Best Practices for Building AI-Based Intent Signal Workflows Versus Using Third-Party Intent Data Providers

·May 21, 2026 06:55 AM·

Hi everyone, Has anyone built an intent signal workflow using AI — Claude, ChatGPT, or similar? Trying to figure out if training my own AI model to catch intent signals is better or using intent data from providers like Bombora, G2, or Apollo. What's working in practice.

10 comments

· Sorted by Oldest

Federico C.
·
·
Hi Sakshi G., yes! I personally create own custom sigmas using clay building AI agent capable to get any data you need
Ozan D.
·
·
Both can work but they solve different problems. Third-party providers like Bombora and G2 give you aggregated intent signals (what topics companies are researching) but the data is shared with everyone using the same provider, so you're all chasing the same accounts.
Ozan D.
·
·
Building your own with Claude or GPT is better for catching signals that are specific to your ICP — things like monitoring job postings for certain titles, tracking tech stack changes, or parsing earnings calls for keywords that suggest budget shifts. The upside is the signals are unique to you. The downside is you need to maintain the data pipelines and the model needs clear criteria for what counts as a signal vs noise.
Ozan D.
·
·
Practically I'd start with providers for volume and layer in a custom AI workflow for the high-value accounts where you want differentiated signals. That hybrid approach tends to get better results than going all-in on either side.
Sakshi G.
·
·
Ozan D. Yes, that's why I wanted to setup own AI agent that we can train. I tried Bombora and G2 but the kind of intent they give is very broad and not specific plus they are very expensive as well.
Sakshi G.
·
·
Do you have any experience in it?
hatem
·
·
Hey Sakshi G. been hands-on with this exact architecture for a while. Sharing what I've learned the hard way: "set up our own AI agent that we can train" sounds like one project but it's actually 4-5 layers that each need to be solved properly — and training the AI is the hardest part, but only after the lower layers are solid, otherwise you're training on noise.
Shuvrajit M.
·
·
Wdym by training own model? Do you mean setting up a workflow to catch the signals? I think the answer lies on what kind of intent you're looking for, platforms like Bonbora has partnerships with other companies where they share user data with Bambora, you won't be able to access those with your own system. But if they're public like job changes, news, reviews, job postings and anything available in the public internet, your own system will work better
Ilya S.
·
·
If you’re sharing sales collateral and use content in prospecting, you may want to try hummingdeck.com for tracking view duration, per-page engagement etc. That would be first-party intent captured from your own content and the way prospects interact iwth it (at the contact level) - as opposed to third party intent which is sometimes unreliable.
Prasun (.
·
·
I've recently explored https://www.factors.ai/, and it's good. Building something custom using Claude is possible, but I feel it's hard to build a reliable solution.

Federico C.
·
·
Hi Sakshi G., yes! I personally create own custom sigmas using clay building AI agent capable to get any data you need
Ozan D.
·
·
Both can work but they solve different problems. Third-party providers like Bombora and G2 give you aggregated intent signals (what topics companies are researching) but the data is shared with everyone using the same provider, so you're all chasing the same accounts.
Ozan D.
·
·
Building your own with Claude or GPT is better for catching signals that are specific to your ICP — things like monitoring job postings for certain titles, tracking tech stack changes, or parsing earnings calls for keywords that suggest budget shifts. The upside is the signals are unique to you. The downside is you need to maintain the data pipelines and the model needs clear criteria for what counts as a signal vs noise.
Ozan D.
·
·
Practically I'd start with providers for volume and layer in a custom AI workflow for the high-value accounts where you want differentiated signals. That hybrid approach tends to get better results than going all-in on either side.
Sakshi G.
·
·
Ozan D. Yes, that's why I wanted to setup own AI agent that we can train. I tried Bombora and G2 but the kind of intent they give is very broad and not specific plus they are very expensive as well.
Sakshi G.
·
·
Do you have any experience in it?
hatem
·
·
Hey Sakshi G. been hands-on with this exact architecture for a while. Sharing what I've learned the hard way: "set up our own AI agent that we can train" sounds like one project but it's actually 4-5 layers that each need to be solved properly — and training the AI is the hardest part, but only after the lower layers are solid, otherwise you're training on noise.
Shuvrajit M.
·
·
Wdym by training own model? Do you mean setting up a workflow to catch the signals? I think the answer lies on what kind of intent you're looking for, platforms like Bonbora has partnerships with other companies where they share user data with Bambora, you won't be able to access those with your own system. But if they're public like job changes, news, reviews, job postings and anything available in the public internet, your own system will work better
Ilya S.
·
·
If you’re sharing sales collateral and use content in prospecting, you may want to try hummingdeck.com for tracking view duration, per-page engagement etc. That would be first-party intent captured from your own content and the way prospects interact iwth it (at the contact level) - as opposed to third party intent which is sometimes unreliable.
Prasun (.
·
·
I've recently explored https://www.factors.ai/, and it's good. Building something custom using Claude is possible, but I feel it's hard to build a reliable solution.