AI orchestration software starts between $9 and €173,30 monthly. For 400 to 2.000 monthly work items, costs run roughly $9 to $1.000 monthly. The amount depends less on staff numbers than on billing units: users, executions, credits, or actions.
This page compares public prices for Microsoft Copilot Studio, Salesforce Agentforce, UiPath, n8n, Make, Zapier, and Relevance AI, plus Bombos as a different offer. Prices were taken from vendor pages/documentation on September 30, 2026, in original currencies without conversion. We calculate identical work for ten, twenty-five, and fifty people, examine price jumps, and identify unlisted costs.
An orchestration platform joins steps, systems, and agents into one process: email arrives, one agent reads, another finds the case, another prepares a reply. See orchestration versus ordinary automation.
The short answer
- Make Core starts at $9 monthly annually billed for 10.000 credits; n8n Starter €20 for 2.500 executions (Make; n8n, 30-09-2026).
- Copilot Studio sells shared organizational capacity: €173,30 monthly for 25.000 credits (Microsoft, 30-09-2026).
- Salesforce charges $500 per 100.000 Flex Credits and 20 per standard action, $0,10 per action (30-09-2026).
- UiPath and Relevance AI list no full orchestration price on their main pages. UiPath Basic starts $25 monthly; Standard/Enterprise require quotes (UiPath, 30-09-2026).
- At 2.000 monthly items and five paid actions each, our example runs from €20 n8n to $1.000 Salesforce for identical action counts.
- One review minute per item, 1.000 items, €60 hourly costs €1.000 monthly, exceeding almost every software bill here.
- Bombos charges €2.000 for three months, then €1.000 monthly excluding VAT, first-task setup included (30-09-2026).
Which pricing models do platforms use?
Four units dominate: users, whole workflow executions, credits/tasks per step, and organization-wide capacity packs. A fifth is a configured service’s fixed monthly fee.
Per user. Accounts cost money. Agentforce charges $5 per user plus credits, or $125 user add-ons for unlimited employee use (Salesforce). Zapier Team supports at most 25 platform users (Zapier).
Per execution. n8n counts whole runs regardless of steps: “All plans include unlimited users & workflows and every integration” (n8n).
Per step. Make says “1 operation equals 1 credit” for ordinary modules; heavier AI modules may cost 2 or 10 (Make). Zapier counts tasks; Relevance Actions are “counted when an Agent runs a Tool” (documentation).
Shared capacity. Microsoft sells 25.000-credit packs across the organization or Azure pay-as-you-go (Microsoft).
The same workflow can mean one unit, five, or twenty-five. Executions, tasks, actions, and credits differ.
What do the seven platforms cost?
Public rates range from $9 to contact sales. Prices below were read September 30, 2026; annual means the monthly amount with annual billing.
| Provider | Model | Public rate | Limits |
|---|---|---|---|
| Copilot Studio | Capacity/usage | €173,30 for 25.000 credits; US pay-as-you-go $0,01 each | Task credits depend on action |
| Agentforce | Credits/conversations/users | $500 per 100.000; $2 conversation; $5 user plus credits; $125 add-on; Agentforce 1 Editions from $550 user | Four different scopes; CRM separate |
| UiPath | Monthly plan | Basic from $25; Standard/Enterprise quotes | Basic limited users/robots; agents at Standard |
| n8n | Executions | Annual Starter €20/2.500; Pro €50/10.000; Business €667/40.000 | Model/setup separate; Business self-hosted only |
| Make | Credits | Annual Core $9, Pro $16, Teams $29 per 10.000; monthly $10,59, $18,82, $34,12 | Some AI modules 2/10 credits; own model bill separate |
| Zapier | Tasks | Annual Professional from $19,99/750; Team $69/2.000 | AI costs 1 to 5 tasks per step and tool call |
| Relevance AI | Actions/Vendor Credits | Annual/monthly Pro $19/$29; Team $234/$349; main page only Enterprise on request | Docs: $80 per 1.000 extra Actions; model via Vendor Credits |
| Bombos | Fixed | €2.000 three-month pilot, then €1.000 monthly excluding VAT | First-task setup included; service rather than build-it-yourself license |
Row sources: Microsoft, licensing guide, Salesforce, UiPath, n8n, Make, Zapier, model pricing, Relevance, Bombos. All accessed 30-09-2026.
Nearly every usage model also has fixed charges, so fixed monthly does not mean unlimited. Entry plans require upgrades or top-ups at growth. Full UiPath/Relevance orchestration starts with quotations rather than calculations.
What does this mean for ten, twenty-five, or fifty people?
Our SME example runs between €20 and $1.000 monthly. Staff affect work volume rather than directly setting costs. Compare identical work.
Assumptions, not measurements: two items per employee per working day, twenty working days monthly. Each request is matched to a case and prepared for review in one run with five paid actions: read, search, match, draft, record.
| Quantity | 10 people | 25 people | 50 people |
|---|---|---|---|
| Monthly items | 400 | 1.000 | 2.000 |
| n8n executions | 400 | 1.000 | 2.000 |
| Paid actions, × 5 | 2.000 | 5.000 | 10.000 |
| Microsoft credits, 5/action | 10.000 | 25.000 | 50.000 |
| Salesforce credits, 20/action | 40.000 | 100.000 | 200.000 |
| Make credits/Zapier tasks/Relevance Actions, 1/action | 2.000 | 5.000 | 10.000 |
Action rates: Microsoft, Salesforce, Make, Relevance, 30-09-2026.
| Route | 10 people | 25 people | 50 people | Qualification |
|---|---|---|---|---|
| n8n Starter annual | €20 | €20 | €20 | Volume fits; model extra |
| Make Core annual | $9 | $9 | $9 | 50 people fill 10.000-credit bundle |
| Zapier Professional annual | $49 | $89 | $129 | One platform user |
| Zapier Team annual | $69 | $119 | $169 | At most 25 platform users |
| Microsoft whole packs | €173,30 | €173,30 | €346,60 | 25.000-credit packs |
| Microsoft US usage | $100 | $250 | $500 | $0,01/credit, no euro price |
| Salesforce standard actions only | $200 | $500 | $1.000 | Licenses/contract separate |
| Relevance Team monthly | $349 | $349 | $589 | 7.000 Actions; extra 3 × $80 |
| UiPath Standard | Quote | Quote | Quote | No public price |
| Bombos after pilot | €1.000 | €1.000 | €1.000 | Fixed guided service; no identical built process |
Zapier’s slider lists 2.000 tasks at $49/$69, 5.000 at $89/$119, and 10.000 at $129/$169 for Professional/Team with annual billing (pricing).
This translates one design into meters rather than ranking quality, connections, or builders. Bombos is a configured service alongside licenses. See build yourself or hire an agency.
What happens with per-user pricing?
Fifty licenses cost five times ten even when few people use them. Ask how many need accounts.
| Monthly license | 10 users | 25 users | 50 users |
|---|---|---|---|
| Copilot Business €18,20 annual, excluding VAT | €182 | €455 | €910 |
| Agentforce $5, credits excluded | $50 | $125 | $250 |
| Agentforce add-on $125 | $1.250 | $3.125 | $6.250 |
Sources: Microsoft, 30-09-2026, then discounted €15,60 annually or €21,84 monthly, alongside an eligible base subscription; Salesforce.
User licenses do not automatically cover background work. Microsoft includes agents only “using the authenticated Microsoft 365 Copilot USL user’s identity.” Triggers other than “When an agent calls the flow” consume credits; Computer-Using Agents are excluded (updated 14-09-2026). Nighttime mailbox processing does not sit in a user’s seat. Calculate accounts first, workload second. Using staff count in both can double-count or miss background work.
Why does the bill jump?
Bundles and blocks create steps rather than smooth growth. One extra item can change tiers.
Microsoft: half-used packs are expensive. €173,30 / 25.000 = €0,0069 per credit at full use; using 10.000 makes €0,0173, two and a half times as much (Microsoft). Ten-person use fills 40%. US packs cost $200 versus $0,01 usage; below 20.000 monthly credits, usage costs less (September 2026 licensing guide). No Dutch usage price permits direct conversion.
Microsoft: capacity runs out. Custom agents disable at 125% prepaid use. Flows block earlier: “When a tenant’s prepaid Copilot Studio capacity is fully consumed, new agent flow runs are blocked” (documentation). Peaks can halt work unless pay-as-you-go is enabled.
Relevance: top-ups can equal a plan. Pro includes 2.500 Actions. For 5.500, $29 + 3 × $80 = $269. At 5.501, four blocks make $349, equal to Team’s 7.000-Action monthly price (docs). Purchased Actions roll over, distinguishing cash spending from consumption.
Make: unused credits expire at period end. Top-ups are 1.000 or 10.000 (pricing).
Do multiple agents cost more than one?
Only paid model/tool calls or executions add cost. Packaging can bill identical collaboration once or five times.
n8n Execute Sub-workflow counts “only the parent (top-level) execution” (30-09-2026). Two thousand parents calling four children remain 2.000 executions. Five separately webhook-triggered workflows mean 10.000. Manual tests and empty polls do not count; scheduled triggers count each firing.
Zapier charges reasoning: “Tasks used per run = (1 × model rate) + (number of tool calls × model rate).” Advanced rate 3, Premium 5, own model key 1 (30-09-2026). One AI step with four tool calls costs 5, 15, or 25 tasks, giving 5.000, 15.000, or 25.000 per 1.000 runs. New paid-plan steps default Premium. At 75 tasks, a run pauses for your approval.
Microsoft agent actions cost 5 credits, while 100 flow actions together cost 13 (documentation). Reasoning-agent execution is almost forty times pricier per step. Agents remain useful precisely outside fixed patterns.
Three agents do not imply triple cost. Your execution log reveals paid units per completed item.
Which costs are absent from price pages?
Model usage, building/maintenance, and human review often exceed SME licenses. Missing means unknown rather than zero.
Models. Own-key Make users pay two bills: “You pay Make for credit usage based on operations, and pay your AI provider for token usage” (Make). n8n Cloud Starter/Pro Gateway credits are “billed from your instance’s prepaid credit balance”; own keys mean “Separate billing per provider” (n8n). Relevance charges $20 per 10.000 Vendor Credits separately (docs). Microsoft reasoning models add 10 credits per 1.000 tokens.
Tests. Zapier full tests consume production-equivalent tasks: “Full-Zap tests currently consume tasks the same way production runs do” (updated 07-09-2026). n8n/Microsoft flow tests do not. Relevance counts failed tool runs as Actions.
Building/management. Someone builds, connects packages, maintains changes, and handles exceptions. See existing software and implementation/guidance.
Review. One thousand extra one-minute checks mean 16,7 monthly hours, €1.000 at an assumed €60 hourly. See errors and review.
Cost per approved item = (licenses + usage + model + recovery + review + management + setup) / approved items. Failed/duplicate items are excluded from the denominator. See full administrative costs.
Why are most price comparisons wrong?
Comparisons mix units, stale prices, or vendor contradictions. Three checked examples on September 30, 2026:
Hostingdirectory, 22-07-2026 claims $19,99 for 2.000 Zapier tasks. Zapier lists $49 Professional. $19,99 buys 750.
Relevance docs state $80 per 1.000 extra Actions and “The Free plan is retired for both new signups and existing accounts” (docs). Its public pricing-new still shows free and $40. We use $80. New Dutch-account terms require a quote.
UiPath Basic $25 sits beside Standard, but agents belong to Standard: “Build, test, run, and share enterprise automations including Agents” (UiPath). “Orchestration from $25” is imprecise. See orchestration versus RPA.
What should a good quote show?
Three counts: accounts, paid units per item, and expected approved monthly items. Without them, fair comparison is impossible.
That is our position derived from the numbers. Increasing the example’s user count fivefold leaves n8n/Make unchanged, increases Microsoft/Zapier/Relevance in steps, and increases Salesforce/per-user costs proportionally. In every case one review minute outweighs most platform differences.
The cheapest license is rarely the cheapest choice. A five-times-task-cost model can win if approval takes ten seconds rather than a minute. Cheap platforms may simply check less. Compare equal quality, authority, and failure handling. See choosing a platform.
Where is this heading?
Prices do not demonstrably fall; meters become more detailed. Four of seven providers changed AI-step billing between June and September 2026.
| Date | Change | Source |
|---|---|---|
| June 15, 2026 | New Zapier AI steps priced by model class; old steps stay 1 task | Zapier |
| July 2026 | Work IQ API use excluded from Copilot user coverage | guide |
| July 15, 2026 | AI by Zapier repeated tool calls generally available, billed per call | Zapier |
| August 2026 | Copilot Studio execution harnesses have different billing rules | guide |
| September 2026 | Relevance retires free, splits Actions/Vendor Credits; n8n adds Gateway credits | Relevance, n8n |
Providers separate ordinary work and agent reasoning, charging by model, tool, or token. Limits follow: Zapier pauses at 75 tasks, Microsoft supports monthly agent limits. Such brakes follow bill surprises.
Our expectation: by September 30, 2027, at least three of seven show estimated money before runs and actual money afterward, including models, in standard management interfaces. Today they show units and separate calculators. Increasingly detailed meters require money per run for budgeting; existing brakes show awareness.
This fails if per-user packs return, model bills remain structurally separate, or custom quotes dominate. Credits alone or calculator links do not count.
Until then, comparisons age within six months. Bombos charges one monthly amount; deciding which agent performs each step is our work.
What can this comparison not yet show?
It prices a specified design rather than actual identical processes everywhere. Only a same-item trial proves that.
- No invoices: no verified comparable Dutch bills; quotes may discount or impose minimums.
- Assumptions control results: five paid actions is chosen. Salesforce flows/Relevance tools can bundle steps; Make can count more.
- Currencies/VAT: five publish dollars, n8n/Microsoft euros. No conversion, VAT not always explicit.
- No UiPath Standard/Relevance Enterprise public price: request quotes.
- Conflicting Relevance prices: $80/$40. Check logs and contract versions.
- Bombos differs: configured guided service rather than license. No published volume cap; ask what your work includes.
How does Bombos approach this?
€2.000 three-month pilot, then €1.000 monthly excluding VAT, same for everyone (Bombos): €11.000 year one. More than n8n/Make licenses, less than hiring. Setup and three-month guidance distinguish the offer.
Chef assigns work, Wegwijzer guides, specialists fit your tasks/packages. Bombos reads email/documents, finds cases, and prepares supported proposals. You approve, edit, or reject in Bombos. Payments, customer messages, and contracts require your approval. Review becomes part of work. Corrections become rules for coworkers.
Your team then teaches new tasks without technical skills or another hired builder. See piloting and more work with the same people.
Sources
Each source was opened on September 30, 2026. Each quotation appears verbatim.
- Microsoft, Copilot Studio prices: https://www.microsoft.com/nl-nl/microsoft-365-copilot/pricing/copilot-studio (30-09-2026)
- Microsoft Learn, “Billing rates and management,” updated 14-09-2026: https://learn.microsoft.com/en-us/microsoft-copilot-studio/requirements-messages-management
- Microsoft, “Microsoft Copilot Studio Licensing Guide,” September 2026: https://go.microsoft.com/fwlink/?linkid=2320995
- Microsoft, Copilot prices: https://www.microsoft.com/nl-nl/microsoft-365-copilot/pricing (30-09-2026)
- Salesforce, “Agentforce Pricing”: https://www.salesforce.com/agentforce/pricing/ (30-09-2026)
- UiPath, “Plans and Pricing”: https://www.uipath.com/pricing (30-09-2026)
- n8n, “Plans and Pricing”: https://n8n.io/pricing/ (30-09-2026)
- n8n, “Understand executions”: https://docs.n8n.io/build/understand-workflows/understand-executions (30-09-2026)
- n8n, “Use Gateway credits”: https://docs.n8n.io/build/understand-workflows/use-gateway-credits/ (30-09-2026)
- Make, “Pricing”: https://www.make.com/en/pricing (30-09-2026)
- Make, “Credits”: https://help.make.com/credits (30-09-2026)
- Zapier, “Plans & Pricing”: https://zapier.com/pricing (30-09-2026)
- Zapier, “AI by Zapier model tier pricing”: https://help.zapier.com/hc/en-us/articles/46425475442829-AI-by-Zapier-model-tier-pricing (30-09-2026)
- Zapier, “Migrating from Agents to AI by Zapier,” updated 07-09-2026: https://help.zapier.com/hc/en-us/articles/47402591569805-Migrating-from-Agents-to-AI-by-Zapier
- Relevance AI, “Pricing”: https://relevanceai.com/pricing (30-09-2026)
- Relevance AI docs, “Pricing”: https://relevanceai.com/docs/get-started/pricing (30-09-2026)
- Relevance AI alternate pricing: https://relevanceai.com/pricing-new (30-09-2026)
- Hostingdirectory, “n8n vs Zapier vs Make,” 22-07-2026: https://hostingdirectory.nl/blog/n8n-vs-zapier-vs-make/
- Bombos, “What does AI cost in a small business?”: https://bombos.ai/en/knowledge-base/what-does-administrative-automation-with-ai-cost/ (30-09-2026)