The $60K SDR problem, solved for people who can't afford one.
If you've ever done outbound yourself, you probably know the pain.
You either hire an SDR, or you spend hours every day finding prospects, researching them, and rewriting basically the same email over and over.
That’s what pushed me to build DeepReachAI.
It researches a lead, looks for actual context around them, and turns that into a personalized message instead of another generic AI-generated email.
Right now, you can click on a LinkedIn profile and get a personalized email in Gmail.
Something I didn’t expect was how much people cared about what we don’t do. We don't automate LinkedIn or scrape WhatsApp. No bots running in the background. The goal is to make personalization easier without putting accounts at risk.
I’m still building this solo, so I’d honestly rather hear what feels clunky or annoying than what looks good.
Also working on something called Orio (Voice agent), which takes this a step further. You give it an instruction, and it can find leads, research them, reach out, and even make calls to book meetings. We’ve already tested the full flow with real calls and are starting to roll it out to active users.
If you're doing outbound, I'd love to hear what you're struggling with.
deepreachai.tech
Hi Aryan, interesting work with DeepReachAI and Orio. I’m an AI Engineer working with AI automation and agents, and I found your approach to personalized outreach particularly interesting. Wishing you the best as you continue developing both products.
Thanks, really appreciate it! Glad you found the personalized outreach approach interesting. Wishing you the best with your AI automation work too :)
The real $60K cost isn't the salary. It's the 6 months to ramp before you know if they're even working. Most early-stage founders find out 8 months in that the hire wasn't right, which puts them a year behind where a well-structured outreach system would have them.
The tools that crack this space usually work well for the first 100-150 contacts and then hit a personalization ceiling — sequences get too templated and reply rates drop fast. The harder problem is data quality. Lead data goes stale faster than sequences run, especially at the domain level.
Curious how you're solving the signal layer. Are you pulling from intent data or is it more first-party behaviour-based?
Yeah, this is actually very close to what we’ve been trying to solve.
I agree that the signal layer is probably more important than the sequence itself. If the lead data is stale, better copy doesn't really save you.
Right now DeepReachAI uses a mix of company/lead research, public signals and context pulled at the time of outreach rather than just relying on a static lead database. That’s also the direction we’re taking Orio in, where you basically give it an ICP + instruction and it does the research before deciding how to approach each lead.
We’re still improving this part, but honestly your point about the 100-150 contact personalization ceiling is exactly the kind of thing I’d love to test against.
If you’re doing outbound yourself, I’d be happy to give you access and let you run a batch through it. Would be much more useful to see how it performs on your leads than just talk about it here.
Interesting framing. The cheapest replacement for an SDR isn’t just automated outreach; it’s a clear qualification gate. I’d make the first touch earn a reply with one specific observation + one binary question, then stop after a short, explicit sequence (for example, three useful touches over 10 days).
A “not now” or no-response state should be treated as a real outcome, not an endless drip. That same clarity is what I’m building into a small $9 close/follow-up kit for designers and devs—different audience, same problem: making the next step easy to decide.
Yeah, I actually agree with the qualification point a lot.
One thing I should clarify though: I don't really see DeepReachAI as an "SDR replacement." It's more like giving a company an AI employee that handles the repetitive outbound work an SDR would otherwise spend hours doing.
The goal isn't to spam someone until they reply either. Finding the right lead, understanding why they might care, deciding how to approach them, and knowing when to stop are all part of what we're building.
The interesting part with Orio is that we're trying to make that whole process happen from one instruction instead of making the user build sequences and workflows themselves.
Your "one observation + one binary question" idea is actually something I'd be curious to test in the system. If you want, I can give you access and you can throw a few of your own leads at it and see if it holds up.
Hey, I ran a quick check on DeepReachAI.
I asked ChatGPT, “best AI tools for personalized cold outreach/prospecting in 2026,” and it came back with Clay, Apollo, Instantly, Smartlead, Common Room, and Unify. DeepReachAI didn’t come up.
What I found interesting is that ChatGPT actually pointed out that the market is shifting more toward “real signals + relevance” rather than generic personalization at scale. That feels pretty aligned with what you’re doing at DeepReachAI, research-driven outreach without the usual automation risk.
So the positioning makes sense. The bigger issue seems to be that DeepReachAI just isn’t showing up in these AI recommendations yet.
I do focused work around this such as finding the threads, directories, comparison pages, and other sources where competitors are getting mentioned, then reaching out to get your product into those conversations.
If that’s something you’re looking at, I can send over the pricing and what the process/timeline would look like.
Appreciate you taking the time to check it out and share that.
Yeah, getting mentioned in AI recommendations is definitely something we'll work on eventually, but it's not something we're actively looking to spend on right now. We're still pretty focused on building the product and getting it into the hands of people actually doing outbound.
Also interesting that ChatGPT framed the market around real signals + relevance. That's pretty much the direction we're taking with DeepReachAI, so that part was good to hear.
Thanks for reaching out though, and I'll keep you in mind when we get to that stage.
A distinction I'd test in the research-to-Gmail flow: a true fact about someone isn't necessarily a reason to contact them. 'You posted about hiring' may be sourced, while 'therefore you need our service' is still an assumption. Showing the source, that inference, and the proposed ask separately would make the draft much easier to review. I'd also ask the sender what they can actually offer before composing; otherwise good research can still produce a vague meeting request. I'm building Portunio around an agent helping the sender clarify an incoming opportunity, so this is a related design question for us. I haven't tried DeepReachAI yet. Does your review step expose the research evidence, or only the finished email?
Yeah, this is a really good distinction. A sourced fact and the conclusion you draw from that fact definitely shouldn't be treated as the same thing.
Right now we show the research/context that went into the email, but I agree there’s room to make the chain much more explicit: “here’s what we found → here’s why we think it matters → here’s what we’re suggesting you say.” That would make reviewing the reasoning a lot easier.
And I agree on the offer too. If the system doesn't understand what you're actually offering, even perfect research can end up producing a generic “would love to connect” email.
You should try DeepReachAI though. Since you’re building Portunio around a similar agent/reasoning problem, I'd actually be really interested in your feedback on where our review flow falls short. I can give you access if you want to play around with it.
One useful guardrail: treat the first 30 sends as a research audit, not a campaign. Log which data point earned the reply, then automate only those fields; a generic fallback should be a hard stop.
The real-call testing is the strongest signal here. Are active users getting materially better outbound results from Orio than from the simpler personalized-email workflow?
Too early to say with confidence, Orio's still in testing and hasn't officially launched yet. Once it's out I'll share whether active users are actually seeing better results than the email only flow or not.
That comparison will be the useful test once Orio is live. What baseline are you using for the simpler email-only flow?
The reaction you did not expect is your actual positioning, so I would move it to the top of the page instead of burying it. We hold the same line on the social side of our own business, and founders who have watched a peer get an account restricted will pick the tool that promises less automation, not more. The harder product problem is what happens when the research turns up nothing real, because a tool that tells you it has nothing worth saying about a prospect is worth more than one that quietly falls back to a generic email.
Good catch, both points are valid. Moving the "no automation" line up top, that's clearly the trust signal people will respond to. And you're right that a tool faking a message on thin research is worse than one that just says "nothing solid here, skip it". That's the actual bar to build toward, appreciate you spelling it out instead of just flagging the vibe.