What AI tools do you use?

We build on top of the tools you already have - HubSpot, Salesforce, n8n, Make - and layer in AI from OpenAI, Anthropic, and others where it makes sense. No proprietary platform, no new vendor to manage.

Is this actual AI or just rule-based automation?

Both, depending on the job. Simple routing and alerts are rule-based - that's the right tool for that. Scoring, summarization, intent detection, and research use real LLMs. We match the approach to the problem.

Do I need to replace my CRM?

No. We build on top of what you have. Whether that's HubSpot, Salesforce, or something else, we connect to it - we don't replace it.

Do I need a technical team to manage this?

No. The automations run on their own. We build it, hand it off, and your team uses the outputs - alerts, scores, summaries - without touching the infrastructure.

How do you handle our data?

Your data stays in your stack. We don't store it, sell it, or route it through proprietary systems. The pipelines we build connect your existing tools directly.

What does onboarding actually look like?

A 30-minute discovery call, then we scope the work and send a proposal. Once you're in, we go heads-down - no lengthy kickoffs or status meetings. You see working software fast.

What if the AI gets it wrong?

It will sometimes, especially early. That's why we build human review steps where it matters and tune based on your data. Accuracy improves over time - it's not set-and-forget.

How is this different from HubSpot's or Salesforce's native AI?

Native AI works within the walls of that one tool. We connect your entire stack and build logic that runs across it - enrichment, scoring, alerts, and summaries that pull from multiple sources and trigger actions anywhere.

HubSpot AI for RevOps: What's Native, What's Missing, and What to Build

HubSpot’s built-in AI features are useful for content generation and basic lead scoring. They are not enough if you want AI that acts on your pipeline instead of just describing it. Here is the honest breakdown of what’s native, where it stops, and what you have to build yourself.

What HubSpot’s native AI actually does

Breeze AI for content. Email copy, sequences, blog drafts, ad variations. The quality is serviceable for first drafts. It’s not a replacement for knowing your ICP, but it eliminates the blank-page problem at scale and meaningfully speeds up sequence creation for reps who hate writing.

Predictive lead scoring. HubSpot’s predictive model learns from your closed-won and churned contacts to score inbound leads. It works reasonably well once you have enough data - in practice, 200+ closed deals is the rough floor before the predictions become reliable. Under that threshold, the model is pattern-matching against noise. For early-stage teams, stick with rules-based scoring until you have the history.

Breeze Intelligence (formerly Clearbit). Firmographic enrichment baked directly into HubSpot. Contact and company records get populated with job title, company size, industry, and tech stack data without a separate Clearbit integration. This is genuinely useful and removes one layer of integration complexity that previously made enrichment more annoying than it should have been.

Conversation intelligence. Transcription and basic call analysis on Sales Hub Professional and above. You get keyword tracking, talk time ratios, and call summaries. Not Gong. But if you’re not ready to add another tool to the stack, it covers the core use case and keeps the data in one place.

AI forecasting. HubSpot will flag deals it identifies as at risk and suggest close date adjustments. The model works from activity patterns logged inside HubSpot specifically - it doesn’t pull in external engagement signals. More on what that limitation means in practice below.

Where HubSpot AI falls short for RevOps

If you’re running HubSpot with 5-10 reps and a real deal volume, you’ve probably already hit these gaps:

No proactive alerting. HubSpot can surface a stale deal on a dashboard if someone looks at the dashboard. It won’t send your AE a Slack message when a deal goes 14 days without contact. The manager still has to go find the problem. That’s the core limitation: HubSpot AI is reactive. It surfaces information when you ask for it. It doesn’t act on your behalf when it notices something wrong.

Rules-based routing only. HubSpot’s lead routing assigns based on criteria you define: if territory equals X, assign to rep Y. That doesn’t account for rep capacity, recent close rate by segment, or whether the rep who’s “next” in rotation just had a bad quarter on this deal type. Real routing intelligence that adjusts dynamically based on fit and performance requires external tooling.

MEDDIC capture is still manual. HubSpot can transcribe calls. It doesn’t extract MEDDIC fields from that transcript and write them back to CRM properties automatically. Your reps are still either skipping the fields entirely or filling them in from memory after the call - which means the data that downstream forecasting depends on is incomplete or wrong.

Forecasting doesn’t cross-reference external signals. If a deal looks healthy based on HubSpot activity but the champion left the company last week, HubSpot’s model doesn’t know that. Cross-referencing CRM data with LinkedIn activity, email delivery signals, or call transcript sentiment requires connecting sources that HubSpot doesn’t pull in natively.

The three workflows worth building

Deal risk alerts in Slack. Build an n8n or Make workflow that runs daily, queries HubSpot for deals matching your risk conditions - activity gap, probability drift, stage stall, close date slip - reads the deal context via API, runs it through Claude with a structured prompt, and sends a Slack message to the deal owner. Not a dashboard notification. A direct message with the specific issue and a suggested intervention.

This is a 2-3 day build. It replaces the Monday morning “let me find the stale deals” ritual entirely, and it works even when your manager is busy or the CRM review gets skipped for two weeks.

MEDDIC extraction from call transcripts. After each call, pull the transcript from your call intelligence tool or HubSpot’s native transcription, send it to an AI model with a prompt structured around your qualification framework, and write the extracted fields back to custom HubSpot properties. The rep reviews and corrects what the AI captured instead of doing the extraction from scratch.

Field fill rates typically go from 20-30% to 80-90% within the first two weeks. The downstream effect on forecast accuracy is significant - when the model has complete deal data instead of mostly empty fields, the predictions actually mean something.

Automated pipeline review brief. Before your weekly pipeline review, run a workflow that pulls all open deals from HubSpot, calculates velocity and risk signals, formats a structured briefing, and delivers it to Slack 30 minutes before the call. The CRO shows up knowing the headlines. The meeting is about decisions - not about asking reps to remember what they did last week.

Most teams who implement this cut their pipeline review time in half. The time recovered goes into coaching the at-risk deals instead of identifying them.

The rule for deciding what to build vs. what to use native

Use HubSpot’s native AI when the task is: generating content at scale, enriching contact and company records, surfacing data on a dashboard someone will look at, or transcribing calls for later review.

Build custom workflows when the task requires AI to act autonomously - sending a message, writing a field, routing a record - without a human initiating the process. That’s where HubSpot’s native tooling stops and where your actual competitive advantage gets built.

HubSpot is excellent plumbing. It stores the data, connects the teams, and runs the sequences. It’s not an autonomous operations layer. The RevOps teams getting the most out of it are the ones who stopped waiting for HubSpot to ship the feature and built the agent themselves.