Summon ← Blog
Sales

What the AI sales employee does all day

Summon · September 28, 2026 · 7 min read

Most founders who try AI for sales aim it at the wrong thing. They want it to close. The real unlock is the other 80% of sales work — research, writing, follow-up, and CRM hygiene — that eats an SDR's day before they ever get to a conversation.

Here is a concrete walk-through of what a Summon AI sales employee actually produces across a working day. Not "AI-assisted outreach." Actual tasks, actual outputs.

Morning: prospect research and list building

The AI sales employee starts with the founder's ICP — a description of the company type, size, and problem that Summon solves. From that brief, it builds a research brief for a new segment: say, bootstrapped SaaS founders between $500k and $3M ARR who have job posts open for an operations role.

The output is a structured prospect list with company name, founder name, evidence of fit (job post, funding status, company description), and a one-line personalisation hook for each. Not a scraped CSV. A researched brief ready for outreach.

What comes out: a prospect brief — 15 to 30 companies with name, fit evidence, and a personalisation hook for each. The founder reviews the list and approves it before any outreach is drafted.

Late morning: first-touch outreach drafts

With the list approved, the AI sales employee drafts a first-touch email for each prospect. Each email is personalised to the hook from the research — not a mail-merge token, a genuine reference to something specific about that company or founder.

The structure: one-line observation on their situation, one-line statement of what Summon does, one-line reason it's relevant to them, one soft ask (a yes/no question, not a calendar link). Under 80 words. Unformatted plain text.

Every draft surfaces in the founder's approval queue. The founder reads, edits if needed, and presses send. The AI employee does not send without approval.

Afternoon: follow-up sequences for open threads

Sales is mostly follow-up. The AI sales employee tracks which outreach threads have gone unanswered and drafts follow-up messages for each one that has aged past a defined window (default: 4 business days).

Each follow-up is short — one new angle, one question, no restating the original pitch. The AI employee surfaces the draft with context: original message date, what was sent, and a suggested subject line variation.

The follow-up rule: three touches maximum before the thread is marked inactive. The AI employee tracks this automatically; founders do not need to manage a spreadsheet.

End of day: CRM updates and pipeline summary

At the end of each day, the AI sales employee logs every interaction — drafts sent, follow-ups queued, replies received — into a structured update. For founders not using a CRM, this is a running deal log. For founders with a CRM, it is a formatted update ready to paste or sync.

The daily summary includes: threads opened today, threads that received a reply, threads moved to active pipeline, threads closed or marked inactive. One paragraph, five numbers, no narrative fluff.

What it does not do

An AI sales employee does not get on calls. It does not negotiate. It does not read the room in a Zoom. And it does not send anything without the founder approving it first.

This is the design. Sales requires human judgment at the moment of commitment. The AI employee handles the preparation and follow-through that humans consistently underprioritise because it is unglamorous and time-consuming.

The volume question

A human SDR runs 30 to 50 personalised touches per day at the top of their range. An AI sales employee can prepare 100 to 200 per day — but outreach volume is not the bottleneck in most early-stage companies. Quality and follow-through are.

The right use of an AI sales employee is not to blast more email. It is to ensure that every prospect on your list gets a thoughtful first touch, every unanswered thread gets followed up, and nothing falls through because the founder ran out of time.

TaskHuman SDR timeAI sales employee
Prospect research (20 companies)2–3 hoursDelivered by morning
First-touch drafts (20 emails)1–2 hoursDrafted for approval
Follow-up queue management30–45 min/dayAutomatic, approval-gated
CRM updates20–30 min/dayDaily summary auto-generated
Pipeline summary for founderAd hoc / weeklyDaily, one paragraph

What this costs

A full-time SDR in the US costs $50,000 to $80,000 per year plus benefits, commission structure, and management overhead. A part-time sales contractor runs $2,000 to $4,000 per month for the research and outreach tasks above.

An AI sales employee on Summon is $500 per month for the full company — all 8 departments including Sales. You are not replacing a closer. You are eliminating the preparation and follow-up drag that keeps founders from doing the relationship work only they can do.

Hire an AI sales employee

Summon's Sales department handles prospect research, outreach drafts, follow-up queues, and pipeline tracking — with a human approval gate on every send.

Start for $500/month →

Cancel any time. No per-seat pricing. All 8 departments included.

Frequently asked questions

Can an AI employee actually do sales work?

Yes — for research-heavy and writing-heavy sales tasks. An AI sales employee excels at prospect research, personalised outreach drafts, follow-up sequences, CRM updates, and pipeline summaries. It cannot make phone calls or read body language in a room, but it handles the preparation and follow-through that consumes most of a human SDR's day.

What sales tasks should I NOT give an AI employee?

Live calls, negotiation judgment calls, and relationship-sensitive deals that require reading a room. Use the AI employee for the top-of-funnel volume work and outreach prep; keep a human in the loop for closing and sensitive conversations.

How does an AI sales employee hand off to a human?

In Summon, every output that requires human action — a send, a meeting confirmation, a negotiation move — surfaces in an approval queue. The AI employee drafts; the founder or a human closer approves and sends. This keeps the human in the loop without making them do the research.