1cloudconsultants.com - News #Business Automation1cloudconsultants.com - News #Business Automationhttps://www.1cloudconsultants.com/blogs/tag/business-automationMon, 02 Nov 2026 16:36:54 -0800http://zoho.com/sites/<![CDATA[Is Zoho MCP Free? What Zia Agents Really Cost in 2026]]>https://www.1cloudconsultants.com/blogs/post/is-zoho-mcp-free-what-zia-agents-really-cost-in-2026Zoho MCP has no licence fee and Zia Agents are free to build, but model tokens and API quotas still count; here is what you pay, checked 2 November 2026.]]>

AI & Automation

Is Zoho MCP Free? What Zia Agents Really Cost in 2026

Published 2 November 2026 | Pricing checked 2 November 2026 | 5 min read

If you are budgeting for AI on Zoho, the headline from Zoho is simple: Zoho MCP costs $0 and Zia Agents are free. That is accurate as far as it goes, and it matters, because you are not adding a licence line for the connection layer or the agent builder. But a free platform still has meters running underneath it, and a budget that ignores them will be wrong by the second month.

This article sets out what Zoho charges, what it does not, and what is left for you to manage: API quotas on the Zoho services involved, and the cost of the language model your agent uses. Every figure comes from Zoho's own published pages, and each page is listed at the end so you can verify it before you commit.

Pricing checked 2 November 2026. Zoho can change these terms without notice, so confirm the figures on the source pages listed at the end before you commit budget.

What you pay for, layer by layer

Zoho MCP and Zia Agents costs by layer
LayerWhat Zoho chargesWhat still limits you
Zoho MCP$0 licence. No per-user charges and no feature tiers.Your existing API quota on each Zoho service involved. Third-party services use their own quotas.
Zia Agent Studio (build, deploy, manage)Free.API access is restricted to the subscription you hold for the apps the agent uses.
Zoho-hosted models, Standard tierFree up to 30 million tokens a month, then $1.00 per 1 million tokens (1 credit per 1,000 tokens).The free allowance resets on the 1st of each month and does not roll over.
Zoho-hosted model, Pro tier (GLM 5)Free up to 20 million tokens a month, then $3.00 per 1 million tokens (3 credits per 1,000 tokens).Available on request, and not listed for the EU data centre.
OpenAI models through Zoho's keysNo free allowance. Vendor rates with no markup, deducted from your prepaid wallet from the first token.Not listed for the EU data centre.
Your own key (BYOK), including Anthropic ClaudeNothing on the Zia Agents side.Your vendor bills you directly at its own rates.

How the pricing works in practice

Zoho MCP: no licence fee, but it uses your API quota

Zoho describes MCP as included at no additional cost for all Zoho users, with no per-user charges and no feature gates. It adds no new billing logic. Each action runs through the standard API of the service involved and counts against that service's existing daily or monthly allocation. If an agent updates a CRM record and then a Desk ticket, each service uses its own quota. Zoho points to your service level agreement for the exact limits, so pull those figures for your editions before you design anything high volume.

Zia Agents: free to build, billed by model

Zoho states that every agent built, deployed and managed in Zia Agent Studio is free, and that you pay only for model usage. Models bill in tokens, the small chunks of text a model reads and writes. Where the model runs decides your bill. On Zoho-hosted models you get a monthly free token allowance, and your prepaid wallet only starts to drain once you exceed it. Credits convert at 1 USD to 1,000 AI Credits, and credits you buy roll over from month to month.

Bring your own key: Claude, OpenAI or Gemini on your own terms

With BYOK you connect an API key from your own vendor, and Zia Agents charges nothing for the connection. Zoho's model configuration page lists Anthropic among six supported BYOK vendors. Its own worked example is a team that connects an Anthropic key and runs Claude Sonnet 4.6: the cost on Zia Agents is zero and Anthropic bills the team directly. The trade-off is data handling. With Zoho-hosted models your data stays within Zoho's infrastructure, whereas with BYOK it is processed by the vendor you connect.

The licence is free. The meters that remain are API calls and model tokens, and you choose who bills you for the second one.

Four worked examples from Zoho's own documentation

Customer Support
An internal support bot on a Zoho-hosted Standard model

Zoho's example runs Qwen 3.5 35B MoE and uses 25 million tokens in a month. That sits inside the 30 million free allowance, so the model cost is zero.

Operations
A high-volume agent on the Pro tier

Zoho's example runs GLM 5 for 25 million tokens in a month, with the first 20 million covered by the free allowance. The remaining 5 million tokens cost 15,000 AI Credits, which is $15.00.

Sales
Creating a lead through MCP

Zoho's example asks MCP to create a lead in Zoho CRM. The cost is one CRM API call from your existing quota, and Zoho adds no MCP charge on top.

Finance
A multi-step workflow across CRM and Stripe

Zoho's example updates CRM, triggers Stripe and sends an invoice in one run. Zoho counts it as one CRM call plus one Stripe action, and Stripe applies its own quota under its own terms.

How to cost an AI build on Zoho before you start

List every API call

Map each action your agent or MCP client will take to the Zoho service it touches, then compare the monthly volume with the API limits for that edition in your service level agreement.

Choose the model route

Decide between a Zoho-hosted model, OpenAI through Zoho's keys, or your own key. Check that the model you want is available on your data centre.

Estimate tokens against the allowance

Multiply tokens per task by tasks per month. Compare the total with the 30 million (Standard) or 20 million (Pro) free allowance, then price any excess at 1 credit per 1,000 tokens on Standard or 3 credits on Pro.

Load the wallet and watch usage

Zoho shows your credit balance and usage by model and agent under Settings, then Plan & Billing. Review it monthly while usage settles.

What this means for UK and Irish businesses

Check your data centre first. Model availability depends on the data centre your Zoho organisation sits on. On Zoho's model configuration page, the EU data centre lists the Zoho-hosted Qwen models and GLM 4.7 Flash, plus BYOK for OpenAI, Anthropic, Gemini, DeepSeek and Cohere. OpenAI through Zoho's keys and GLM 5 are not listed for the EU. Zoho also says support for more data centres, including the UK, is on the way, so recheck the list before you design around it.

Budget in the right currency and add VAT. Zoho publishes these rates in US dollars and says local taxes such as VAT are charged on top. For a UK or Irish finance team, that means some exchange rate exposure and a VAT line on any credits you buy. Both belong in the business case from the start.

Our position: plan around API quota, and treat the model as a choice rather than a given. We are a Zoho Authorized Partner and a Zoho AI Enabled Partner, and we favour Claude for most of our customers' Zoho work. We say so openly, and the cost logic above applies to whichever model you pick. If you want a second opinion on an AI design before you build it, we are happy to have that conversation without the sales pressure.

Sources, all from zoho.com and checked on 2 November 2026
  • Zoho MCP
  • Zia Agents
  • Zoho Partner UK
  • Zoho AI Enabled Partner
  • AI
  • AI Pricing
  • Business Automation
  • Zoho API Limits

Planning AI on Zoho?

Tell us what you want your agents to do and we will help you work out the API and model costs before you build. Happy to have that conversation without the sales pressure.

Book a Zoho Discovery Call
]]>Mon, 02 Nov 2026 11:15:15 +0000<![CDATA[Why AI Should Clarify Your CRM Data, Not Disguise It]]>https://www.1cloudconsultants.com/blogs/post/why-ai-should-clarify-your-crm-data-not-disguise-itAI can make every customer record read the same, masking the signals your team relies on: here is how to keep AI a clarifying tool, not a flattening one.]]>

Why AI Should Clarify Your CRM Data, Not Disguise It

Every business using Zoho CRM now has AI touching customer data somewhere: a Zia-generated summary, an AI-drafted follow-up email, a chatbot transcript condensed into a case note. Used well, this saves your team hours. Used carelessly, it can quietly erase the very detail that made a record useful in the first place.

That detail matters more than it looks. A customer's own words, their tone, how long or short their message was, whether they wrote in frustration or in passing, all of this used to tell a rep something about how to respond. When AI rewrites every interaction into the same polished, neutral summary, that signal disappears, even though the record looks tidier than ever. The question for any business adopting AI in its CRM is not whether to use it, but how to make sure it sharpens the picture rather than flattening it.

Three Ways AI Touches a Customer Record

ApproachWhat It DoesRisk to Data Quality
Manual entryA person types notes in their own words, at their own lengthInconsistent formatting, easy to skip fields under time pressure
AI-generated summaryZia or a similar tool condenses a call, email or ticket into a short noteCan standardise every record into the same tone, losing urgency or nuance
Agentic interpretationAn AI agent reads the record and takes or recommends the next actionCompounds any signal already lost at the summary stage, since the agent only sees what was written down

How the Signal Gets Lost

What AI Summaries Tend to Flatten

Ask an AI tool to summarise a support ticket or sales call and it will usually produce something accurate, concise and, on its own, indistinguishable in tone from every other summary it produces. That consistency is often the point. But it means a genuinely urgent, frustrated customer and a mildly curious one can end up described in near identical language, unless someone has deliberately asked the AI to preserve tone and urgency as fields in their own right.

Where Zia Fits In

Zia's role in Zoho CRM is to draft, summarise and suggest, not to make the final call. That is a sensible design, but it only works as intended if the team using it treats Zia's output as a first draft rather than a finished record. Left unedited, a Zia-generated note can sit in the system looking authoritative while quietly omitting the one line that mattered most.

Rebuilding the Signal Through Structured Fields

The fix is not to abandon AI summaries, it is to separate what AI is good at (structure, brevity, consistency) from what a human still needs to supply (judgement calls on urgency, sentiment and priority). Adding a small number of structured fields, such as a manually set urgency flag or sentiment tag, keeps that judgement visible in the record instead of hoping it survives inside a paragraph of prose.

AI should sharpen the signal in your customer data, not flatten it into generic prose.

Where This Shows Up in Practice

Sales
Lead notes that all sound the same

When every AI-summarised call note reads with the same even tone, reps lose the early warning signs of a hesitant buyer versus a ready one.

Customer Support
Tickets that hide genuine urgency

A furious customer and a mildly annoyed one can generate near identical AI-written ticket summaries unless urgency is captured as its own field.

Marketing
Segments built on flattened data

Segmentation built from AI-smoothed records can miss the outliers and edge cases that often make the best case studies and referrals.

Operations
Agentic workflows acting on thin signal

An agent that triggers actions from CRM records will only ever act as well as the signal left in those records by the humans and AI tools before it.

A Practical Process for Keeping the Signal

1
Capture the raw input first

Keep the original call recording, email or transcript attached to the record, even after a summary is generated, so nothing is truly lost.

2
Let AI draft, not decide

Treat Zia-generated summaries as a starting point for a human to edit, not a final entry to leave untouched.

3
Add structured fields for judgement

Use dedicated fields for urgency, sentiment or priority so those judgements survive independently of the prose summary.

4
Audit the workflow, not just the output

Periodically check whether AI-assisted records are missing the detail that used to inform how your team responded, and adjust the process accordingly.

What This Means for UK and Irish Businesses

Most businesses adopting AI inside Zoho CRM are focused on speed: faster ticket resolution, faster follow-up, faster reporting. That is a fair goal, and Zia and similar tools genuinely deliver on it. The point of this article is not to argue against that adoption, but to flag a cost that is easy to miss because it does not show up as an error message. Records that look complete can still be missing the judgement a human once supplied automatically.

This is exactly where the kind of Zoho consultancy we provide adds value beyond the initial setup. Configuring the right structured fields, deciding which processes genuinely benefit from AI drafting versus full automation, and building in a review step for AI-generated content are governance decisions, not technical afterthoughts. Getting them right at the outset is considerably cheaper than untangling a CRM full of well-formatted but under-informative records eighteen months later.

If you are rolling out Zia or any AI-assisted workflow in your Zoho environment and want a second opinion on where the structure needs reinforcing, we are happy to have that conversation without the sales pressure.

Related Reading

ZiaZoho CRMZoho Partner UKAIBusiness AutomationAgentic AIData GovernanceCRM Data Quality

Want a second opinion on your AI-assisted workflows?

We will look at where structure is helping and where it might be quietly costing you signal, then talk through what is worth changing.

Book a Zoho Discovery Call
]]>Mon, 21 Sep 2026 09:06:00 +0000