AI Agent ROI Before You Scale
- The Future of Agentic AI For Customer Support Explained
- 13 Best Agentic AI Platforms in 2026 for Autonomous Work
- 13 Best AI Agent Integrations to Automate More in 2026
- 14 Best Browser Use Agent Tools for AI Web Automation 2026
- AI Agent vs Assistant: Key Differences Compared (2026)
- How To Measure AI Agent ROI With Proven Metrics
- The Ultimate AI Automation Tutorial for Beginners (2026)
- 13 Best AI Agents for Marketing Teams in 2026
- How to Set AI Agent Permissions Without Losing Control
- What Are Parallel AI Agents? A Complete 2026 Guide
Enterprise Use Cases for Agentic AI and What They Reveal About Which Platform Category You Actually Need
Use cases are not inspiration. They are stress tests. Each one should do one job: reveal which platform category survives your actual workflow and which one collapses the moment work crosses a tool boundary.
The critical insight here is that evaluation itself is not a neutral act: the 102-day average B2B evaluation cycle, layered against a market where that very timeline is the norm, creates a compounding trap specific to small teams. The extended trial process doesn’t just delay the benefit; it actively consumes the same founder and ops capacity the platform was meant to free, so prolonged evaluation becomes arithmetically indistinguishable from permanent inaction. That trap is made worse by a pricing dynamic teams encounter at the door: enterprise suites increasingly bundle agentic AI behind expensive licensing tiers that force an all-or-nothing commitment before the productivity value of those features has been proven.
Paying for seats you can’t yet justify is a bet, and for a 25-person team it is often a bet that kills the initiative before it starts.
Customer Support Escalation and Resolution
The failure is almost never inside the ticketing system. It happens the moment a resolution requires pulling context from the CRM, checking an email thread, and updating the helpdesk record in one continuous action. According to the HubSpot State of Customer Service, cross-system coordination spanning CRM, email, and helpdesk remains a gap that single-agent tools cannot bridge without human intervention. The seam between systems stays human, and that seam is where support quality actually lives.
*Key takeaway: 93% of customers are likely to make repeat purchases with companies that offer excellent customer service. A broken handoff between tools is a revenue event.*
One widely cited industry projection holds that AI will power the vast majority of customer interactions in the coming years, so the infrastructure holding those interactions together carries more weight as automation scales.
Where single-agent tools stall at the seam, SkyDive’s multi-teammate architecture is built for exactly this crossing. A support teammate with multi-channel access can surface the CRM record, read the relevant email thread, draft the resolution, and update the helpdesk record as one continuous workflow, not three separate manual steps. Granular permissions mean the right teammate touches the right system, and memory and persistence mean the context from the first touchpoint is still present at the last. Crucially, that institutional knowledge doesn’t walk out the door when a team member leaves: judgment is made persistent by design, so the next person who picks up a ticket inherits the full resolution history, not a blank slate.
Sales Outreach and Follow-Up Sequencing
A typical B2B close requires six to eight touchpoints across email, LinkedIn, and CRM. Most single-agent tools handle the first email, then the sequence stalls, the contact record never updates, the follow-up never schedules, the reply never gets flagged. Enterprise suites can handle this, but require a dedicated implementation sprint to wire the sequence together. For a 25-person team without an in-house engineer, that sprint never happens, especially when accessing the relevant agentic features requires upgrading to a bundled licensing tier whose cost can’t yet be justified against unproven output. The use case reveals the category mismatch before a single dollar is spent.
SkyDive addresses both failure modes:
- Scheduled runs keep the sequence moving without manual intervention.
- Pre-built teammate templates compress the setup time that would otherwise require an engineering sprint.
- Because the platform is multi-teammate by default, a sequencing teammate and a CRM-update teammate can coordinate the same outreach motion in parallel, each owning a defined step, each operating within its permissioned lane, so the workflow doesn’t stall at a tool boundary.
Internal Ops and Cross-Functional Coordination
Founders consistently describe the same loop: something needs to happen across Slack, email, and a shared project space, and the cost of coordinating it is almost as high as the cost of doing it. Status updates replace focused work. Context gets lost between handoffs. The person who owned the last step has moved on, and nobody has a reliable record of where things stand.
SkyDive’s granular permissions and team collaboration model is designed to break that loop structurally. Complex, multi-step workflows can be coordinated across teammates without losing track of who owns what at each stage, reducing the back-and-forth communication overhead that keeps teams in status-update mode instead of execution mode. Memory and persistence carry context forward across steps and sessions, so the coordination cost doesn’t reset every time the workflow touches a new tool or a new person. The result is ops that accumulate institutional knowledge rather than eroding it.
Next steps
If your real problem is spending weeks on vendor trials only to discover the shortlisted platform still needs an engineer to run it, the path forward starts with matching your team size and technical capacity to the right category before you evaluate a single vendor.
The three-category taxonomy of agentic AI platforms (enterprise orchestration, developer frameworks, and multi-teammate platforms) means that most “agentic AI” shortlists are structurally misleading before a trial ever begins, because all three categories compete under the same label. The 102-day average B2B evaluation cycle means that for a 15-to-50-person team, every week spent comparing vendors is a week the same operational work lands back in the same inbox. Together, they point to one action: deploy a narrowly scoped workflow this week rather than running another extended comparison.
Start with AI teammates to see a multi-teammate deployment built for non-technical operators. Describe what needs to happen, and a teammate starts doing it the same afternoon.
Frequently Asked Questions
What is the difference between agentic AI and traditional RPA or chatbots?
RPA automates rigid, pre-scripted workflows where every step is defined in advance, and chatbots accelerate a human who is still doing the work. Agentic AI is the only category where the system owns the outcome, it perceives a goal without being told each step, selects and sequences tools on its own, and handles failure without asking you what to do next.
What are the three types of agentic AI platforms and how do I know which one fits my team?
The three categories are enterprise orchestration suites (built for 200-plus person organizations with IT-led rollouts that take months), developer frameworks (built for teams with available engineering staff who write custom agent logic), and multi-teammate platforms (built for 15–50 person funded startups where a non-technical operator needs work running the same day). Your team size is the most important filter: enterprise suites are overkill for a lean startup, and developer frameworks require an engineering bench most small teams don’t have.
Why do so many AI trials fail in week two even when the demo looked great?
Week one of any trial is a controlled environment, one task, one tool, one polished outcome. Real work crosses functions: it touches the CRM, then email, then Slack, then a calendar. The failure is almost always a category mismatch, not a capability gap, and 74% of companies have yet to show tangible value from AI investment for exactly this reason.
What features does a platform actually need to qualify as genuinely agentic?
Every credible agentic AI platform must have multi-step task execution, tool and API integrations, persistent memory across sessions, multi-agent coordination, and the ability to act autonomously without step-by-step human instruction. A platform missing any one of these is not truly agentic, it is a smarter chatbot wearing a bigger label.
If I’m a non-technical founder, what’s the most important thing to test before committing to an agentic AI platform?
The single most important filter is whether you, not an engineer, can deploy it yourself using plain language. Most feature comparison tables bury this entirely because it is hard to score on a matrix, but it is what separates real delegation from expensive demos. Platforms that require developers to model system architecture before anything runs are built for the engineer, not the operator.