AI Marketing
AI marketing services are the prompts, guidelines, editing and campaigns that put AI inside the marketing work your team already does. Eicra Soft Limited in Dhaka runs six of them for B2B companies: brand voice, prompt strategy, prompt engineering, AI content editing, email personalization and campaign management. A named human editor checks every output before a customer sees it, and you keep the prompts, guidelines and data.
Book a 30-minute scoping call See the six services
Built in your own CMS, email platform and ad accounts · NDA and DPA signed before any data moves · Delivered from Dhaka on UTC+6
Each one is a separate engagement with its own scope, price and service page. We recommend starting with the voice guideline and the prompts, then adding editing once both are in use. Every service starts with a written diagnostic, so you can stop after the verdict if the problem turns out to be smaller than expected.
One written voice guideline that your writers and your AI tools both follow, with word lists, examples and rewrite rules. It is built from your own published copy and tested on one real channel before the full guide is written.
A written prompt system for the team: which prompts exist, who owns them, how they are versioned, tested and retired. It stops every writer prompting in a different way, so quality no longer depends on who wrote the brief that day.
Production-ready prompts for each recurring marketing task, tested against real briefs, with inputs, guardrails and an expected output. Each prompt is scored by your editors on your own content before your writers use it, and the prompts stay in your accounts.
Human editors check AI drafts for facts, claims, voice and readability, and log every change against the guideline you approved. Anything that cannot be checked against a source you supplied is removed or flagged for you before it is published.
Segments, dynamic blocks and subject-line variants built in your own email platform, with suppression rules and a documented test plan. Every journey is tested on seed inboxes and signed off by your marketing lead before any contact receives it.
Campaign calendars, bid and budget rules, creative rotation and weekly reporting, run in the ad accounts you already own. One named owner keeps the brief and the change log, and every spend change is signed before it goes live.
Start from the symptom, not the tool. If the AI output sounds nothing like you, fix the voice guideline first. If quality swings from writer to writer, fix the prompts. If a wrong claim could reach a customer, add editing. If nobody owns the numbers, start with campaign management.
| Your situation | What you need | Where we start |
|---|---|---|
| AI drafts sound nothing like your brand | Brand Voice Guidelines | Voice audit of 20 published pieces |
| Every writer prompts differently and quality swings | Prompt Strategy | Prompt inventory and owner map |
| The same task is rewritten from scratch every time | Prompt Engineering | Three prompts built and tested |
| Claims, facts or tone reach customers unchecked | AI Content Editing | Sample edit on your own drafts |
| Every email goes to one list with one message | Email Marketing Personalization | Segment and suppression review |
| Spend is live but nobody owns the weekly numbers | Campaign Management | Account audit and weekly report setup |
Five questions. Read across your answer and note the service named. Whichever service appears most often is where to start — and if Brand Voice Guidelines appears twice or more, start there, because the other services are all graded against it.
| Question | Option A | Option B | Option C |
|---|---|---|---|
| 1. How is your brand voice written down? | A full guideline writers follow → Prompt Engineering | A few notes in a slide → Brand Voice Guidelines | It lives in one person’s head → Brand Voice Guidelines |
| 2. Who owns the prompts your team uses? | One owner, versioned → Prompt Engineering | Everyone, in their own way → Prompt Strategy | Nobody; we type freehand → Prompt Strategy |
| 3. What checks AI output before it is published? | A named editor, every time → Prompt Engineering | A quick skim by whoever is free → AI Content Editing | Nothing at all → AI Content Editing |
| 4. How personal is your email programme? | Segments and dynamic blocks → Campaign Management | One list, one message → Email Marketing Personalization | We barely send email → Email Marketing Personalization |
| 5. Who owns the weekly numbers? | A named owner, reported weekly → AI Content Editing | Reported monthly, roughly → Campaign Management | Nobody owns them → Campaign Management |
Three rules decide the order, and we apply them on every engagement. We say so when the cheaper step is the one you actually need, because a voice guideline often fixes more than a new prompt. The five steps below the rules are the same on every service.
A model copies the instruction it is given. If the instruction says “professional and friendly”, every competitor using the same two words gets the same paragraph back. A guideline written for your brand gives the model something only you can supply.
The guideline that fixes this is specific: preferred words, banned phrases, sentence length, reading level, and three worked examples of a rewrite.
A prompt that lives in someone’s browser history is not an asset. A prompt with a name, an owner, a version, defined inputs and an expected output is. It can be tested, improved and handed to a new writer without losing what already works.
That is the difference between output you can predict and output you have to re-check from scratch every single time.
Models state wrong facts confidently. The cost of that is not the draft; it is the correction, the refund, or the regulator’s letter afterwards. A named editor costs less than any of those, and the edit log shows what was caught.
So every published piece passes a named editor who checks claims against your sources and records what was changed and why.
| Step | What happens |
|---|---|
| Scope | A free 30-minute call on the problem, the channels and the tools you already use, ending in a diagnostic quote. |
| Diagnose | A written verdict on your current content, prompts, email or campaigns; from $99 for voice, prompts and email, $1,500 for a campaign audit. |
| Pilot | One channel, workflow or campaign built and tested on real work before anything else is priced. |
| Production | The remaining work added one piece at a time, each tested, documented and signed off. |
| Managed Ops | Optional monthly upkeep: re-tests, updates and a report, with a named owner; cancel any month. |
In February 2024, Gartner predicted that traditional search engine volume would drop 25% by 2026 as search marketing loses share to AI chatbots and virtual agents (Gartner press release, 19 February 2024). Whatever the final figure turns out to be, the volume you cannot control makes the quality of what you publish the thing you can. A page that answers clearly and names its sources is easier for search engines and AI assistants to quote.
| Item | Without | With Eicra Soft Limited |
|---|---|---|
| Brand voice | A short list of adjectives | A written guideline with word lists, examples and rewrite rules |
| Prompts | Typed fresh by each writer | Named, versioned and tested against real briefs before use |
| Facts and claims | Trusted to the model | Checked by a named editor against the sources you supply |
| One list, one message | Segments, dynamic blocks and suppression rules in your own platform | |
| Reporting | A screenshot when someone asks | A weekly report against the metric agreed at kickoff |
| Ownership | Prompts stay in the agency’s account | Prompts, guidelines and templates handed over as files you keep |
| Data handling | Pasted into whichever tool is open | NDA and DPA first, named least-privilege access, removed at handover |
| Contracts | A freelancer invoice | Company contract with GDPR Article 28(3) processor terms where they apply |
Buyers of AI marketing services ask twelve questions before the first call. Cost comes first, because that is the question everyone opens with. Then come where to start, quality, accounts, facts, AI search, writing and handover, followed by next steps, switching agencies, data and what we will not promise.
More answersEach service is priced in fixed steps, and the price of each step is agreed before work starts. The written diagnostic costs $99 for brand voice, prompt engineering and email personalization, credited to the pilot. Pilots are $1,200 for brand voice and $399 for prompt engineering or email. Campaign management starts with a $1,500 audit. Model usage and tool subscriptions are paid by you, directly to the provider.
Brand Voice Guidelines, in almost every case. Prompts are written against the voice and editing is graded against it, so both are guesswork until the guideline exists. If you already have a written, current voice guideline that your writers actually follow, start with Prompt Engineering instead.
Not if the guideline is specific. Generic output comes from generic instructions, so we build word lists, banned phrases, sentence-length rules and worked examples, then a named editor rewrites anything that misses them. You see the edit log, so you can check which rule was applied.
Yours, wherever the platform allows: your CMS, your email service provider and your ad accounts. Our access is named and least-privilege, and it is removed at handover. You receive the prompts, guidelines, templates and reports as files you keep, not as a login you rent.
A named human editor, before publishing. Every claim, number, quote and product detail is checked against the source you supply, and anything unverifiable is removed or flagged rather than published. The edit log records what changed and why, so an approver can audit a piece in minutes.
Yes, as part of content editing rather than as a trick. Assistants quote pages that answer a question directly, name their sources and use clean structure, so we edit for answer-first paragraphs, sourced numbers, schema and clear headings. Placement inside an assistant cannot be bought or guaranteed.
Both, and the split is agreed at kickoff. Drafting is done with your prompts inside your tools; editing is always human. You can keep drafting in-house and buy editing only, because the editor and the guideline are where the quality difference actually comes from.
The voice guideline, the prompt library with versions and owners, the editing checklist, email segments and templates, campaign rules and the reporting sheet. Everything is exported into your own accounts and documented, so your team can run the programme without us from the first week.
Four things happen, in this order.
Yes. The diagnostic starts with a review of what your current agency built: prompts, guidelines, email journeys and ad accounts. Nothing running is paused or deleted without your written sign-off, and the handover is done one service at a time, so nothing goes dark while it moves.
Work starts under an NDA, with processor terms under Article 28(3) of the GDPR wherever personal data is involved. Access is named and least-privilege, in your own accounts, and removed at handover. Customer data is never pasted into public AI tools, and certifications are listed only when they are held.
We do not promise rankings, a place inside an AI assistant's answer, open rates or revenue figures, because those depend on your market, your list and the platforms themselves. We commit to the work: written guidelines, tested prompts, edited content, and reports you can check against your own accounts.
Send us three pieces your AI wrote. We will show you what a voice guideline would change.
Request A QuoteBook a 30-minute scoping callYour request is subject to our Terms and Privacy Policy.
