Ask a sales rep what they did yesterday and listen for how much of it was selling. Usually it’s not much. Updating the CRM, digging through old email threads for a pricing detail, rewriting the same follow-up for the fourth time, chasing legal about a contract that’s been “nearly done” since Tuesday.
Salesforce has tracked this for years. In its 2022 State of Sales survey of about 7,700 sales professionals, reps said they spent just 28% of their working week actually selling. Everything else was admin, internal meetings and paperwork.
That’s the gap the current wave of AI sales tools is chasing. The pitch has moved well past chatbots. The idea now is AI built for sales specifically: software that reads your CRM, ranks your pipeline, drafts the outreach, summarises the call and keeps track of the contract after signature. Some of it delivers, and some is a fancier autocomplete with a sales-flavoured logo.
The useful question isn’t which tool is “best.” It’s which part of your sales process is actually broken.
A quick reality check on what “AI” means here
Under the label “AI sales tool” you’ll find a few quite different technologies:
- Predictive models that score leads or forecast revenue from historical CRM data
- Generative AI that writes emails, proposals and call summaries
- Conversation analytics that listen to recorded calls and look for patterns
- Agents, the newest category, which take a goal (“research these 40 accounts and draft first-touch emails”) and work through the steps with less hand-holding
Vendors blend these freely, so it’s worth asking which one a given feature actually uses. A lead score built on three years of your own closed-won data is a very different thing from a score built on generic firmographics.
CRM busywork: the least glamorous, most valuable use
Nobody gets excited about CRM hygiene. It’s still where AI earns its keep fastest.
Good systems pull details out of calls and emails, suggest field updates, flag deals with missing information and nudge reps toward the next step. When the CRM is accurate, everything downstream gets better: forecasts, territory planning, handovers to customer success.
Lead scoring sits right next to this. Old-style scoring was a rules spreadsheet: +10 for a VP title, +5 for opening an email. AI scoring looks across far more signals, such as site visits, email engagement, past interactions, company traits and deal history, and learns which combinations actually led to wins.
One caveat that vendors rarely put in bold: the scores are only as good as your data. If half your closed-lost deals were never logged properly, the model learns from a distorted picture.
Outreach: AI makes it easy to be annoying at scale
Generative AI turned personalised email from a skill into a button. A rep can now write fifty tailored openers before lunch, each mentioning the prospect’s industry, role and some recent company news.
The trap is obvious once you’ve been on the receiving end. Fifty AI-tailored emails to the wrong fifty people is still spam, just with better grammar.
Mailbox providers have tightened the rules, too. Since 2024, Gmail and Yahoo have required bulk senders to authenticate their domains, offer one-click unsubscribe and keep spam complaint rates below 0.3%. Blast poorly targeted AI copy and your domain’s reputation can sink. Then even your good emails land in junk.
The teams getting this right use AI to research and prepare, and let a human decide who actually gets contacted and what the message says.
After “yes”: where deals leak value quietly
Most sales tech goes quiet the moment a deal closes. That’s a mistake, because the contract is where a lot of revenue slips away.
Think of renewal dates nobody noticed until the auto-renew window had passed. Notice periods buried on page 14. Payment terms that sales agreed to and finance never heard about. None of it is dramatic, and all of it costs money.
This is where AI contract management fits in. GetAccept’s version reads agreements, both its own and contracts uploaded from elsewhere, and pulls out the fields that matter: parties, start and end dates, renewal terms, notice periods, contract value and payment terms. Each extracted field carries a confidence indicator. You can click through to the exact passage it came from, which matters, because nobody should trust an AI-extracted renewal date without being able to check it. The system then flags auto-renewals and expiring notice windows before they lapse.
GetAccept also covers the stretch before signature, with proposals, digital sales rooms, e-signatures, and AI tools that draft content and build business cases from call transcripts. The appeal is having one thread from the first proposal to the renewal reminder, instead of five tools that don’t talk to each other.
The main players, sorted by the job they do
| Tool | Where it’s strongest | Best fit |
| Salesforce (Agentforce / Einstein) | Forecasting, opportunity scoring and agents built into Salesforce CRM | Teams already deep in Salesforce |
| HubSpot (Breeze) | Copilot, prospecting agents and data enrichment in one CRM | Growing companies running marketing and sales on HubSpot |
| Gong | Call recording, conversation analytics, deal risk | Sales leaders who want to know what’s really said on calls |
| Outreach | Sequences, engagement workflows, prioritisation | Outbound-heavy teams with structured cadences |
| Apollo | Contact database plus prospecting and outreach | Teams that want data and engagement in one place |
| GetAccept | Proposals, deal rooms, e-signature, AI contract data | Teams whose bottleneck is the late stage and post-signature |
A few notes the table can’t fit.
Salesforce is in the middle of a branding shift. Einstein, which has been around since 2016, still powers the predictive features like lead and opportunity scoring. Since Dreamforce 2024, though, the company’s AI push has centred on Agentforce, its platform for autonomous agents. If you’re already on Salesforce, the integration advantage is real. If you aren’t, it’s a big ecosystem to adopt just for AI.
HubSpot bundles its AI under Breeze. That means a Copilot assistant inside the CRM, pre-built agents for prospecting, content and customer service, and Breeze Intelligence for filling gaps in contact and company records. It’s at its best when marketing and sales already live in the same HubSpot instance.
Gong is a different kind of tool altogether. It isn’t trying to be your CRM. It records and analyses customer conversations, so managers can coach on real calls and spot deals going sideways, rather than relying on whatever reps type into notes fields.
Outreach and Apollo overlap in sales engagement, but Apollo bundles a large contact database, which appeals if you’re currently paying separately for data and sequencing.
How to choose without regretting it
Start with the bottleneck, not the feature list. Ask your reps where their week actually goes, then match the answer to a tool:
- Hours lost to CRM entry: start with automation and call summaries
- Pipeline full of junk: look at scoring and prospect data
- Outbound that isn’t landing: consider engagement tools, but fix targeting first
- Deals stalling late, or renewals slipping: proposals, deal rooms and contract tools
Then check integration before anything else. An AI tool that doesn’t sync properly with your CRM doesn’t save time. It creates another data silo, plus a weekly ritual of copying things between tabs.
It’s also worth knowing how often these projects go wrong. In one recent survey of development agencies, around two-thirds said their AI clients had a failed attempt behind them. Treat that as directional, since it comes from agencies describing their own clients. The causes they named were poor data, weak workflow fit and missing governance, not bad models. In other words, the boring stuff.
Finally, ask the uncomfortable security questions. Where does your customer data go? Is it used to train anyone’s models? Who can see call recordings? Sales data is full of pricing, contacts, and contract terms, which is exactly what you don’t want leaking.
Where this is heading
The next step is already visible: AI agents that handle multi-step work, such as researching an account, drafting outreach and logging the results, with a person approving the output rather than doing every step. Salesforce, HubSpot and others are all building in that direction.
That may well push the 28% figure up. It won’t replace the part of selling that actually closes deals: understanding a buyer’s problem and earning their trust. The best tools free up time for exactly that. The worst ones just help you send more email.
Related: How to Choose a Perth Digital Marketing Agency in 2026: What to Ask About AI
