Automate the busywork.Keep the growth.

Custom AI systems that take support, sales admin and back-office work off your desk. Built and maintained by one person — so whoever understands your problem is the one who solves it.

See how it works
3+ yrs
In business operations & program management
6
Product & automation programs led end to end
20+
Cities covered by systems I rolled out

From programmes led in-house before founding Lytheo — not client results. Where that applies.

Automatingcustomer support
support-agent.workflow5 steps
  1. on message.receivedEmail, chat or WhatsApp
  2. classify.intentBy topic and urgency
  3. retrieve.answerFrom your docs and policies
  4. draft.replyCited to the source used
  5. resolve | escalateFull context to a person
Showing Support agent, 1 of 4

Builds on

OpenAIAnthropic ClaudeGoogle Geminin8nMakeZapierPythonGoogle Apps ScriptGoogle SheetsWhatsApp Business APITelegramSlackAirtableNotionHubSpotPlaywright

What I build

Custom AI. Real business problems.

Not off-the-shelf chatbots. Systems designed around your workflows and wired into the tools you already run.

AI support & sales agents

Agents that answer from your own documentation, refuse what they cannot source, and escalate with the whole conversation attached. The boundary is set by you, in writing, before anything goes near a customer.

Answers from

  • Your help centre and policy pages
  • Product, pricing and order data
  • Past tickets your team resolved

Hands to a person when

  • Nothing in your docs covers it
  • Refunds or credits above your threshold
  • The customer asks for a person

Not sure where AI fits? Start with the audit.

A structured teardown of your operations with every automatable workflow scored — and an honest list of the ones not worth touching.

How it works

From audit to autopilot in weeks.

Fixed scope. One fixed price. Working software every week. You always know what ships next and what it saves.

  1. Audit

    Week 0 · Free

    Thirty minutes on where your hours actually go. I map the repetitive workflows, time them, and score each one by what automating it would return. You keep that map either way — including the honest note on which processes to leave alone.

  2. Blueprint

    Week 1

    A written proposal: what gets built, which systems it touches, what it costs to run each month, the timeline, and one fixed price for the whole engagement. Nothing starts until you approve it. The price does not move unless you change the scope.

  3. Build

    Weeks 2–4

    I design, build and test inside your own accounts, with evaluations and guardrails on anything using a language model. You see working software every week rather than a status update — which means you can redirect early, while redirecting is still cheap.

  4. Run & improve

    Ongoing

    Launch, monitor, iterate. Thirty days of support included with every build. After that: take a care plan, hand it to your own team using the documentation, or run it yourself. All three are genuinely fine.

Track record

Where the experience comes from.

Lytheo is new. The operating experience behind it is not — and it is more useful to you if I show exactly where it was earned.

3+ yrs

In business operations & program management

6

Product & automation programs led end to end

20+

Cities covered by systems I rolled out

What these numbers are

These figures come from programs I led in-house at a fintech lender and a services marketplace before founding Lytheo — not from client engagements. Independent client results will be published here as engagements complete, with permission.

Selected work

Systems I have actually shipped.

Financial services

Manual checks removed from the sales floor

The back-and-forth between sales and underwriting went away. The answer arrived during the conversation instead of after it. The system ran in production through business hours.

PythonGoogle Apps ScriptGoogle Sheets

Built in-house as Program Manager at a fintech lender

Read the detail
Content operations

Keyword research to publish-ready draft, unattended

A draft arrives ready to edit rather than ready to assemble. It is the clearest example of what I build for clients: the expensive part of a job is rarely the thinking, it is the handoffs.

n8nAnthropic ClaudeREST APIs

Personal build — my own automation, running on my own stack

Read the detail
Insurance operations

A failure queue that clears itself

Nobody watches the queue by hand any more. Failures are found and re-run automatically, so the queue clears whether or not someone remembers to look.

Google Apps ScriptREST APIsPostman

Built in-house at a fintech lender

Read the detail

See all six projects

Why trust a new studio

No client logos yet. Commitments instead.

Lytheo is new, and pretending otherwise would be the fastest way to lose your trust. So here is what goes into every engagement agreement — enforceable, not decorative.

You own everything

Code, workflows, prompts and documentation transfer to you in writing on final payment. Everything runs in accounts registered to you. There is no Lytheo platform you must keep paying for to keep what you bought.

The audit is free, and yours regardless

You leave the first call with a written map of your automatable workflows and what each is worth — whether or not you hire me. If the honest answer is that nothing is worth automating yet, you get that in writing too.

One fixed price, agreed up front

No hourly billing. No change fees for my own misestimates. The price moves only if you change the scope, and only with your written agreement first.

You pay against working software

Milestones are demonstrable systems, not percentages of elapsed time. Nothing is invoiced before it ships.

Your data stays yours

Enterprise API tiers where your inputs are excluded from model training. Least-privilege access. An NDA signed on request before you tell me anything sensitive. On exit, credentials revoked and my working copies deleted.

No lock-in, no notice period

Care plans run month to month. Cancel whenever you like, keep everything, and I hand over cleanly to whoever comes next.

Automation is not about replacing people. It is about letting people spend their time on work that deserves it.

Ajay Kumar Emani, founder
Business first
Process second
AI third

In that order, deliberately. Most automation projects fail because they start with a tool and go looking for a use. I start with the business outcome, model the human process that produces it, and only then decide whether AI belongs anywhere in it. Often it does not — and the honest answer is a simpler system.

Plain language

Weekly updates you do not have to decode. If you need me to translate a status update, I wrote it badly.

Documented

Every workflow and integration written up so your team, or any developer you hire later, can take it over cleanly.

Honest scope

If off-the-shelf software solves it, or a process is not worth automating, I say so. Including when it costs me the project.

Built around how you work

No recycled workflows. Systems fit your actual process — usually the reason off-the-shelf did not.

Pricing

Fixed scope. No surprises.

Quote-only by design. Every engagement is scoped on a free 30-minute call, then quoted as one fixed price for the whole piece of work. No hourly billing. No discovery fee. No retainer you cannot leave.

Free 30-minute call

We scope the problem together and identify which workflows are worth automating first — and which are not. No charge, no obligation, and the audit map is yours either way.

Fixed quote within 48 hours

One number for the entire engagement with the scope written out explicitly, plus an honest estimate of what the system costs to run each month in third-party fees.

Weekly builds, milestone invoices

You see working systems every week and pay against milestones as they ship — never in advance of delivered work.


Audits run about a week · most builds ship in two to four weeks · month-to-month care plans available, never required

Why there is no price list

Because a published price for custom work is either fiction or a floor nobody meets. Two businesses asking for invoice automation can differ tenfold in how many systems are involved and how many exceptions need handling. Rather than advertise a number that changes the moment we speak, I quote once — after I understand the problem, and before you commit anything.

Running costs are yours, and visible before you commit

AI model usage, automation platform subscriptions and hosting are billed by those vendors to accounts you own. I never mark them up or bill them through me. The blueprint includes an estimated monthly running cost with the assumptions written out, so you can check the arithmetic yourself.

Lytheo

Ajay Kumar Emani · Founder

3+ yrs
In business operations & program management
6
Product & automation programs led end to end

About

A studio of one. Deliberately.

I'm Ajay Kumar Emani. I spent about three years in program management and business operations at a fintech lender and a services marketplace. That meant living inside the repetitive work that quietly eats a team's week, and answering for it when it did not get done.

The systems I built there were not clever. Each one removed a manual step that was costing real money: an eligibility check that meant sending a message and waiting, a failure queue someone had to remember to clear, a report put together by hand every month. They worked because I understood the process first and chose the tools second.

Lytheo is that work, done for other people's businesses. One person, on purpose. Whoever looks at your operations is the same person who builds the system, and the same person who answers when it breaks. No account managers. No handovers. Nobody learning your business from a brief.

  • Direct accessYou talk to the person building it, at every stage.
  • Operations backgroundAbout three years running business operations, so my first question is what a process costs you — not what would be interesting to build.
  • Priced on paybackBefore I build anything, I work out what the process costs you now in hours and money. If a system would cost more than it saves, I tell you not to build it.

Before you book

Frequently asked questions

Models from OpenAI, Anthropic and Google, orchestrated with n8n, Make or Python, and wired into whatever you already run — HubSpot, Slack, Google Workspace, Airtable, Notion, WhatsApp Business, your CRM or helpdesk. Everything is built inside accounts you own, so you hold the keys and the billing from day one.

The audit takes about a week. Most builds ship in two to four weeks with working software to review every week. Larger multi-system projects are phased deliberately so something is saving you time by week three rather than everything arriving at the end.

No, it is not used for training. I build on enterprise API tiers where inputs are excluded from model training by default, use least-privilege access so a workflow can only reach the specific data it needs, and sign an NDA before you share anything sensitive. Your data stays in systems you control, and when an engagement ends, credentials are revoked and my working copies are deleted.

No. Handover includes documentation written in plain English and a recorded walkthrough for your team. If you have an in-house developer they get the full source and a technical session, but nothing is built on the assumption that you have one.

Every build includes thirty days of support. After that you can take a month-to-month care plan covering monitoring, provider changes and small iterations, or run it yourself from the documentation. Both are genuinely fine — care plans have no minimum term and no notice period.

Two honest reasons and one honest caution. First, you work directly with the person building your systems — no account managers, no handoffs, nothing lost between the person who understood your problem and the person writing the automation. Second, my background is business operations and program management rather than pure engineering, so the starting question is what a process costs you rather than what is technically interesting. The caution: I have no client case studies yet, which is exactly why the audit is free and the commitments on this site are written into the contract rather than just displayed on a page.

Yes — most enquiries come from the US and UK. Calls run from early morning to late evening India time, which covers UK business hours and US mornings comfortably. Invoicing is handled in your currency, and engagement terms and data-handling arrangements are set up to suit your jurisdiction, including GDPR and UK GDPR requirements where they apply.

Then I tell you on the free call, before you have spent anything. Processes that run rarely, change shape every time, or genuinely need human judgement at each step will cost more to automate than they return. A meaningful share of audit calls end with one thing worth building and several things worth leaving alone — and you keep that written assessment either way.

Ready to get the hours back?

A free 30-minute audit call. You leave with a written map of your automatable workflows and what each one is worth — whether we end up working together or not.

Replies within 24 hours · No sales team — you talk to the builder

Chat on WhatsApp