Hey everyone!
In my original post, Maximizing CRM Usage: Proven Strategies to Empower Your Sales Team, we focused on the hard part most teams struggle with first: driving adoption and getting clean, consistent data into your CRM so reps actually use the system every day.
Hopefully by now, you’ve been able to get your team not just into the CRM, but using it as part of their daily rhythm. Fast forward a couple of years and the landscape has shifted.
Assistants like Breeze (HubSpot’s native AI), along with tools like ChatGPT, Claude, and others, can now automate a huge amount of the admin work that used to sit on your reps’ shoulders.
So what happens when logging and data quality are no longer the bottleneck?
This “Part II” is all about that next step: using assistants, custom assistants, and agents inside HubSpot to turn your CRM into a real prospecting engine for your sales team. The approach below isn’t a one‑size‑fits‑all blueprint, but a strategic example of how you can think about structuring AI‑powered prospecting in HubSpot, then adapt the concepts to your own motion, markets, and maturity.
Turn Your ICP Framework into Smart Properties That Power Everything
If you’ve already done the work to define an Ideal Customer Profile (ICP) by vertical and tier, that framework shouldn’t live in a slide deck—it should live in HubSpot.
Here’s a pattern you can use.
Step 1: ICP framework → smart Company Vertical property
Take your ICP framework and:
- Document how to recognize each vertical based on company attributes, such as:
- Firmographics (size, revenue, region, funding, tech stack, etc.).
- Business model and key signals you’re already tracking in the CRM.
- Capture that logic as clear instructions that can drive a smart property:
- Company Vertical (Smart)
Configure a smart property that uses your ICP rules + existing company data to classify Company Vertical automatically. The goal is a consistent, AI-powered vertical on every company record—without asking reps to memorize the rules.
Step 2: Company Vertical → Account Tier (Smart) → native ICP Tier and Industry
Once you have a reliable Company Vertical (Smart), use it as the engine for your next layer: an Account Tier (Smart) property.
That Account Tier (Smart) property should:
- Pull directly from Company Vertical (Smart) plus a handful of supporting fields.
- Pack multiple tiering frameworks into a single configuration, because the way you define Tier 1/2/3 will often differ by vertical (and that nuance really matters).
- Output a simple, rep‑friendly Account Tier label on the company.
From there, use workflows to map into native HubSpot properties:
- Let Account Tier (Smart) drive the native Ideal Customer Profile tier (Tier 1, Tier 2, Tier 3) via workflow.
- Update the native Industry property where needed.
Because ICP Tier and Industry are native HubSpot properties that show up everywhere (reports, lists, filters, scoring, sequences, views), making them your “source of truth” means:
- Reps can easily slice their views by Industry and ICP Tier.
- Managers get out‑of‑the‑box reporting by vertical and ICP fit.
- Workflows, sequences, and prospecting plays can all key off fields that behave consistently across the platform.
Your ICP framework stops being just a strategy doc and becomes live infrastructure that HubSpot can use across your entire go‑to‑market.
Build Personas That Your Prospecting Agent Can Actually Use
Verticals and tiers tell you who to go after. To sharpen the message, you’ll also want clear personas that the Prospecting Agent can pull from.
Step 1: Map personas for your go‑to‑market
Start by defining practical personas per vertical:
- Titles & roles (e.g., VP Sales, RevOps Leader, Marketing Director, Founder).
- Primary pains for each persona in that vertical.
- Desired outcomes they care about (revenue, efficiency, forecasting, CAC, retention, etc.).
- Proof points and angles that resonate most with them.
A spreadsheet is fine to start, as long as it’s specific and real-world.
Step 2: Represent personas in HubSpot (and AI settings)
Next, push those personas into HubSpot in a way the platform (and your agents) can use:
- Fill in HubSpot’s default Persona property with your mapped personas.
- Mirror those same personas and descriptions in your AI settings, so that Breeze, the Prospecting Agent, and other AI features all share a consistent understanding of your audience.
- Use job title, seniority, department, and vertical to:
- Auto‑suggest or tag the most likely persona.
- Let reps confirm or override with one click.
Now every contact isn’t just a job title—they’re tied to a persona your agents actually understand.
Don’t forget your global AI settings: all of HubSpot’s AI tools (including Prospecting Agent and smart properties) pull from these. Make it a habit to keep them up to date as your ICP and personas evolve.
Step 3: Configure the Prospecting Agent to use your context
There isn’t a single “magic prompt” for the Prospecting Agent. Instead, you configure it with:
- The fields you feed in:
- Company Vertical (Smart)
- Account Tier (Smart) / ICP Tier
- Industry
- Persona
- Recent activity and lifecycle context (net‑new vs existing customer)
- Custom instructions and guidelines that tell it:
- How to talk to each persona (“pain points to lean into,” “outcomes to highlight”).
- How to adjust based on vertical and ICP Tier.
- How to use your brand voice (“use our brand tone—clear, helpful, no hype,” etc.).
- What it shouldn’t do (compliance, claims, topics to avoid).
With the right fields and configuration, the Prospecting Agent can generate outreach that is:
- Persona‑aware
- Vertical‑aware
- Tier‑aware
- And consistent with your brand voice
—rather than generic AI copy.
Configure Prospecting Plays Inside the Prospecting Agent
Once your smart properties and personas are in place, the next step is to configure plays directly inside Prospecting Agent so it can go find and work the right accounts for you.
1. Use Audience settings to define who the play should watch
Inside the Audience tab of a play, use your CRM data to tell the agent where to focus:
- Target companies: build segments using properties like Company Vertical (Smart), Account Tier (Smart) / ICP Tier, geography, product fit, etc.
- Signals: choose the buying signals you want the agent to monitor across those accounts (e.g., funding, tech investment, executive hiring, product launch, expansions).
- Target personas: select the personas you want to sell into so the agent knows which contacts to pull at each company.
This is where you turn concepts like “Tier 1 SaaS expansion” or “Tier 2 Manufacturing net‑new” into actual plays the agent can run against your database.
2. Use Selling context and instructions to define how the play should sell
Next, move into the Selling context and instructions for the play:
- Fill in what your company sells, your value proposition, and the pain points you solve so the agent has accurate talking points.
- Specify which products and services you want this play to focus on.
- Add advanced instructions that guide tone and behavior (for example: “use our brand voice,” “be consultative, not pushy,” or “focus on efficiency and ROI for Operations personas”).
You’re not writing a single mega‑prompt—you’re configuring structured context + guidelines that Prospecting Agent uses every time it researches an account and drafts outreach.
3. Let the play handle research and drafting, with tier‑based send rules
With the audience and context set, your play can now:
- Monitor the target accounts for the signals you selected.
- Pull in the right personas at each company.
- Draft:
- Personalized emails and sequences.
- Call talk tracks tied to the value props and pains you defined.
- Follow‑up messages when there’s new activity or a signal fires.
You can still control how much autonomy the agent has by tier:
Tier 2 & Tier 3: Auto‑send (within guardrails)
- Let the play auto‑send outreach when it meets your guardrails (brand voice, compliance, length, etc.).
- This maximizes coverage on mid‑ and lower‑priority accounts without overloading reps.
Tier 1: Review first
- Keep a human‑in‑the‑loop for your highest‑value accounts by requiring review and approval before messages go out.
- Reps start from an 80%‑complete draft instead of a blank page, but still make the final call.
Configured this way, Prospecting Agent plays turn your CRM data, signals, and personas into always‑on, AI‑assisted prospecting instead of one‑off manual campaigns.
Use Custom Agents to Tighten BDR → AE Handoffs
One of the highest‑impact, low‑friction use cases for agents is tightening the handoff from BDR to AE.
Here’s the key distinction:
- Assistants: manually prompted by a human (“Hey AI, do X for me.”).
- Agents: can be triggered automatically (e.g., workflows) and rendered as custom cards where reps already work.
For handoffs, you want a BDR → AE handoff agent, not just an assistant.
Step 1: Define a consistent handoff structure
Document a simple structure for every qualified opportunity, for example:
- Why we reached out (trigger, signal, list, or event).
- What we learned (persona, pains, timeline, budget hints).
- What they engaged with (pages, assets, events, sequences).
- Next best action (what the AE should do in the first call).
This becomes the blueprint for your agent.
Step 2: Let the handoff agent draft the note
Set up a workflow so that when a BDR:
- Sets a qualified meeting, or
- Moves a lead/deal into a certain stage,
your Handoff Agent is triggered.
Have the agent:
- Look at recent activity (emails, calls, meetings, pages, forms).
- Pull in key properties (ICP Tier, vertical, persona, deal size, source).
- Generate a structured, scannable handoff note right on the contact or deal record.
Step 3: Keep humans in control
Make sure:
- BDRs review and edit the agent’s draft in seconds instead of writing from scratch.
- AEs get a tight, consistent brief before every discovery call—less time digging through timelines, more time preparing to add value.
Research at Speed with the Data Agent
Your reps shouldn’t be spending 20 minutes on basic research before every call. That’s what the Data Agent is for.
Here are a few high‑value ways to use it:
Company Snapshot
- “Summarize this company in 5 bullets based on CRM + recent activity.”
- “Given their industry, tier, and product mix, what are the top 2–3 likely pains?”
Persona & Use Case Alignment
- “Based on this contact’s persona and vertical, what 3 problems should we focus on in the first email?”
- “Draft three outreach angles that align with their persona and ICP tier.”
Account Planning
- “Given similar closed‑won deals, what buying committee roles might be missing from this account?”
- “Surface any recent intent or engagement that suggests they may be close to a buying decision.”
The goal isn’t to replace rep judgment—it’s to start every interaction from a much stronger baseline, in seconds instead of minutes.
Other High-Impact Ways to Help Sales with AI
Beyond prospecting and handoffs, here are a few more ways assistants and agents can support your team:
Call and Meeting Prep Assistants
Use AI to:
- Auto‑generate agendas and talk tracks based on:
- Recent activity.
- ICP Tier.
- Persona and prior objections.
- Produce call summaries and next steps that can be dropped straight into deals or follow‑up emails.
Follow-up Drafting
After a call or demo, have an assistant:
- Summarize the conversation.
- Propose 2–3 follow‑up email drafts with recap + clear CTAs.
- Tailor content to stage, tier, and vertical.
Playbook Acceleration
Turn your best:
- Scripts
- Objection handling
- Discovery frameworks
into prompts + playbooks that assistants and agents can reference.
New reps ramp faster because your expertise is embedded in the tools, not just in tribal knowledge.
Treat This Like a Program, Not a “Set It and Forget It” Feature
One important mindset shift: this is not a one‑and‑done project.
Agents and smart properties will make mistakes:
- They’ll mis‑classify a few companies.
- They’ll occasionally get the tone wrong.
- They’ll misunderstand an edge-case persona or vertical.
That’s normal—and expected.
What matters is that you:
- Monitor where they succeed and fail.
- Document mistakes when they happen.
- Look for patterns (by vertical, persona, tier, product line, etc.).
- Make incremental tweaks to instructions, fields, and workflows.
In many cases, yes—you’re building the plane as you fly it. That’s okay, as long as you’re intentionally tuning it over time.
At a bare minimum, I’d recommend:
- Running a monthly AI optimization session (60 minutes is plenty to start) where you:
- Review where your agents and assistants did well.
- Collect examples where they got it wrong or missed the mark.
- Decide on specific optimizations:
- Added context.
- Updated instructions.
- Adjusted field inputs.
- Consider forming a small AI optimization panel:
- A few sales leaders.
- A RevOps or operations owner.
- 1–2 subject matter experts (SMEs).
This group owns keeping your agents sharp, aligned with your strategy, and continuously improving.
Don’t Waste the AI You’re Already Paying For
If you’re on HubSpot today, you almost certainly have AI capabilities and credits bundled into your subscription:
- On HubSpot Pro, that’s typically 3,000 AI credits included.
- On Enterprise, it’s 5,000 AI credits.
Prospecting Agent runs, Data Agent research, and smart property evaluations all draw from that same pool.
Every:
- Targeted prospecting play,
- Research request, and
- Smart property update
is a concrete way to put those credits to work instead of letting them quietly expire.
A few practical starting points:
1. Start with one or two concrete workflows
For example:
- ICP framework → Company Vertical (Smart) → Account Tier (Smart) → native ICP Tier & Industry → Prospecting Agent plays by vertical, tier, and lifecycle.
- BDR → AE handoff agent that turns activity + properties into clean, consistent notes.
2. Put guardrails in your instructions
Bake into your agents and assistants:
- Approved messaging pillars.
- Tone guidelines (“use our brand voice…”) and persona‑specific themes.
- “Never say” lists and compliance constraints.
3. Measure impact in simple terms
Track:
- Time saved per rep.
- Increased coverage of Tier 2–3 accounts.
- Lift in meetings booked or opportunities created from AI‑assisted outreach.
You’ve already done the hard work by getting your CRM into shape and (hopefully) driving adoption.
Now it’s time to let assistants, custom assistants, and agents do more of the heavy lifting—and to take full advantage of the AI credits already included in your subscription, instead of leaving them on the table.
Cheers,
Josh