Lync Agent/Harsha Tejas
The ones that arrived sixty seconds ago, and the ones that went quiet months ago.
Every new enquiry answered in under sixty seconds, over text and phone, day and night.
Every old enquiry can still turn into revenue. From 3,000 leads up, no CRM needed.
Figures from six database reactivation campaigns across four client engagements. Pipeline means everyone ready to buy, booked or qualified, multiplied by the client's own average deal value. Full breakdown and methodology below.
Campaigns run in
“Someone fills in our form at 9pm. We see it the next morning.”
That gap is the whole problem. By then they've contacted two competitors, and whoever replied first usually gets the business.
Speed to Lead→“Honestly? Some enquiries never get called back at all.”
Between 40 and 50% of enquiries never get followed up by anyone. It's rarely a people problem, so hiring more of them rarely fixes it.
Speed to Lead→“We called them once. They didn't pick up, so that was that.”
44% of salespeople stop after one try, but most sales take five. Those leads aren't dead, the follow up just stopped.
Database Reactivation→“We've got thousands of old leads and nobody has touched them in a year.”
You paid for every one of them. They didn't stop being interested, they stopped hearing from you.
Database Reactivation→Follow-up figures from industry research, not our own campaigns: 40-50% of inbound leads are never followed up, 44% of salespeople stop after one attempt, and 80% of sales need five or more touches.
Two clients, their own numbers, measured against their own baseline over the same period.
Methodology. Both figures reported by the client, measured against their own baseline over the same period. Results are not attributed to a named client. Two engagements is a small sample and we are not going to present it as more than that.
Two of these produced a few hundred thousand and one produced thirteen million. They are all here because a table with only the good rows in it is a brochure.
| Sector | Hook | Leads | Response | Ready to buy | Pipeline |
|---|---|---|---|---|---|
| Financial servicesSingle firm | New bond offer | 14,644 | 43.3% | 382 | $13.2M |
| Travel and tourismCruise packages | Competition | 8,648 | 23.3% | 316 | $420K |
| Travel and tourismCruise packages | Direct offer | 4,874 | 31.5% | 93 | $124K |
| Financial servicesAustralia | SpaceX IPO access | 9,923 | 30.7% | 122 | $1.27M |
| Financial servicesEurope | SpaceX IPO access | 5,650 | 32.4% | 20 | $348K |
| Boiler companyUnited Kingdom | Free air conditioner offer | 21,831 | 22.9% | 99 | $1.07M |
| Six campaigns, four engagements | 65,570 | 30.1% | 1,032 | $16.5M | |
Single firm · New bond offer
Cruise packages · Competition
Cruise packages · Direct offer
Australia · SpaceX IPO access
Europe · SpaceX IPO access
United Kingdom · Free air conditioner offer
Methodology. Six campaigns across four client engagements. Pipeline is total ready to buy, meaning booked appointments plus qualified prospects, multiplied by the client's own average deal value, reported by the client.
Ten second version
Move the two numbers. No form, no email. To pressure test it against your actual list, the call is one click away.
Speed to Lead and Database Reactivation are productised because they are the two problems almost every business has. The rest of our time goes on things that only one business has. If you have an idea and no idea whether it is possible, that is a good call to book.
The repetitive thing your team does forty times a week, done by a machine instead. Usually the fastest payback of anything here.
You keep the CRM and the tools you already pay for. The intelligence gets added to them rather than replacing them.
Support, onboarding, internal assistants. Built against your actual documents and your actual rules, not a generic wrapper.
Most of the interesting work starts as a sentence beginning "can AI do this". Bring the sentence and we will tell you honestly whether it can.

Harsha Tejas, founder and lead engineer.
Three years building AI in production, starting with computer vision before ChatGPT existed. I did not learn this on clients: I spent the first stretch as the technical partner behind other AI agencies, building the systems they put their logo on.
When your agent mishandles an objection at two in the morning, I can go into it and fix the thing itself.
Two drone systems: automated warehouse inventory, and surveillance. Both still running today. This was before ChatGPT existed, which matters mainly because it means I learned this as engineering rather than as prompting.
Worked with some of the largest names in American tech on model training and evaluation across more than ten industry projects. This is where I learned what these models are actually reliable at, and where they quietly fall over.
Started as the technical partner behind other AI agencies before putting my own name on anything. Now building voice and SMS systems for businesses in the US, UK and Australia.
Not here? Email it over and you will get a real answer back.
A text inside ten seconds and a call inside thirty. If they do not pick up, the text thread carries on for seven days, then stops.
Yes. It sits behind whatever you already use and books into your existing calendar.
Longer than a reactivation campaign, because it has to sit inside your intake and be tested against real enquiries before it is trusted with them.
If they ask outright, it says yes. It will not announce it in the first sentence, because that ends the conversation before they hear about the thing they enquired about. You can have it disclose upfront if you would rather.
A setup fee to build and wire it in, then a monthly maintenance fee to keep the agents tuned as real enquiries come through. That monthly fee is the only recurring cost. The numbers move with your enquiry volume, so you get them on the call.
They go onto a do not contact list straight away, and you get that list. It is yours to keep.
No. Everyone contacted previously gave you their details and asked to hear from you, which is the distinction that matters in the US, UK, Australia and the EU. The AI only ever speaks to people who reply first. We do not buy lists and we do not scrape.
From 3,000 previously engaged leads. Below that the economics usually do not work, and we will tell you so rather than take the setup fee.
Engaged within the last twelve months ideally. Up to twenty four works, depending on how strong the new offer is.
No. A spreadsheet of names and phone numbers is enough. If you do have a CRM, it writes back into it.
Usually inside a week. The list already exists, so the work is in the offer and the agent rather than in plumbing.
A small setup fee, then a commission on the calls the campaign actually books. There is no monthly fee, so you are not paying for a system between campaigns. The numbers move with your list size and deal value, so you get them on the call.
Do not contact list, straight away, and the campaign never messages them again. You get that list alongside the booked appointments, so it carries across anything you run in future, with us or without us.
Built on
Nothing exotic. Anything else with an API can be wired in.











You are on with the person who builds these, not a setter working from a script.
Prefer to write first? harsha@lyncagent.com.
What we cover