How Recruiters Can Use LLMs to Personalize For Prospects
Key Takeaways
"Hey [Insert Candidate Name], just following up” emails in 2025 get totally ghosted by candidates and clients.
Large Language Models (LLMs) can personalize your outreach, automate the boring parts of recruitment sourcing, and make selling your service feel like less of a beige chore.
LLMs help you source talent at scale without giving up strong connections with your candidates.
Staffing agency recruiters: god bless you all. You are part headhunter, part salesperson, and part therapist. Between candidate sourcing, client pitching, and following up on seemingly dead leads, the prospect of cold outreach on a Monday morning is like a root canal in hell.
Yet you go to that dark place anyway, because personalization isn’t optional anymore. Manually customizing every email and LinkedIn message is destroying you, but what else can be done?
If you haven't started customizing LLMs to take this off your chest, this article may be your magic bullet.
Are Large Language Models A Recruitment Software?
An LLM is basically a recruiter who’s read every job post, LinkedIn profile, and company blog in existence without the 15 cups of coffee.
Modern AI models like Chat GPT can not only generate human-sounding language, but they can even interpret human behavior. Translation: they’re insanely useful as a recruitment software.
How many times have you needed to set the exact right tone in your email to a prospect? "I need them to see the urgency here, but also know that I've got their back."
And how many times has Jennifer from Alexandria had completely different needs and tonality than Dan from Cheviot?
Keeping those conversations separate through customization can help you isolate not just files and memory stored, but also tones and conversation structures that play such an intricate role in your agency's process.
The Secret to New Business: How Recruiters Are Using LLMs
Old-school cold emails like “I can help you find your next dream job” don’t cut it anymore.
By using an LLM, you can pull in unbelievably important attributes by scraping private and candidate-provided data. You can reference hundreds of job postings and analyze changes to team sizes, all in seconds.
All of that culminates in a VERY compelling message to your candidate: the time is NOW to make a career move, and it's based on the data.
This kind of marketing for recruiting firms increases open rates, and it opens doors.
But that's just the first step. You can also create tailored content for your own social media or website.
Create a custom GPT that uses a knowledge base of your past social media posts, and ask it to create another one in that same tone of voice. As a matter of fact, ask it to create a 365-day content calendar!
Lastly, adjust your follow up in a way that resonates with that person.
In very rare cases (like Ask-Dwight Pro), you can actually pair an account-level (or, in your case, a candidate-level) knowledge base with a general knowledge base that's specific to you.
That way, you can connect the dots between how you support candidates best with how that candidate best feels supported. It's a win-win for everyone, and your follow up emails are an absolute homerun every time.
Personalization at scale is the new power move. It's a business centered around humans, and to scale your actual HUMANITY??? Now that is a level up.
LLMs won’t replace your gut instinct or industry know-how. But they'll write faster than you and research better than you. Anyone pretending otherwise is going to get lost on the way up.
And the customization they're offering in 2025 will make you more effective at candidate sourcing, more creative with your recruitment, and allow you to convert prospects into placements on a scale you've never seen before.
Lead Scoring: Where Can AI Help?
Key Takeaways:
“Lead Scoring” in sales is the process of assigning a “score” to each lead to indicate likelihood of ultimately converting that lead.
People have historically defaulted to a totally unstructured process for scores as opposed to any sort of centralized system. The rise of AI assistance can help create customization at scale.
AI companies like Ask-Dwight have tailored their large language models (LLMs) around helping salespeople customize scoring for themselves, allowing them to self-standardize with what they value most.
“Here goes Kyle again on his ‘AI helps everything in sales’ spiel,” I know, I know. But the truth is, AI (and, specifically, LLMs) can help take a sales team from a ragtag team of underachievers to award winning efficiency FACTORIES.
If you know where to use it, AI can avoid the de-humanization of your sales process and still unlock efficiencies you never thought possible.
Here at Ask-Dwight, we believe lead scoring is on that frontier.
What is Lead Scoring?
Lead scoring is a process by which you assign a “score” to your leads, which helps you focus on the best performers.
In short, it makes you more efficient with your time.
For instance, you might spend half your day on a lead that just got the discovery call on the books and is led by your best friend from college.
Alternatively, the lead who’s ghosted you for four months and needs the quarter to “get liquid” might take the backseat of your mental load in favor of other prospects.
It doesn’t guarantee you won’t waste your time, but it does try to analytically design your day based on what will earn you the most ROI with your time.
If you’ve not heard of or tried lead scoring before, close this article immediately and go give it a shot! It’ll change your life.
How Do People Traditionally Score Leads?
If you’re a little more seasoned in the sales world, you might know lead scoring as a more personal or company-wide process, which may help you early on with your daily task management but can get tricky as you evolve across different teams and leaders.
But what if that process was more customized?
What if you said to yourself, “maybe my company is scoring leads with the BANT model, but I know I’m converting sales every time a person engages more than once a month?
Do I leave those sales on the table, just because my company can’t meet those needs?
Or what if you’re the sales lead and your teammates are sharing a MILLION of these examples. How do you create a centralized system of lead scoring that also allows for some customization depending on your team individuals’ strengths and weaknesses?
These are issues that come up every single day, even with sales teams actively using AI.
It’s not enough to score leads like a lumbering dinosaur of the 20th century. You have to marshal your resources and personalize each person’s lead scoring process.
Do that, and you may find yourself shattering quotas this year.
How to Score Leads At Scale With AI
Now what’s a fun lead scoring exercise without a little AI right???
Here’s what you do:
Log into any LLM (we prefer Ask-Dwight, but you can use any).
Ask it to provide you with a baseline lead scoring framework. You can base it off of your company’s model, or you can ask it to provide you an outline to fill out.
Adjust that framework to include any specific needs you may have. For example, “factor in companies that received our Christmas card this year, companies with fewer than 500 employees, and women-owned businesses.”
Reassign weights to each allocated factor to better align to your unique goals and sales strengths.
Lastly, ask the model to export as an uploadable document to support your Sales Playbook. Download and use that to support a new custom GPT- or, with Ask-Dwight, just upload to your “My Playbook” section and prioritize your pipeline in any chat!
The reality is that your competitors are already using AI to streamline their lead scoring. You can either beat them at it or lose to them at it.
You can bet your bottom dollar they’ll have their teams spending time on what matters most, both as a team but also as individuals.
In the end, you’re not really using AI here to automate.
You’re using AI to beat your competitors at an efficiency chess match, with AI as the cheat code.
Game on, suckas.
What is a “Sales LLM?”
The fact that this question is in play is a difficult reality to face for the good people of Jasper, Gong, and friends. While your typical '“do-all” sales AI tool is diving Scrooge McDuck style into their piles of gold, others have quietly carved out a specialized niche for LLMs that speak specifically to sales teams’ unique needs.
All of this came from a simple question: if a large language model (LLM) ONLY existed for sales teams, how might that shape its growth & development?
What if it were solely focused on optimizing a sales process?
Would the “web search” function matter as much as the “file upload” experience?
Would the model be trained to interpret the acronym “AE” as “account executive?” Or would it default to advice on your American Eagle shopping intentions?
These are the questions asked when sales-focused LLMs were created.
Why Are People Seeking Alternatives to Chat GPT?
“Well Chat GPT alternatives aren’t hard to come by! Just look at Claude. Look at Perplexity. At Gemini!”
Let me answer this very clearly: Chat GPT may have general knowledge alternatives, but exactly 0 have proven themselves industry specialists. Here at Ask-Dwight, we believe that’s the future.
The future is looking at not only a sales team AI, but a B2B sales team AI. A B2C sales team AI. And I don’t mean general AI with 8 million bells & whistles (shoutout to the players at the intro to this article!). I mean freaking LLM specialists.
People who use Chat GPT are looking to save time. They’re looking for uniquely tailored responses to their situations.
So you tell me: what’s the best way to provide that?
You get specialized. You get specific with your training, and you tailor your LLM’s learning using responses from the masters of that industry.
When we did that at Ask-Dwight, we got way further along in the time saving for sales teams. Forget hours, we were focused on saving time down to the minute… the second!
That’s how salespeople operate. But when they’re focused on serving the “everyman,” Chat GPT doesn’t have a prayer of serving people at that level.
What Do Specialized Large Language Models Do Differently than Custom GPTs?
“If you want specialization, just use a Custom GPT!”
Now to be fair, this is the closest we will get to Chat GPT serving salespeople correctly. But Custom GPTs fail in a few key ways:
While they’re providing tailored output, the entire experience should be tailored… not just the conversation
Custom GPTs still require significant refinement from sales teams, which may not be intuitive. Would you trust 22 year old Braxton with writing your custom GPT? What about 65 year old Henry?
Custom GPTs still don’t impact the overall direction of the LLM at large. Your Custom GPT won’t help Chat GPT make decisions about what features to add next.
The vision for specialized LLMs does all of these things. It makes the sales process not just easier, but absolutely natural. Collaborating with your LLM should be like talking to a sales associate, not a robot yes-man.
Why Not Use Established Sales AI Tools?
There’s no doubt that well-known sales AI tools own the integration space. You can use their internal LLM to source and relay your own data.
But those integrations can break, the LLM may be hard to find, and (the biggest offender) often the tool simply leans on the trajectory of Chat GPT to decide how the LLM will grow in the space.
If you need a quick data point here or there, using these LLMs is fine of course. But a real partner in sales should be something more.
Now here’s the homer take: the best sales LLM is Ask-Dwight, hands down.
It tailors output to your industry, and acts as a true partner in sales. You can replace Chat GPT with it or simply add Dwight to your tool belt.
Join the transition to industry-customized LLMs today, and say hello to your new best salesman: Dwight.
Sales AI Email Generators: Why You Need Personalization
Numbered are the days where people will turn to Chat GPT for their every sales AI (Artificial Intelligence) needs. Don’t get me wrong, Open AI built an incredibly powerful tool that changed the world, but the idea of sales-focused output being customized to you without a heavy lift on your end is almost laughable.
In almost any sales role, cold email outreach is a staple in your funnel, and a process whose importance can’t be overstated.
When you use an AI to help you generate that email, you’re essentially asking a personal assistant to research and draft that outreach on the fly.
Close, but what if you used a Sales assistant?
The specific sales training, the background, the experience helping salespeople close valuable deals… this is the future I’m proposing in the world of LLMs (Large Language Models), and email generation is the perfect entry point.
How To Use AI With Email Generation
Start with this: enter as much data as humanly possible (within compliance rules) about what you’re selling and who you’re selling to.
Note any recent context and get good at describing emotional states of all stakeholders.
Once your AI tool has a clear picture of the situation, it’ll be way more equipped to help you analyze and respond to emails.
Next, brainstorm some top-of-mind thoughts for your response. Are you for or against what they’re suggesting? Do you have a clear response, or are there pieces to circle back on? Digest your own thoughts, organize them, and lay them out in priority order for the AI.
From there, allow the software to create your first draft.
This part is INCREDIBLY important, so don’t skip it: REFINE what the LLM gives you.
What would you do if a sales associate sent you a response that needed tweaking? You wouldn’t rewrite it from there. Being the amazing leader you are, you’d tell them what’s wrong and have them fix it themselves.
Hold an advanced LLM to that same standard. It improves their knowledge of what you’re working on, and it improves your ability to design excellent prompt sequences for a complex machine. Win-win.
Lastly, make final adjustments. In this last stage, you should realistically only edit in one or two places. But make it your own: reference a joke from before, or a comment about your city’s weather or something.
Humanize it.
Then send away! Each email will be unique, but the process won’t be:
Compile good input data
Generate an email draft using the LLM
Refine output data
Humanize refined data
Send
A few attempts later, and you’ll be an expert!
Why A Sales-Specialized Tool?
Here’s where most AI sales tools screw up: they create tools for the every man. "Market share” this and “active users” that… for the personal side, it’s all good. But you’re a professional.
AI in 2025 can and should be customized by industry. Sales AI is no different at all.
Ironically, “traditional” AI like Salesforce and Gong don’t have any problems creating automation. In fact, they’re elite at it. Hats off to them!
But do they do a better job with their in-house LLMs than Chat GPT?
Nope. They’ve yet to successfully develop an LLM that accurately represents the desired UX of a salesperson.
The hope is that will change moving forward, and Ask-Dwight is at the cutting edge of those efforts. However, if you do choose to stick with a Chat GPT, a Claude, or another more general LLM in the field, be sure to double down on your Custom GPTs to ensure your background is organized and accessible to help you specialize.
Top AI for Writing Cold Emails
In order from most valuable to least, here are the best AI to help you out with cold emails:
Ask-Dwight- yeah, yeah, we’re biased, but hear me out: Dwight’s directness and fully-customized sales background makes email drafts clear but polite, and uniquely tuned to salespeople’s needs for accomplishing sales tasks.
Gong.io- also a great option! Super skilled at using your historical data to create value-adding emails.
Lavender.ai- great for reviewing language to reduce spam folder placement. Not so good with doing the actual writing.
There you have it! These cold emails won’t be writing themselves, so consider the information laid out in this article, and if you have any questions, feel free to hit up Dwight on your way out!
LLMs For Salespeople: Turn Your AI Into An Industry Specialist
We all love a good Chat GPT, a Claude, a Gemini and, of course, the sheer MAGIC of what they’ve created. But they’ve stopped well short of the finish line. Large Language Models (“LLMs”), like other artificial intelligence (AI), SHOULD be specialized.
Not only should they be specialized, they should be uniquely GIFTED to support those specialized fields. It’s not enough to slap together a Custom GPT or Claude project. LLMs should be designed for an industry. Shaped by it over time.
We all love a good Chat GPT, a Claude, a Gemini and, of course, the sheer MAGIC of what they’ve created. But they’ve stopped well short of the finish line. Large Language Models (“LLMs”), like other artificial intelligence (AI), SHOULD be specialized.
Not only should they be specialized, they should be uniquely GIFTED to support those specialized fields. It’s not enough to slap together a Custom GPT or Claude project. LLMs should be designed for an industry. Shaped by it over time.
Without that level of personalization, the only “shape” you’ll be in is “square.” Only dad joke in here, I promise.
How Do Common Large Language Models Specialize?
Now that I’ve said that, let me backtrack before I tell you what we’re doing with Ask-Dwight.
If you want to customize your solution, start with a custom GPT. For consistency’s sake, I’ll use Chat GPT’s vernacular for this since they’re the clear frontrunners for this technology.
With a pro plan on almost any LLM, you can create customized “instructions” for your GPT to follow. For instance, you could:
Design a curated outfit for the day using only your wardrobe.
Create a fully personalized workout plan to fit your diet and your exercise schedule.
Schedule watering & other care for your interior or exterior plants around the house.
Even in business, there’s countless ways to customize large language models to work for your unique needs. Your team could use custom GPTs to:
Rewrite sales email drafts using your language, and optimizing for a specific tone or purpose.
Run a custom sales playbook through your LLM, to seamlessly apply vetted persuasion tactics that follow modern best practices to your communication.
Analyze and forecast pathways to conversion for your prospects at scale and within limited time frames.
These are wide applications within a normal Custom GPT experience, although Ask-Dwight and a very few others are actually exploring the limits of these ideas to the fullest.
The reality is that very few AI companies have EXCELLED in these spaces, and when they have they’ve been unable to package and release it to the public in a way that sticks.
How Does Normal AI Specialize Better Than LLMs?
While the world of LLMs seems frozen under the spell of 5-10 big players, then a half a million smaller players, your non-LLM artificial intelligence has done an incredible job specializing!
Look at companies like Gong and Jasper, who are absolutely crushing the specialized world of AI. You have unlimited examples of specializing just from normal AI, which you can (and should) take advantage of TODAY:
Transcribe calls automatically and summarize key points.
Personalize your interactions with prospects given context within your CRM.
Assess sales risk to determine if your pipeline prospects are at flight risk.
Under normal conditions, AI is unbelievably agile and able to specialize seamlessly to an industry- so what’s the missing piece?
It’s the dark, underbelly of the rock nobody wants to look under. It’s the fear of the unknown with LLMs, and how to best leverage them.
The 10 weekly calls with the Product teams, ending in a “let’s circle back next week,” and the reason these larger companies can hide behind a version 25329 release while discounting sales enablement through LLMs as “vibe coding.”
Customized LLMs are the brand-new, hot pink convertible parked in front of the school.
Get in losers. We’re customizing for our people.