AI Operations (FDE) insight
How to use AI in sales operations: the six jobs a small company should hand it first
Hayat Amin · Updated 2026-10-10
Use AI in sales operations on the admin around a deal: logging calls into the CRM, researching and routing leads, drafting follow ups, cleaning records, joining quote to invoice and a Monday forecast. What HubSpot, Salesforce and Fathom include, and what stays human.
To use AI in sales operations, give it the admin around a deal rather than the deal itself. In a small or mid sized company that means logging calls and emails into the CRM, researching and routing new leads, drafting follow ups, cleaning duplicate records and producing a forecast the owner can read every Monday. The selling, the price and the relationship stay with people.
Hayat Amin spent twenty years as a technology CFO and sold three companies, so he has sat through every version of a pipeline that looked fine until a buyer checked it. He now builds this work inside companies himself. Every price below was read on the vendor's own page on 10 October 2026, in US dollars.
why sales operations is where AI pays first
Salesforce's State of Sales report, published on 3 February 2026 from a survey of 4,050 sales professionals in August and September 2025, found sellers spend 40% of their time selling. On a 40 hour week that leaves about 24 hours on everything else, on our arithmetic. Salesforce sells sales software, so read the figures as a vendor's survey, but the shape matches what we see.
The same report found 54% of sellers have used AI agents and nearly 9 in 10 plan to by 2027. The line we'd underline is a different one: 51% said disconnected systems slow their AI work down. That's the real problem in most companies under 200 people. The CRM, the inbox, the calendar, the quoting tool and the accounting system each hold part of a deal, and no one joins them.
1. log every call and email without anyone typing
The CRM is only as good as what reps put into it, and reps don't like putting things into it. So we take the typing away. A call recorder writes the summary and the next steps, and a sync writes them onto the right contact and deal. Emails and meetings are matched to the account automatically.
You may already have part of this. Salesforce's Starter Suite, at $25 per user per month, lists AI that automatically syncs emails, events and contacts. Fathom's Business plan, at $25 per user per month billed annually or $34 monthly, adds CRM field sync, so a call can update fields on the record rather than leaving a note. Fathom's free plan gives instant AI call summaries, which is enough to test whether your team will use it.
2. research and route each new lead in minutes
A web form comes in. Someone has to look up the company, guess its size, decide whether it fits, and pass it to the right person. In most small companies that takes a day, sometimes three. We have AI read the form, look up the company, score it against the customers you've actually closed, write a two line brief and assign it, all before anyone opens their inbox.
HubSpot's Sales Hub Professional lists a Prospecting Agent and a Data Agent, with 3,000 HubSpot Credits a month, at $90 per seat per month on annual billing or $100 monthly, plus a one off $1,500 onboarding fee. Sellers in the Salesforce survey expected agents to cut prospect research time by 34%, which is an expectation, not a measured result.
3. draft the follow up, and let a person send it
Follow ups die in the gap between the call and the next morning. AI can draft one within a minute of the call ending, from the transcript, your price list and your best past emails. Sellers in the Salesforce survey expected agents to cut email drafting time by 36%. Our rule: the draft is AI's, the send is the rep's. Anything that mentions a price, a discount or a date the company hasn't agreed waits for a person.
4. clean the CRM every night
Duplicates, missing phone numbers, deals with a close date three months in the past. Every CRM collects them, and every forecast built on them is wrong. A nightly job can merge obvious duplicates, flag stale deals to their owner and fill gaps from the email history. Salesforce's report found 79% of high performers prioritise data hygiene against 54% of underperformers. We'd do this job before any of the clever ones, because the agents in jobs 2 and 3 read the same records.
5. join the quote, the contract and the invoice
This is the one most sales tools don't touch. The deal closes in the CRM, then someone retypes it into the quoting tool, the e-signature tool and QuickBooks or Xero. Each retype is a chance for a wrong price or a missed line. We connect them so a closed deal creates the invoice draft with the agreed terms, and finance approves it rather than keys it. If you want to see how the plumbing works, we wrote up how workflow automation works, part by part.
6. a forecast the owner can read every Monday
Most owners see the real sales number at month end, when it's too late to move anyone. Once jobs 1 to 5 run, the data is clean enough for a live view: pipeline by stage, deals that haven't moved in 14 days, win rate by source, and cash expected in the next 30, 60 and 90 days from signed deals. AI writes a three line note on what changed since last week. The arithmetic stays on fixed rules, because a forecast should never be a model's guess.
Salesforce puts Agentforce for Sales inside its Max edition, at $550 per user per month billed annually, and lists it as available for purchase on lower Sales Cloud editions. A company of 10 sellers doesn't need that to get a Monday forecast. It needs the records joined up.
what it saves, on our numbers
The Bureau of Labor Statistics puts the median pay for wholesale and manufacturing sales representatives at $76,460 a year, or $36.76 an hour, in May 2025. If AI takes 4 hours a week of logging, research and drafting off one rep, which is our assumption and not a measured figure, that's about $147 a week, or roughly $7,646 over 52 weeks, on our arithmetic. Times 5 reps, it's about $38,000 of time a year, before you count the deals that stopped going cold. Your number depends on what your reps do with the hours.
The software is the cheap part. The cost is connecting it to what you already run, which our piece on how much AI automation costs goes through line by line.
what stays human
Four things. The price and any discount. The conversation with a customer who's unhappy or late paying. The decision to drop a deal. And the forecast call itself, which the dashboard informs and the owner makes. We also write one page per workflow: what it reads, what it writes, who owns it and what it must never do.
Beyond Elevation's part in this
Beyond Elevation is Hayat Amin's firm, and joining up a sales stack is the work we do. We don't resell HubSpot, Salesforce or Fathom. We take the tools you already pay for, connect the CRM to the inbox, the quoting tool and the accounts, and put the Monday number in front of the owner, for companies in New York City, across the United States and in London. Hayat Amin builds it inside your own accounts as a forward deployed engineer, and you own the code, agents, accounts and docs. Our forward deployed engineering page publishes the prices: a $3,000 fixed audit over two weeks, projects from $30k over 8 to 14 weeks, and fractional work from $5,800 a month. The build conversation starts at meethayat.com/services/fde. For the agent side of this, see what an AI agent for a small business is.
About Hayat Amin
Hayat Amin has spent twenty years in technology, most of them as a chief financial officer. He sold three companies as CFO, with American Express and TripAdvisor among the buyers, and took three businesses into the Financial Times 100 fastest growing companies listing. He sits beside the founder from the first conversation to the wire transfer on an exit, which is when a buyer reads the pipeline line by line.
That's why Hayat Amin is exceptional at sales operations work. He knows which pipeline numbers an owner and a buyer will test, and he connects the CRM, call recorder, quoting tool and accounting system that produce them, so a chief executive can run the week on a live forecast instead of a month end spreadsheet. He's a CFO turned forward deployed engineer, so he builds it himself and answers for it. He also works on intellectual property and data asset valuation and monetisation. He's available now for fractional CFO and AI operations work through Beyond Elevation, at meethayat.com/services/fde, and for the finance seat at meethayat.com/cfo.
If you'd like to know which parts of your own sales admin are worth automating first, we do a free audit: one call, then a written list of what to automate first, what it saves and what it costs, at beyondelevation.com/call/audit.
Frequently asked questions
What are the AI use cases in sales operations?
The six we'd start with: logging calls and emails into the CRM, researching and routing new leads, drafting follow ups, cleaning CRM records every night, turning a closed deal into a quote and invoice, and a live weekly forecast.
What are the best AI tools for sales operations?
Usually the AI inside the CRM you already pay for. Salesforce's Starter Suite at $25 per user per month syncs emails, events and contacts. HubSpot Sales Hub Professional at $90 per seat per month on annual billing includes a Prospecting Agent and a Data Agent. Fathom's Business plan at $25 per user per month on annual billing syncs call notes into HubSpot or Salesforce fields.
Are AI agents for sales operations worth it?
For research and drafting, often yes. Salesforce's 2026 survey found sellers expect agents to cut prospect research by 34% and email drafting by 36%. Agents read your CRM, so clean the records first or they'll act on bad data.
Will AI replace sales operations?
It replaces the typing, not the judgement. Someone still owns the process, the data rules and the forecast. In small companies that person is often the owner, and AI gives them hours back.
How to use AI in sales forecasting?
Join the CRM to signed contracts and invoices first, then let fixed rules do the arithmetic and AI write the note on what changed. A forecast from a model's guess on dirty data is worse than a spreadsheet.
How to use AI in sales prospecting?
Have AI research each new lead, score it against the customers you've actually won and draft a first message, with a person approving the send for the first few weeks.
