AI filters are powerful, but they are not the right tool for every condition. In many cases, the information you need is already available through standard filters, or simply cannot be verified by a web search. Using AI filters in these situations leads to inaccurate results, wasted processing, or both.

This article covers real-life examples of unsuccessful use use cases of AI filters. These will show you when AI filters are not an optimal choice, in addition to providing guidance on what behaviour would enhance your results.

1. They must have Python on their LinkedIn profile

Why it doesn't work: LinkedIn profile data is not publicly accessible. You need an account to view it, and Topo's AI agent cannot log into LinkedIn to verify skills listed on someone's profile.

What to do instead: Use the Skills filter and type "Python" (case sensitive). This filters directly on LinkedIn skill data that is already available in the platform.

2. Asking to verify that people hold senior positions

Why it doesn't work: Two things go wrong here. First, this is a condition about the person, but the filter is set at the company level, so the AI will try to check whether the company is "senior", which makes no sense. Second, even if switched to person level, seniority is already structured data in the platform.

What to do instead: Use the Seniority filter and select Director, CXO, VP, or whichever levels match your targeting.

3. The company has no operations leaders

Why it doesn't work: AI filters work by running a web search. "Company X has no operations leaders" is not something you can confirm through a Google search. The absence of a department or role is not publicly documented on the web.

What to do instead: Use the Operations headcount filter and set it to 0. This uses real data to surface companies with no one in that department.

4. Find companies that hire mobile engineers

Why it doesn't work: Hiring is a signal, not a static company attribute. AI filters are not designed to detect signals. Topo has a dedicated search type for this.

What to do instead: When building your search, write your prompt naturally (e.g. "companies that hire mobile engineers"). The platform will automatically generate a hiring signal filter that checks whether the company has posted relevant job openings, by default in the past 15 days.

5. Is this person an operational manager (criteria set at company level)

Why it doesn't work: Job title information is not something a web search can reliably confirm. This is a person-level condition, but the filter is configured at company level. The AI will try to verify whether the company is an operational manager, which produces meaningless results.

What to do instead: Use the Job titles filter or contact-level keywords (the ones with the people icon on the right) to target operational managers directly. You can even use the “Generate more” button in the filter “Job Title” to have AI infer similar job titles.

6. This person is located in North America

Why it doesn't work: Location is one of the most basic filters available. AI results will not be relevant for this type of ask. There is no reason to use AI filters for something the platform can filter on directly with structured data.

What to do instead: Use the Locations filter at person level and select United States, Canada, Mexico, or simply "North America" from the dropdown.

7. This person is not in pre-sales

Why it doesn't work: Whether someone works in pre-sales is determined by their job title. This is structured data the platform already has. AI filters cannot reliably verify a person's role through web search, and using them for this will produce unreliable exclusions.

What to do instead: Compile a list of pre-sales job titles (Pre-Sales Engineer, Pre-Sales Consultant, Sales Engineer, etc.) and add them to the Exclude Job Titles filter.

8. The company is SaaS, Fintech, DevTools, AI/ML, or Cloud Software

Why it partially works: This is a mixed case. "SaaS" as a broad category already maps to the industry filter Software Development, so using AI filters for that part is unnecessary. However, the sub-verticals (Fintech, DevTools, AI/ML, Cloud Software) are not available as standard industry filters, so AI filters can add value there.

What to do instead: Use the Industry filter for the broad category (Software Development), and keep AI filters only for the sub-vertical conditions that filters cannot cover.

9. Company has between 10 and 75 attorneys

Why it doesn't work: The exact size of a specific department is not something AI filters can reliably determine from a web search. The platform already provides department-level headcount data through its filters.

What to do instead: Use the Legal headcount filter and set the range to 10-75. This gives you accurate, data-backed results.

10. Must work in a university / doesn't work in a middle school

Why it doesn't work: Both conditions can be handled through standard filters. The type of institution someone works at is captured by industry and company name data, making AI filters unnecessary.

What to do instead: Use the Industry filter set to "Higher Education" and add an industry exclusion for "Primary and Secondary Education". Alternatively, use the Company name filter with "University" as an inclusion and exclude typical middle school naming patterns.

What these examples have in common

Every example above shares the same mistake: using AI filters for something the platform already handles through filters or signals. AI filters shine when the information you need lives on the open web and is not captured by any structured data field. When a filter, a signal, or a headcount metric already exists for your condition, that path will always be faster, more accurate, and more reliable than asking AI to verify it through a web search.