A company turns on the AI features in its Zoho applications. A month later, somebody asks what changed, and the honest answer is: not much.
When AI underdelivers in a Zoho estate the cause is rarely the model. It is the four foundations the model reads: filled-in data, a process written where the system can follow it, permissions that say who may see what, and an owner for each. The review scores exactly those four on four Zoho CRM screens.
Each chapter starts the video at that point.
A company turns on the AI features in its Zoho applications. A month later, somebody asks what changed, and the honest answer is: not much. The suggestions were odd, the summaries were thin, and the sales team stopped looking at them in the second week. The instinct is to blame the model. In our experience it is almost never the model. It is the four things the model reads, and they can be scored in an afternoon on four screens that are already in Zoho CRM. That is what an AI readiness review is, and this video walks through the four screens so you can score your own before you spend the next budget.
Everything on screen is a mock demo of Zoho CRM with sample data. Six minutes, four scores. The first screen is a report. Not a dashboard, a plain list: every lead and account created in the last ninety days where Industry is empty, or Lead Source is empty, or the company has no phone number and no website. Those are the fields any model would use to decide who a record is. On this mock report, two hundred and ten of three hundred and forty new leads have no industry.
That is not a data quality problem in the abstract. It is the model being asked to prioritise leads it cannot tell apart. The score is a percentage: records with the fields the AI needs, over records created. Under sixty is red. The fix is upstream, at the three doors records come in through, and we have a separate video on exactly that. The review only has to score it. The second screen is the Blueprint list, under Setup, then Process Management. A Blueprint is the sales process written as states and transitions the system enforces: what has to be true before a deal moves from proposal to negotiation, who may move it, and what is required at each step.
On this mock org the list is empty. The company has a sales process. It is in a slide deck and in the head of the sales manager. The AI cannot read either. When it summarises a deal, it does not know that a deal with no decision-maker named should never have reached negotiation, because nobody told the system. The score here is simple: how many of the processes the business depends on are written in the system, as a Blueprint or at least as required fields by stage. Zero is red.
One is a start. The point is not to automate everything. It is that the model reads what the system knows, and the system knows what was written down. The third screen is the Profiles list, under Setup, then Security Control. Zoho ships two profiles: Administrator and Standard. On this mock org that is still all there is, four years in, and thirty-one users are on Standard. Why does that matter for AI? Because an agent acts with the permissions of the person who runs it, and it will answer whatever that person could have looked up.
If every user is Standard and Standard can see every deal, every field and every note, then the agent can be asked anything by anyone, and the first time it reads the acquisition deal aloud to a new hire, the permissions problem becomes a people problem. The score is whether the profiles describe the organisation: a profile for sales, one for support, one for finance, with the fields each may not see marked. Two profiles for thirty users is red. We have a separate video on what an agent can see; the review only has to notice that nobody decided.
The fourth screen is a report of records by owner, and one more column: the date each owner last touched a record. On this mock org, four hundred accounts belong to two people who left the company, and the Industry field, the one that was empty on screen one, has no owner at all. Nobody is responsible for it being right. This is the score people skip, and it is the one that predicts the other three. Data gets filled in when somebody owns it. Processes get written when somebody owns them.
Permissions get decided when somebody owns the decision. The score is a list: for each of the first three foundations, a name. Not a team, a name. Blanks are red. A company that can fill in the three names is usually a quarter away from the AI working; a company that cannot is a quarter away from a cleanup. Put the four scores on one page. Data, thirty-eight percent. Process, zero of three. Permissions, two profiles. Ownership, no names. That is the readiness review, and it took four screens. It also says what to do, in order.
Name the owners first, because the other three do not move without them. Fix the doors the data comes through, so the percentage climbs on its own. Write the one process the business most depends on as a Blueprint. Split the profiles. Then turn the AI back on and watch what changes. The budget for the model is the smallest line in that list. The review exists so you spend the other lines first. If you want to score your own, the four screens are in your Zoho CRM today: a report of empty fields, the Blueprint list, the Profiles list and a report by owner.
If you want us to do it with you, the AI Readiness Review is a ninety-minute working session and a written scorecard across those four foundations, fixed scope, no obligation. We are CodeStringers, a Zoho authorised consulting partner. The link is below.
We design, build, integrate and operate AI-driven, platform-agnostic integrated business solutions for operationally complex companies. Start with a conversation; we will tell you if we are a fit.
