AI ruined sales. Or did it really?

The real story is more interesting.

AI did not ruin sales. Humans ruined sales by pointing AI at the wrong target. Instead of asking it to understand people, we asked it to send more. Instead of depth, we optimised for volume. And the rest of the ecosystem - inbox providers, IT admins, security teams - quietly started fighting back.

Before we built Sera, we wanted to understand that shift properly. So we tested 33 different sales outreach tools with AI features. Here is what we found and how it led us to build a network of agents instead of “tool number 34”.

Scale vs Quality: how AI got wired wrong

Almost every sales platform that claims “AI-powered outreach” is built around the same assumption:

If you can send more, you will sell more.

So the AI is pointed at the parts of the workflow that increase volume:

• Writing more low quality emails

• Auto generating more follow ups

• Pushing more contacts into sequences

• Adding a thin layer of [First Name] and [Company Name] tokens and calling it "personalization"

The outcome is predictable:

• Volumes go up

• Quality goes down

• Everyone’s inbox starts to feel like a bad LinkedIn DM.

Even when the copy is not terrible, it still feels mass produced. That is how you end up in two places you really do not want to be:

1. The spam folder

2. The Promotions tab that busy people rarely check

Promotions is an under rated graveyard. Messages are not technically spam, but to the recipient they might as well not exist. A lot of AI written outreach lives there.

When we ran our experiments across 33 tools, the pattern was very clear. The tools that made it easiest to crank up the volume also made it easiest to disappear into Spam and Promotions forever.

Global numbers tell the same story. Recent reports estimate that around 45 to 47 percent of all email traffic is spam - roughly 150 to 170 billion spam emails every single day.

Takeaway: Give the world a powerful engine and most people will use it to build a leaf blower, not a violin.

Every power creates a counter power

Whenever one side in a system gets a new power, something else eventually evolves to balance it out. AI gave senders a new power: the ability to generate convincing, fluent emails in any language, at any scale. The counter power appeared in exactly the right place: inside the inbox.

Gmail and friends said "nope"

Consumer providers like Gmail looked at the explosion of bulk email and reacted the only way they could. They started to:

• Enforce stricter authentication (SPF, DKIM and DMARC)

• Track complaint rates and engagement much more aggressively

• Separate "real" mail from promotions and bulk outreach

• Penalize sudden spikes of activity from a domain

To the average user, this shows up as:

• More mail going directly to spam

• A Promotions tab that quietly absorbs sequences and mass emails

• Fewer obviously automated emails in the primary inbox

Takeaway: If your strategy is "send as much as possible and hope something sticks", the inbox has become an active opponent, not a neutral pipe.

What Outlook and IT teams are doing about it

Now look at the audience most B2B teams actually care about: executives at companies with 200 or more employees. Those inboxes are usually sitting behind:

• Microsoft 365 or similar corporate email infrastructure

• Organisation wide anti spam and anti phishing policies

• Custom rules designed by internal IT teams

Microsoft gives admins very fine grained tools to tune spam, “bulk mail” and graymail filtering, choose where suspicious messages go, and apply more aggressive policies to protect their users.

On top of the default spam and phishing filters, internal IT teams often:

• Turn up the sensitivity on bulk behavior

• Add rules that block or quarantine unfamiliar senders who hit multiple employees

• Deploy advanced threat protection tools that score every incoming message

Their job is to keep employees safe from phishing, malware, and unnecessary noise. Your outbound sequence is just another unknown signal in that noise.

So the bigger and more attractive the company, the more likely it is that:

• A high volume, low quality approach never reaches the executive at all

• Messages get stuck in quarantine or internal "junk" folders long before the user sees them

• Domain reputation risks compound with every bad campaign

Takeaway: Scale used to feel powerful. In this environment, it looks more like self sabotage.

Meanwhile, The Bad Guys Upgraded Too

There is a darker twist. While honest sales teams struggle with deliverability, criminals are happily using the same AI to get better at phishing. A recent report found that more than 80 percent of phishing emails in late 2024 and early 2025 showed some use of AI.

AI removes the classic “tells” - broken English, strange phrasing, copy paste templates - and makes malicious emails smoother and more targeted. Other research shows AI generated phishing attempts now outperform human written ones by a wide margin. To cope, inbox providers have to keep turning the dial up. The collateral damage hits anyone who still believes “more emails equals more pipeline”.

We tested 33 sales outreach tools and here is what we found

When we started building Sera, we did not want to guess. We wanted to see what actually worked in this new environment. So we went hands on with 33 different sales outreach tools that offered AI features. Over several months we:

• Analyzed and ran real campaigns

• Measured reply rates

• Watched what happened to deliverability and domain health

• Saw which conversations led to real opportunities

Here is what stood out.

1. Tools were optimised for activity, not outcomes

The default experience everywhere looked similar:

• Bigger lists were easier than sharper segments

• More steps in a cadence were easier than better timing

• Analytics highlighted opens and send volume more than positive replies

The more a product celebrated “activity”, the faster domain reputation eroded.

2. Volume hurt more than it helped

When we did what the tools made easiest - large lists, generic copy, high frequency - we saw:

• Higher rates of messages landing in Spam or Promotions

• Increased negative responses

• Very few meaningful conversations with the right people

When we deliberately did the opposite - small lists, heavy research, careful sending - we saw:

• Higher positive reply rates

• Fewer deliverability incidents

• Leads replying with “this actually makes sense” instead of “unsubscribe”

The winning pattern was clear. The whole stack needed to be optimised for better bets, not more shots. So we took everything we had learned from those 33 tools, cherry picked the genuinely useful capabilities, and used that as training material for something else entirely.

AI agents replacing sales tools and specialists

Rather than adding one more tool to the pile, we built Sera as a set of specialised agents that run tools for you. Traditional tools are different. You still need someone who knows how to drive them. You buy the software, then you either:

• allocate your best teammate to live inside sequences, filters and reports, or

• hire a specialist who spends their days tweaking settings, reading deliverability blogs and trying to reverse engineer what went wrong.

So you end up paying for the tool, paying for the operator, and still getting hit by spam filters and weak reply rates. High cost, low signal.

With Sera, you are not given a cockpit and told to figure out what works. You are effectively hiring a small, opinionated AI team that already understands how to operate in this new inbox reality.

We trained our agents on the best humans in outreach. Working in sync they are already outperforming some of the experts who taught them. Here are the six core agents.

List Building Agent - sources verified companies and decision makers from more than 300 million companies and over 1 billion professionals, based on your ideal customer profile. Instead of a bloated list, you get a focused set of people who are actually worth talking to.

Enrichment Engine - expands each contact with firmographics, tech stack, revenue and market data so you always reach right fit prospects. Every row in the list comes with enough context that your outreach can sound informed from the very first line.

Research Analyst - scans the web, LinkedIn and press to uncover buying signals, timing and angles that make your message resonate. Funding rounds, hiring waves, product launches, strategic shifts; this agent turns background noise into reasons to reach out now instead of “sometime”.

Decisionmaker Identifier - makes sure you are speaking to the person who can actually say yes. Instead of blanketing a whole department, this agent hunts down the real decision maker or champion inside the account and confirms they are the right starting point.

Deliverability Guard - warms new domains, validates addresses and optimises pacing to keep deliverability high, often above 93 percent. It watches for patterns that would push you into Spam or Promotions and adjusts behaviour before inboxes start pushing back.

Outreach Writer - writes researched, multilingual, natural emails that feel authentic, not automated. It pulls in everything the other agents have learned and turns it into short, specific messages that connect your value to what is happening in the prospect’s world.

Crucially, these agents are trained on the strategies that worked best when we used those 33 tools, not on the high volume habits that burned domains.

You do not need a power user who knows every deliverability trick. The agents behave like that person by default, so you are not stacking tool costs on top of headcount costs just to stand still.

Quality as the default setting

Most platforms treat low volume, high quality, highly researched outreach as an expert move you grow into. Sera treats it as the starting point.

• Volume is kept intentionally low

• Research is non optional

• AI is used to reduce waste, not inflate it

Every send is treated as a decision, not as background noise. That is also how you work with the modern inbox instead of fighting it. You build a sending pattern that looks like thoughtful, human communication, not like an automated cannon.

AI did not kill sales. Lazy sales did.

So did AI ruin sales?

Not exactly.

We, humans, took a powerful technology and pointed it at the wrong metric. We asked it to help us shout louder instead of to help us choose our words, and our recipients, more carefully. The inbox responded, security teams responded, buyers tuned out.

The next decade in outbound will belong to the teams that do the opposite:

• Fewer emails

• More research

• Respect for deliverability and admin rules

• AI used as a strategist and operator, not a spam engine

That is the bet we made with Sera: AI not as a leaf blower, but as a careful, composed partner that knows when to speak, what to say, and just as importantly, when to stay quiet.