BOAT (Business Orchestration and Automation Technologies) is the category Gartner has used since 2025 to group the platforms that orchestrate processes, systems, bots, APIs, human tasks and AI agents from a single place. The 2026 Magic Quadrant, published on 14 September, evaluates 20 vendors. For an SME the practical reading is to consolidate tools, not to buy the biggest one.
What BOAT is and where it comes from
BOAT is a market category, not a product. Gartner defines it as "a consolidated software platform that orchestrates and automates disparate business processes and tasks, with varying degrees of autonomy and complexity, across the enterprise's systems", according to the quotation reproduced by UiPath in its press release on the report. The first edition of the Magic Quadrant was published in October 2025; the second on 14 September 2026, with twenty platforms evaluated.
What sets the 2026 edition apart is the entry condition: to be considered BOAT, a platform must natively offer orchestration and governance of AI agents, multi-agent coordination and connectivity through APIs, through user interfaces and through protocols such as MCP. In other words, Gartner has decided that the control layer for automation and the control layer for agents have to be the same one.
If you have followed the evolution of the labels (iBPMS, RPA, hyperautomation, LCAP), BOAT is the one that brings them together. It does not change what a good BPM does; it changes what analysts expect it to include as standard.
Email, OCR, an approvals spreadsheet, a robot that keys data into the ERP, a shared folder. Bring your list of tools and in 30 minutes we will tell you which ones are surplus, which ones stay and where to start consolidating. No commitment.
What a BOAT platform consolidates
BOAT absorbs six categories that for a decade were bought, licensed and governed separately. The table summarises what each one did and what happens to it inside a consolidated platform.
| Previous category | What it did | Inside BOAT |
|---|---|---|
| BPM / iBPMS | Model, execute and measure processes with human tasks and rules. | The core: the process is still the unit of orchestration. |
| RPA | Robots that mimic clicks and copy data between screens with no API. | One more type of task inside the process, with its own log and retry. |
| iPaaS / integration | Connect applications through APIs and events. | Native connectors; the process calls the ERP or the CRM with no intermediate tool. |
| IDP / OCR | Read documents and extract data. | The first step of many processes; what matters is what happens after reading. |
| Low-code (LCAP) | Build applications and forms without programming. | The way processes and their screens are built. |
| AI agents | Classify, summarise, draft, decide with a language model. | Process steps with a confidence threshold, human validation and a trail. The 2026 entry condition. |
If you want the detail of how agents and deterministic processes coexist, AI agent security in business processes develops it; and RPA and BPM explains why the robot must live inside the process and not the other way round.
Why now: agent sprawl is the new problem
Tool consolidation has been on every IT director's wish list for years. What has made it urgent is AI agents. Three figures sum it up:
- Everyone is adopting them, almost nobody is operating them. Forrester, in The State of Agentic AI in 2026 (3 June 2026), notes that three out of four business leaders say they are adopting agentic AI, but only a minority have it in real production beyond chatbots; and that more than half suffer uncontrolled agent proliferation even after applying governance frameworks such as the NIST AI RMF.
- Many projects will be cancelled. Gartner predicts that more than 40% of agentic AI projects will be cancelled before the end of 2027 because of rising costs, unclear business value or inadequate risk controls.
- The answer is a control layer. As Appian quotes from the 2026 BOAT report, Gartner predicts that by 2030, 70% of companies will adopt a centralised platform to manage AI agent sprawl.
The logic is the same as it was for robots five years ago: an agent that lives outside a process has no defined input, no structured output, no threshold at which it asks a person for help and no log. Inside the process, it has all four. We explain it step by step in from prompts to processes and on the AI agents governed by BPM page.
Eight symptoms of tool sprawl in an SME
You do not need an analyst to know whether you have the problem BOAT describes. If you recognise three or more of these symptoms, you have it.
- A process crosses three tools or more and nobody can say where a specific case file is without asking two people.
- Nobody knows how many robots or automations are running, who created them or what happens when they fail at three in the morning.
- The Zapier, Make or similar account has more than fifty automations and half of them were built by someone who has since left.
- The OCR reads well, but what comes out of the OCR goes into a spreadsheet from which someone approves by email.
- There are at least two generative AI pilots running in different departments, with two different vendors, with no shared traceability and without IT having reviewed them. This is the shadow AI scenario.
- Approvals live in email, and the history of who approved what sits in personal inboxes.
- There are three different user and permission models for automating, and offboarding an employee means touching all three.
- Automation licences add up to more than the ERP and nobody can attribute the spend to a specific process.
None of these symptoms is cured by buying one more tool. They are cured by removing some.
What the Magic Quadrant does not tell an SME
The Magic Quadrant evaluates twenty global vendors on their ability to execute and their vision, and its leaders (Appian, Pega, UiPath and company) are platforms designed for large organisations, with prices, implementation timelines and project teams to match. That does not make them bad: it puts them in a different league. The criteria that will decide your purchase are not those of the quadrant.
| What counts in the Magic Quadrant | What counts in an SME |
|---|---|
| Functional breadth and global scale. | Time to the first process in production: weeks, not quarters. |
| Ecosystem of integrators and partners. | Your own team being able to build and change processes without an integrator. |
| Multi-year product vision. | Predictable pricing per user or per process, with no volume surprises. |
| Multinational presence. | EU hosting, GDPR, ENS and support in your language and your time zone. |
| Generic orchestration capabilities. | Local compliance already solved: VERI*FACTU, B2B e-invoicing, time tracking, certified digitisation. |
| Agent orchestration at massive scale. | AI agents with human validation at the steps that matter and data that never leaves Europe. |
The good news is that the BOAT thesis (one platform, one security model, one trail) is independent of the vendor's size. What an SME has to demand is the same functional scope with the friction of an SME tool. Modern low-code BPM platforms already cover that scope; the question is which ones do it with IDP, signature and governed agents inside and which ones sell them as separate modules.
Ten questions for evaluating an orchestration platform
These ten questions separate a consolidated platform from a set of modules sharing the same logo. Ask for them to be answered with a demo on one of your processes, not the vendor's.
- Are processes modelled in standard BPMN and exportable? If the process design can only be read inside the product, it is a dependency.
- Are human tasks, approvals and deadlines part of the engine, or are they handled with emails and external reminders?
- Is document reading (IDP) native and does it feed the process directly, or is it another vendor with another invoice?
- Which connectors exist for your ERP, your CRM and your payroll, and what happens when there is no connector: API, RPA, or "contact us"?
- Can an AI agent be a process step with a confidence threshold and human validation, and is what it decided and why recorded?
- Where do the AI models run and where is the data stored? For a European company the answer must be the EU, as LLM and RAG with European data details.
- Are electronic signature and the document archive inside, with evidence and legal retention, or are they integrations?
- Is there a single audit trail for people, robots, integrations and agents, that can be consulted without requesting a report?
- How is it billed? Per user, per process, per execution or per document volume, and what happens when you grow.
- How do you leave? Full export of data, documents, trails and processes with no ticket and no fee, as the Data Act already requires from 2027.
If the vendor answers all ten with the same product, you have an orchestration platform. If it answers with four products and a partner, you have what BOAT is meant to replace.
A 90-day consolidation plan
Consolidating is not migrating everything at once. It is proving with one process that the target platform does the work of three tools and, from there, repeating.
| Weeks | What is done | Deliverable |
|---|---|---|
| 1-2 | Inventory of tools that touch processes: annual cost, owner, processes they cross, what happens if they disappear. | One sheet with every tool and its real cost. |
| 3-4 | Map of processes against tools. Choose the first process: the one that crosses the most tools and has a clear metric (cycle time, errors, rework). | Chosen process with its baseline measured. |
| 5-9 | Rebuild the whole process on the target platform: capture, IDP, rules, approvals, ERP integration, signature, archive. AI agents only at the steps where they add value and always with human validation at first. | Process in production with a pilot group. |
| 10-12 | Measure against the baseline. Cancel the tools that process no longer needs. Decide the second process. | Before/after comparison and the first licence cancellation. |
| 13 | Decision: continue with the consolidation plan or stop. With data, not with a vendor presentation. | Documented decision. |
A good first process is usually supplier invoice approval: it crosses email, OCR, approvals, ERP and archive, and its metric (days until the journal entry) is indisputable. To estimate the saving before you start, the ROI calculator gives you the reference figures.
How Dokuflex handles it: the BOAT thesis at SME scale
Dokuflex, a low-code BPM platform with AI, does not appear in the BOAT Magic Quadrant, which evaluates twenty global vendors. What it shares with the category is the thesis: a single platform where processes, documents, integrations, signature and AI agents live, with a single trail and a single permission model.
- Processes in BPMN with human tasks, rules and deadlines in Dokuflex low-code BPM, built by the business team itself and exportable in the standard.
- AI document reading (IDP) and archiving inside the same case file, with document management and legal retention solved.
- Integrations with ERP, CRM and payroll through connectors and APIs in the low-code iPaaS layer, with no intermediate tool.
- AI agents as process steps, with a confidence threshold, human validation and a record of every decision, as described on the AI agents page; models run in Europe.
- eIDAS electronic signature, certified email and Spanish compliance (VERI*FACTU, AEAT certified digitisation, time tracking) already inside, not as a separate project.
- Per-user pricing and exit with no fee. Full export of data, documents, trails and processes from the interface.
What Dokuflex does not do is build the inventory for you. That sheet with the tools and their cost has to be done by your team; what we do is review it with you and tell you, process by process, what would fit inside and what would not.
Bring your list of tools and we will rebuild one of your processes in 30 minutes →
Frequently asked questions
What does BOAT mean at Gartner? +
BOAT stands for Business Orchestration and Automation Technologies. It is the market category Gartner created to group the platforms that orchestrate and automate business processes, systems, bots, APIs and human tasks from a single place, and which from 2026 onwards must also include native orchestration and governance of AI agents. The first BOAT Magic Quadrant was published in October 2025 and the second on 14 September 2026, with 20 vendors evaluated.
Is BOAT the same as BPM? +
Not exactly. BPM is one of the disciplines BOAT absorbs. A BOAT platform includes process management (modelling, execution, human tasks) and adds what used to be bought separately: robotic process automation (RPA), application integration (iPaaS), intelligent document processing (IDP), low-code development and, now, AI agent orchestration. A modern BPM with integrations, IDP and governed agents performs the same function in practice; what changes is the analysts' label.
Why did Gartner create the BOAT category? +
Because the previous categories (iBPMS, RPA, LCAP, iPaaS, hyperautomation) described tools that companies no longer wanted to buy separately. Keeping three or four contracts, three security models and three code bases to automate a single process is expensive and fragile. With the arrival of AI agents the problem gets worse: you need one place from which to govern them. Gartner describes BOAT as the consolidated platform that acts as the control layer for all of that automation.
Does an SME need a BOAT platform? +
An SME does not need a Magic Quadrant vendor, but it does need the thesis behind it: a single platform where processes, documents, integrations and AI agents live, with shared traceability. The Magic Quadrant evaluates twenty global vendors designed for large organisations, with prices and implementation timelines to match. What an SME should look for is the same functional scope on a platform with an implementation time of weeks, predictable pricing, European hosting and local compliance already solved.
Which tools does an orchestration platform consolidate? +
The usual ones in a mid-sized company: the workflow or BPM engine, the RPA robots that copy data between screens, Zapier- or Make-style automations, the OCR or IDP that reads invoices and documents, the document management system where case files end up, electronic signature and, increasingly, the AI agents that classify, summarise or draft. Consolidating does not mean throwing out the ERP or the CRM: it means that process logic, human tasks and traceability live in a single place and that the other tools connect to it.
What is the risk of AI agents without orchestration? +
The risk is twofold. On one side, sprawl: Forrester notes in its 2026 report that three out of four companies say they are adopting agentic AI, but very few have it in real production, and that more than half suffer uncontrolled agent proliferation even with governance frameworks. On the other, cancellation: Gartner predicts that more than 40% of agentic AI projects will be cancelled before the end of 2027 because of costs, unclear value or insufficient risk controls. An agent inside an orchestrated process has an input, an output, a confidence threshold, human validation and a log; outside it, it is an experiment.
How do I start consolidating automation tools? +
With an inventory and a single process. First, list every tool that touches a process (workflow, RPA, automations, OCR, signature, agents) with its cost, its owner and the processes it crosses. Second, pick a process that today crosses three tools or more, for example supplier invoice approval or employee onboarding. Third, rebuild it end to end on the target platform, measure cycle time and errors before and after, and only then cancel the tools it no longer needs. Ninety days are usually enough for the first process.
Sources
- Appian, BOAT Defines a New Future for Automation: publication of the 2026 Magic Quadrant (14 September 2026), twenty vendors evaluated and Gartner's forecast on centralised platforms by 2030.
- UiPath, press release (18 September 2026): Gartner's definition of BOAT and the requirements for agent orchestration and governance.
- Gartner Peer Insights, Business Orchestration and Automation Technologies: market definition.
- Forrester, The State of Agentic AI in 2026 (3 June 2026): adoption versus real production and agent proliferation.
- Gartner, press release (25 June 2025): more than 40% of agentic AI projects cancelled before the end of 2027.
Fewer tools, one process visible end to end
Bring your list of automation tools and a process that today crosses three or more. In 30 minutes we will rebuild it in Dokuflex and tell you what you could cancel. No commitment.